The Quest for IBM PS/2 Tower 5.25 Drive Bay Bezels

August 11, 2025 by Dan Vincent in podcast

How do you fill a hole in a computer’s heart? Here, in Userlandia, the solution is 3D prints.

My favorite vintage computing acquisition over the past few years might well be my IBM PS/2 Model 80. It’s an ivory monument to corporate hubris that’s built like a tank and can withstand being run over by one too. There’s something romantic about owning the official PC of “it’s somebody else’s money.” How could I not love this big brute of a box? But they say every rose has its thorn, just like every case has its flaw. And most owners of model 60, 65, or 80 towers have encountered it when installing a CD-ROM drive.

You might remember from the 386 Tower of Power video that these cases have an externally accessible full-height 5 1/4 inch drive bay. A hard drive, floppy drive, tape drive, or any kind of drive can move on up to a de-luxe drive bay in the sky. Just screw on a pair of AT rails on to a drive and slide it on in. But wait—that blanking plate’s only good for hard drives! Installing an externally accessible half-height floppy or CD-ROM drive leaves a gaping hole in the front of the case. Even a full-height optical drive needs a piece of trim to look its best installed in these PS/2s. It’s all because IBM needed to make a compromise.

The original IBM PC 5150’s drive bays were all externally accessible. Therefore all drives, even non-removable hard disks, needed integrated bezels to protect against dust ingress. As hard drives grew in popularity and floppy drives shrank in size IBM changed their bay arrangements with the PC AT. Instead of two full-height (or four half-height) external bays, there were now two external half-height bays and one internal full-height bay. This opened the PC AT up to more hard drive options and eliminated the need for hard drive beauty covers.

But bezels bounced back in a big way when IBM launched the PS/2 line with a new design language. Sharp, angular covers for the new 3.5 inch floppy and hard drives gave these desktops a distinctive look that set them apart from the aging XT and AT. But the Models 60 and 80 ditched the traditional desktop form factor in favor of a tower design aimed at server or workstation customers. These roles had an almost insatiable hunger for storage space, and so any serious machine needed the ability to install capacious full height hard drives or large format removable cartridge disk drives.

To accommodate both types of storage devices in the same chassis IBM devised a clever and innovative solution: they cut a hole in the front of the case. Now either internal or externally accessible drives can be installed in the same bay! IBM figured most customers would install a second hard drive in their towers so a blanking plate is included to keep noise in and dust out. For those that needed removable storage, all they had to do was pop out the blanking plate and slide in their half-height or full height drive. Easy as pie! Unfortunately that leaves you with a small problem: a big gaping hole in the front of the case! How unsightly.

So what do we do with big gaping holes? Why, we cover them up, of course! IBM offered several beauty covers for PS/2 towers starting with their 200MB optical WORM drive introduced alongside the Model 80 in 1987. A half-height bezel soon followed with the CD-ROM and 5 1/4 floppy drive kits. If you didn’t need a whole kit IBM’s service manuals listed part numbers for each bezel so you could call your IBM dealer and order one. But these weren’t the only bezels available! While doing some research to determine when IBM introduced CD-ROM drives as an option for the PS/2, I stumbled across all sorts of PS/2 tower bay bezels. Check out this dual half-height cover used in Tam Thi Pham’s model 80 could cover either a RAID array or clean up a CD-ROM. I wonder if this is IBM official because they bear a striking similarity to Model 95 blanks. Further searching uncovered more interesting half-height covers—even one for slimline drives! And the absolute king is this dual Bernoulli drive on this eBay listing with Iomega’s custom trim ring. This isn’t a hack—Iomega specifically aimed this drive at PS/2 towers! Neat.

But procuring these bezels thirty years after their heyday isn’t an easy task. Most of these trim pieces met their demise when the PCs they were installed in were scrapped. They rarely show up on eBay, and when one does you usually have to buy an expensive computer to get it. So when we can’t find rare and expensive doodads, we go for the next best thing: 3D printed replacements!

Since I still couldn’t find a real bezel by the time I started filming the OS/2 install video, I started looking for printable models. Alas, none were available at the time. A few months later I searched again and found one created by Mike1978, one of the admins of the UK IBM Collectors Discord and all-around cool dude. It would do the job, but my inner perfectionist wanted something that looked more like IBM’s bezels. But I’m not a 3D modeler—my CAD skills never graduated beyond the noodling I did in middle and high school drafting classes.

So I did what anyone lacking in a certain skill does: I posted about it on the interwebs. Back in December I posted on Bluesky asking if anybody local would be interested in collaborating with me on a 3D printable bezel design. My cool (and local) chum GutBomb threw his hat in the ring, but before we even worked out a time to meet up to make some test models I got another reply from the maestro of Micro Channel himself: Tube Time. He came out of nowhere with newly created models for half-height and full-height drive bezels. They even have the little dimples! I’m always surprised by the awesome generosity of this community. Time and again people contribute solutions and fixes to solve problems big and small and they often do it just for the thrill of it. Tube Time, you have my eternal gratitude.

Now that I had a model I could print a bezel. I don’t own a 3D printer, but I do have the next best thing: a buddy who owns a 3D printer! My pal Treble takes on almost any printing challenge and he was happy to oblige with these bezels. After sending him the models I got a box in the mail a week later with some samples. And would you look at that: it’s a perfect fit! It snaps right in to the slot and provides a clean lip above the drive just like the original. I couldn’t be happier!

Well, okay, I could be happier. The fit may be perfect, but the color is not. Treble was limited by the filament colors he had on hand and neither pure white nor bone white are much of a match for IBM’s ivory finish. Color matching is a common problem facing 3D printing enthusiasts, so much so that they end up commissioning custom colors. That’s exactly what Joe’s Computer Museum did with their new platinum filament, but it didn’t exist at the time we were working on these prints. Nevertheless, we can solve this problem with a little bit of rattle can-do attitude.

Painting 3D printed parts is one way to get a better color match, and the truly obsessed would special order a can of RAL 1013 spraypaint. But I wager that most people are fine with “close enough” and I was curious how close I could get with off the shelf paint. I made a trip to the local home improvement store to check out my options. Most of the shades were too white or too beige for my liking, and I wound up settling on Rust-oleum Heirloom White. I’ve been told that Ivory Bisque is a good match for an IBM, but I couldn’t find it. Regardless, after spending some time in the great outdoor paint booth I think the results are… acceptable. It’s brighter than I’d like and it’s a shade too warm, but it’s a closer match than the pure white or bone white filaments. It’ll do.

With a finished bezel installed I feel like my Model 80 is complete. I can have a real optical drive without an ugly gaping hole, and the color’s close enough for government computer work. I know some folks might wonder why the bezel isn’t fitting flush with the drive, and that’s because IBM’s original bezels weren’t flush fit. There’s variances among various drive manufacturers, and an overlap is a safer strategy than risking a gap. Genuine IBM bezels have nicer overlaps thanks to beveled edges and sometimes rounded corners and reproducing those features on a 3D printer is trickier than sticking to a flat edge. But the upside to 3D printing is that if you want a flush fit you can tweak the model to widen the opening by a millimeter or two to accommodate your favorite drive.

Now, I could stop here and say I’ve completed my quest. But I wasn’t done yet. The original bezels are so rare that increasing the supply is the right and honorable thing to do, and I wanted to match Tube Time’s generosity with some of my own. So I had Treble print out fifteen more bezels to bring to VCF East. I put a little label with an explanation on each one and dropped ten of them on the free pile. They were all gone by the end of Saturday, so clearly they were a hit! I also handed out a few to cool people like Steve and LGR and BigBadBiologist. I feel like a regular Danny Bezelseed here! … All right, that was a stretch. Still, I want these old boxes to look their best, and hopefully folks who picked up the freebies rushed home to install them into their old Micro Channel machines.

So if you find yourself in need of a 5.25 bezel for a Model 60, 65, 80, or Academic System, head over to TubeTime’s page on Thingiverse where you can download a half-height, full-height or blank panel model to fit most any drive you’d like to install. Your tower will thank you for helping it look its best.

How Much Video Memory Does A Mac Color Classic II Have, Anyway?

July 25, 2025 by Dan Vincent in text-only

You’ll often see it posted on the internet that the Macintosh Color Classic II and LC / Performa 550 use the same logic board. They both have a 33MHz 68030 CPU, support up to 36MB of RAM, and implement a full 32-bit data bus. But here’s a little weirdness for you: Apple's official spec pages (along with the Color Classic II / Performa 275 Service manual) say there’s 256KB of onboard video memory optionally expandable to 512K by adding a 256K VRAM SIMM. So how could it be the same board as an LC / Performa 550, which come with 512KB of onboard VRAM? These machines are expandable to 768KB by installing a 256KB VRAM SIMM, just like the LC / Performa 520 and LC III and LC III+. People trust these spec lists as a source of truth—they’ve been in the Apple Knowledge Base for decades at this point. Wikipedia cites them, Apple history sites reference them, databases like EveryMac rely on them, and I’ve seen folks wonder about this discreipancy. Did Apple really leave the extra VRAM off of the Color Classic II’s logic board? Or do both computers use the exact same board, which would mean those spec pages and manuals were wrong for over thirty years? I needed to find out.

If you guessed that a recent purchase is what sent me tumbling down this rabbit hole, well… you’d be right. At the MIT swap meet last Sunday I found a logic board for a color all-in-one Macintosh which was in desperate need of a bath.

All this was mine for the cool cost of ten dollars thanks to the Rhode Island Computer Museum.

A quick inspection showed a 33 MHz 68030 hiding under the dust and grime and wondered if it was from an LC 550 or maybe something else. I did some on-the-spot Googling for LC 550 and Color Classic II boards, and sure enough it looked exactly like the board on Recap-a-Mac’s Color Classic II guide, complete with matching ROM and half-AA battery! ID number 820-0368-A was printed on the guide’s board and that matched up with the ID number pritned on my board. Meanwhile searching for LC 550 boards found a board with ID number 820-0595-A which looked similar but had a 4.5V alkaline battery. That lined up with some old memories of handling Performa 550 boards with an alkaline battery. I thought I’d found a diamond in the rough so I paid the ten dollar asking price and brought it home. After a rinse, scrub, and dry-down I popped it in my Color Classic and it booted just fine. But a curious thing happened when I opened the Monitors control panel: I could choose thousands of colors instead of 256! I thought Color Classic IIs only had 256K of VRAM, and I had yet to transfer over the VRAM SIMM from my old Color Classic board. So how was I getting thousands of colors? Did I actually get an LC 550 board, or was something else afoot? What, exactly, is a Color Classic II board anyway?

First, let’s establish a baseline. The original Color Classic definitely has 256KB of onboard VRAM. Just look at the original board from my own Color Classic here:

Macintosh Color Classic Logic Board with VRAM highlighted

There’s two 128Kx8 chips fabbed by Micron, and you can see the data sheet for these MT42C8128 modules here. Adding a 256KB VRAM SIMM brings it up to 512KB total. For another example of all-in-one color Mac’s VRAM here’s my Macintosh TV board.

Macintosh TV Logic board with VRAM highlighted

There’s four Mitsubishi M5M482128J 128Kx8 chips, bringing the total to the spec sheet maximum of 512KB VRAM.

Now here’s my freshly recapped Color Classic II board (or possibly LC 550 board, more on that distinction later).

Macintosh Color Classic II / LC 550 logic board with VRAM highlighted

And sure enough, there’s four Mitsubishi M5M442256AJ 256Kx4 dual ported VRAM chips for a total of 512K VRAM. I went back and double checked Recap-a-Mac’s guide, and that board also has 256Kx4 VRAM chips. I never paid attention to the chip markings before, but that’s what they are. Bruce at Branchus Creations runs Recap-a-Mac and he's a pretty smart fellow who’s refurbished his share of old Macs. He says on the Color Classic II page that he's recapped customer CC II boards, and I have no reason to doubt that because he’s recapped Color Classic II boards on a live stream! The board in the stream might even be the same one pictured in the guide.

But a sample size of one isn’t enough for me, so I searched for more examples of “real” Color Classic IIs. I remembered that Colin from This Does Not Compute imported a Performa 275 from Japan, and luckily he recorded a video about it. I scrubbed through the video looking for logic board shots and saw his board had all the same traits as my board. It was ID number 820-0368-A with four VRAM chips, a half-AA PRAM battery, and the ROM identifiers matched too. And since he imported this machine from Japan I’m pretty confident that the board is original to the machine. I also found another video where Tom // Drives unboxed a Japanese Color Classic II. Freeze-framing his logic board removal shows another board that matches Colin’s, Bruce’s and mine.

Now I needed to look at LC 550 boards. You might remember my earlier Googling showed an LC 550 board equipped with a 4.5V battery, and because I had handled similar boards I thought that’s what an LC 550 board was. Said board was posted on TinkerDifferent by user Fizzbinn and its model ID is 820-595-A. But when I could search at home with less time pressure I noticed a bunch of LC 550s with boards that looked exactly like the CC II board and ID 820-0368-A. There’s a Worthpoint eBay listing archive with photos of an LC 550 that match the Color Classic II boards. Same with Apple Rescue of Denver’s listings for recapped logic boards. Same with Ancient Electronics and their Color Classic which was equipped with a board pulled from an LC 550.

As it turns out there’s two versions of the LC 550 logic board—Fizzbin’s board has REV B printed on the serial number sticker. Rev B board differences are the ID number (820-595-A), the battery header, and a velcro pad to mount the alkaline battery. I’m betting these rev B boards were introduced around the same time Apple switched the LC 550 to tray-loading CD-ROM drives after the launch of the LC 575 in February 1994. It’s a reasonable guess to say that the closer your build is to 1995 the greater the likelihood your LC 550 will have a revision B board. I plumbed YouTube again and found videos where people pulled out or reworked their logic boards. CamelProd666 pulled a revision B board out of a tray-load Performa 550. Matt’s Workbench pulled an 820-0368-A board out of his caddy-load LC 550. Bruce at Branchus repaired an LC 550 Revision A board that looks identical to the Color Classic II. BigBadBiologist’s recent recap of a 820-0368-A board is also identical to the Color Classic II. And somehow there’s even Rev A boards that have what look like factory bodges to use a 4.5V battery, like this one seen in RetroTechIIfx’s video or on powercc.org.

Lastly I decided to look at LC 520s just to be thorough. An LC 520 logic board is supposed to be the same design as the LC 550 except a 520 board has a 25 MHz 68030. After investigating some examples of LC 520s I’d say that’s a true statement. LC 520 boards have the same board ID of 820-0368-A as revision A LC 550 boards. Garth Beagle restored an LC 520 for Marchintosh and his board looks exactly like a Rev A LC 550. The same goes for this new old stock Apple Service kit for an LC 520 logic board that was archived on WorthPoint. And a 520 board restored by CayMacVintage also looks identical to a 550. The only difference is the 25MHz CPU.

Lastly, ROMs aren’t a differentiating factor between all these boards. Every 520, 550, and Color Classic II board I saw had Hi and Lo ROMs marked 341-0534 and 341-0533 respectively. That renders the 520, 550, and Color Classic II identical from a software standpoint. So how can you tell an LC 520 board apart from an LC 550 or Color Classic II?

Apple’s June 1994 Service Module Identification Manual show the LC550 and CC II logic board as part number 661-1830, and a note is attached to verify that the CPU is 33 MHz. An LC 520 board with a 25 MHz CPU is Apple part number 661-0812.

Color Classic II / LC 550 Identification

But another way to tell LC 520 boards apart from LC 500 boards is the serial number schema. The LC520s I’ve seen have serial numbers formatted like EZxxxxxxDY7 and DWxxxxxxDY7 depending on the board revision, while Rev A Color Classic II and LC 550 boards use a DWxxxxxx1X9 schema. I checked all the boards I found on the web and in videos and confirmed that all the Color Classic II and LC 550 boards followed the DWxxxxxx1X9 schema. As for the rev B LC 550 boards with the 820-595-A board ID, they also used DWxxxxxx1X9. Given the same serial schemas and Apple replacement part numbers those are two more points in favor of the Color Classic II and LC / Performa 550 using the same logic boards with 512KB of VRAM.

Copy and Paste, Sloppy and Waste

Given the evidence in front of me I think it’s safe to conclude that the same 33 MHz 820-0368-A logic boards were installed in both the LC / Performa 550 and Color Classic II. So there’s no difference between “a Color Classic II board” and “a LC 550 board” because they are one and the same. Apple didn’t make specific logic boards with only 256KB of VRAM for the Color Classic II, which means Color Classic IIs have always come with 512K of onboard VRAM expandable to 768K and that Apple’s official spec sheet and service manual are wrong and wrong.

I’m not surprised that such a goof-em-up happened because the Color Classic II and Performa 275 were only sold in Asia, Canada, and some PAL countries. Given the overwhelming plethora of slightly different models Apple was churning out at the time it’s no surprise that mistakes could be made when updating a spec sheet for a speed bump. I bet there’s other mistakes lurking in the spec sheets for other 1990s beige Macs. It’s also possible that Apple intended to make a variant of the logic board for the Color Classic II with two fewer VRAM chips but backtracked after realizing that producing another board variant with all the associated logistics was more expensive than the cost of two chips and then forgot to update the spec sheet. That seems unlikely to me, though—Occam’s Razor and such.

But the service manual is a different story. Almost all the specifications were clearly carried over from the original Color Classic manual and never updated to reflect its new capabilities. The processor speed is updated to 33MHz, but the bus width is still listed at 16 bits. It incorrectly lists the maximum amount of memory as 10MB, the data bus as 16 bits, and the PDS slot as 96 pins along with the incorrect VRAM specs. The RAM upgrade steps still show dual 30 pin SIMMs which would make an unfamiliar tech go “wait a minute.” More stale information is in the sound section. The original Color Classic logic board’s sound output jack only supports mono output duplicated across both channels while the Color Classic II supports true stereo sound with discrete left and right channels. I validated this in two ways. First, I checked the System 7 Sound control panel and the original Color Classic board shows mono, while the Color Classic II / LC 550 board shows stereo. Second, I played a left / right stereo test sound file with SoundApp. The original Color Classic board played a duplicated mono downmix in my headphones while the Color Classic II / LC 550 board played true stereo sound. Incidentally, this means that an internal stereo speaker mod won’t work with an original Color Classic logic board. If you plan to do a stereo mod you must also plan a logic board upgrade.

It’s obvious that whoever made this manual simply did a search and replace for Color Classic to change it to Color Classic II / Performa 275. But at least they updated the logic board part number to 661-1830, which matches the logic board part number for the LC 550 in its service manual.

Bit Depth, Resolutions, and RAMDACs

The LC 550 / CC II Logic Board with 768K of VRAM installed in my Mac TV is capable of thousands of colors.

The result of the spec sheet’s inaccuracy meant Apple actually undersold the Color Classic II’s video capabilities. Since it has 512KB of VRAM standard it can display thousands of colors without spending extra money on a VRAM upgrade. Installing a 256KB VRAM SIMM doesn’t actually do anything for a stock Color Classic II because of limitations in its Andberg ASIC’s video generator and the Ariel RAMDAC. Introduced with the Macintosh LC, the Ariel chip replaced the Brooktree Bt478 CLUT/DAC (also known as VDAC) used in the IIci and IIsi. Ariel variants are found in many 030 machines with built-in video like the IIvi, IIvx, Performa 600, LC III, the DuoDock, and the 520/550. The Mac TV has a slightly different version called Ariel+. Overall Ariel isn’t that different from the VDAC since VDAC was a Brooktree Bt478 and Ariel is pin-compatible. Whether Apple rebranded another Brooktree part, licensed IP from Brooktree to design their own DAC, or commissioned Brooktree to design Apple-secific RAMDACs isn’t really clear.

Ariel improved upon the VDAC by introducing 16-bit color depth to display thousands of colors, but that’s a hard ceiling for that DAC (and the ASICs that generate the video). Adding more VRAM to an Ariel system can enable higher resolutions, but not higher bit depths.1 Speaking of higher resolutions, a LC III with 768KB of VRAM can display 832x624 at thousands of colors, but a Color Classic II or LC 550 can’t because their internal monitors aren’t multisync and the analog board’s sense connections tell the logic board what resolution to use. Yes, I know the Color Classic with a IIe card can switch to the 560-wide Apple IIe display mode, but that’s best saved for a bonus section.2

Other contemporary entry level Macs like the LC 575, LC 475, and Quadra 605 were also limited to 16-bit color despite the logic board’s ability to accept 1 MB of VRAM. That’s because Antelope’s RAMDAC is also limited to 16-bit high color just like Ariel. Originally developed for the Quadra / Centris 610, 650, and Quadra 800, the Antelope was Apple’s attempt to cut cost while eking more performance out of the Direct Access Framebuffer used in the Quadras 700 and 900 which did support true color modes. According to Apple displays engineer Dale Adams, Antelope dropped the 32 bits-per-pixel true color modes because they had high VRAM requirements and added significant cost to the chip design. Customers who demanded true color support would skip right past the built-in video of the Quadra 700 or 900 and buy an expensive dedicated graphics card. Onboard video support for true color graphics wouldn’t return to Apple’s cheapest Macs until the Power Macs 4400, 5500, and 6500.

There is a practical application for installing more VRAM in a Color Classic II, and it’s a VGA mod. After performing a VGA mod a Color Classic II with 768KB of VRAM can display thousands of colors at 640x480 just like an LC 550 driving its native monitor. I have a hunch that such a combination is rare, since the type of people willing to put in the effort for a VGA mod are also the same type to seek out a Mystic upgrade. Or if you’re like me and have a Hook-style chassis floating around you can leave the VRAM SIMM installed and just swap around as your mood dictates. Regardless, I think you can confidently say that the Color Classic II and LC 550 logic boards are in fact identical. Now who can I call at Apple to get that spec sheet fixed?

The following is technical bonus content that could be its own post but I’m including here because nerds might want to read it. You’ve been warned!

A Dive Into The Mac’s History of Onboard 24-Bit True Color Video

Footnote 1: To write accurately about the capabilities of the LC 550 and Color Classic's video modes I had to read about Apple's video chip and DAC history.

Other contemporary entry level Macs like the LC 475, LC 575, and Quadra 605 were also limited to 16-bit color despite their logic boards’ ability to accept 1 MB of VRAM. These systems used the Antelope RAMDAC—also known as the AC843 (Stripped) RAMDAC—which is also limited to 16-bit high color like Ariel. Antelope emerged from efforts to cost-reduce the video architecture of the Quadra 700 and 900 into something less expensive for the upcoming Quadra / Centris 610, 650, and Quadra 800. According to Apple video engineer Dale Adams, stomers who needed true color support didn’t use the built-in video anyway and bought expensive accelerated graphics cards instead. Adams also wrote that omitting the 24-bit true color modes and convolution support simplified the RAMDAC and reduced the amount of VRAM required which lowered the cost of a base system for those who didn’t mind 16-bit color while those who really needed true color could spend their money on an add-in card.

So what’s the timeline of support for true color on Macintosh built-in video? Let’s start with the Quadra 700/900, which supported 24 bit color and whose video was provided by the Direct Access Frame Buffer chip and the AC842 RAMDAC. The DAC’s even got a fun code name: AC/DC! The Quadra 950 updated these chips to the DAFB II and a revised AC/DC RAMDAC numbered AC843 (see page 6). These improved chips boosted performance, introduced support for 1024x768 resolution, and added more bit depth options to the existing resolutions. For the Quadra / Centris 610, 650, and Quadra 800 Apple integrated the DABF into the djMEMC memory controller and replaced the complete AC843 AC/DC with the stripped AC843 Antelope. The djMEMC was further simplified into the MEMCjr found in the LC 475, 575, and Quadra 605. One of the changes was removing extra VRAM bank connections in the djMEMC that were unused in the 610/650/800. Theoretically a Quadra 650 could be hacked with more VRAM and a full AC843 to get 24-bit video modes, but while it’s been discussed over at 68KMLA I don’t think anyone’s actually tried it.

To bring this back to the Color Classic II and the LC 550, their video setup is very similar to these Quadras from a high-level perspective but use 68030 generation tech. Video generation and memory timing are handled by the Ardberg ASIC (see page 20) which drives the Ariel RAMDAC to generate analog signals for display on a CRT. Since Ardberg is a derivative of the LC III’s Sonora ASIC, it also uses Sonora’s video architecture (see page 7). You can draw a straight line from the V8 gate array which handles video in the original LC and LC II to the VASP chip in the IIvx and IIvi, to Sonora in the LC III, then Spice in the original Color Classic, and lastly Ardberg in the LC 520, 550, and Color Classic II.

Onboard video support for 24-bit true color modes would return with the launch of the Quadra 660AV and 840AV thanks to their CIVIC chip (see page 34), but those were expensive machines that replaced the top end Quadra 700 and 900/950. The Power Mac 6100, 7100, and 8100 used a funky DRAM-based system for onboard video which was limited to 16 bits per pixel, but most 6/7/8100 series Power Macs had either an Apple PDS VRAM expansion card or an AV option card which carried over Civic and other chips seen on the Quadra AVs. Depending on the amount of VRAM on the expansion card the system could support true color output. The AV Option card added 2MB of VRAM and a Civic chip for true color support along with video capture hardware. When desktop Power Macs moved to PCI they ditched the PDS and DRAM setup entirely for new chips that supported true color at a variety of resolutions, provided you installed enough VRAM. The 7200 and 8200 used the Platinum and Iridum combo with the DACula RAMDAC (see page 14) while the 7500, 8500, and 9500 use a duo called Chaos and Control to drive the RaDACal RAMDAC (see page 20).

Meanwhile clone makers were using off-the-shelf video controllers from the likes of ATI and were blowing Apple away in terms of resolution and color capabilities in less expensive machines. Apple’s in-house clone, the Power Mac 4400, also used an ATI chip and has 24-bit true color capability(see page 38).

Consumer and education Macs were a different story. Ariel and Antelope were used in LCs and Performas until the launch of the Performa/LC 580 and 630. These Macs introduced the Valkyrie chip which I would consider to be a complete “video” chip. Apple separated the frame buffer and memory controller functions out of the system ASIC and combined them with the RAMDAC functions to form the Valkyrie. It wasn’t a particularly good chip, mind you, but it did have a neat acceleration that gave Marathon a performance boost. It also adds support for scaling and overlaying video for the Apple TV/Video System. Because Valkyrie is a consolidation of the MEMCjr DABF implementation and the Antelope DAC, it shares the same limitations of 1 MB of VRAM and a maximum color depth of 16 bit.

Valkyrie was eventually replaced by the Valkyrie-AR, and that chip’s RAMDAC does support 24-bit color modes according to its design documentation. But it needs 2MB of VRAM to display 24-bit color at 640x480 resolution and as far as I know it never shipped with that amount. The models that used that chip—the Power Mac/Performa 5400, 6360, and 6400—only ever had 1 MB of VRAM. Apple eventually put ATI graphics chips and 2MB of VRAM into the Power Mac 5500/6500 which finally closed the door on Mac desktops that couldn’t support 24-bit true color.

The Hidden Apple IIe Video Mode

Footnote 2: This bonus section delves into the special Apple IIe monitor mode.

The Apple IIe mode is handled differently on the Color Classic and LCs with the 12-inch RGB Display versus the all-in-ones with a 640x480 display or a Quadra 605/LC 475. According to page 23 of Apple’s Color Classic developer tech notes the IIe card generates a video timing signal for the Apple II display and sends it through pin B31 of the PDS slot when installed in an LC, LC II, or original Color Classic. Starting with the LC III the Apple II video mode timing signal is generated on the logic board by the Omega clock chip (see page 32). Omega’s clock functions were moved into the Spice ASIC on the Color Classic and the Ardberg ASIC on the LC 520, 550, and Color Classic II (see page). The 68040 machines—the LC 475, 575, and Quadra 605—rely on a separate clock generator chip called Gazelle (see page 45). So how does each computer’s video system handle that timing?

For the Color Classics it looks pretty simple: the IIe video mode has a faster horizontal clock to get the extra horizontal pixels. The vertical timing doesn’t change because the Color Classic already runs at 384 vertical pixels and 60Hz (see page 23). It isn’t explained why the user can’t manually select this mode on the Color Classic (or CC II) despite its ASICs having the ability to generate the timing signal without the IIe card, but I have a hunch as to why. The Apple IIe uses non-square pixels, and since the geometry of the monitor doesn’t change when switching between these two modes the result is narrower pixels that look “correct” for Apple II video in the same geometry. If you used the 560x384 mode for a Mac desktop you’d need to reduce the vertical size of the display to maintain square pixels. The Color Classic’s monitor is small enough that some users might not care for the resulting empty space that would leave on their monitors. The Color Classic's geometry controls are also hidden inside the case, which makes it difficult to adjust them on the fly! IMO the user should be able to make that choice, but Apple made it for them.

This is also the situation when using the Apple 12 inch RGB display with the LC, LC II, and LC III. When using that display those computers output a 560x384 at 60Hz mode for the Apple IIe card, just like the Color Classic (see page 20). What if you’re using something other than the 12 inch RGB display? Keep reading!

The 14 inch all-in-ones are slightly different. Their normal display mode is 640x480 at 67 Hz. On the LC 520 and 550 the monitor and analog board support switching to a 560 x 384 mode at 60 Hz. The LC 520 technote describes the timing for this secondary display mode in diagrams onsee page 27.

I have to imagine there’s some kind of geometry compensation on the 520/550 analog board for this mode, but I don’t have an Apple IIe card to validate this. The Macintosh TV would have the same rectangular pixel conundrum when converting television video, but I don’t have the skills to test how this ends up working on the analog board side of things.

This differs from the LC 575, where the developer technote says the following on page 30:   “When the Apple IIe Card for the Macintosh LC is installed in the expansion slot, the computer generates an Apple IIe video display with 560 by 384 pixels. The Apple IIe display uses the standard video timing but displays only a 560-by-384-pixel area in the middle of the display.”

That’s definitely different than the 520/550. So how does that work? If the video timing doesn’t change, how does it display the lower resolution? The solution is a bordered mode. The LC 475/Quadra 605 technote gives us a clue on page 28 and the LC 575’s architecture is largely the same as those machines:

“The Macintosh LC 475 and Macintosh Quadra 605 computers can provide a 560-by-384 display on any size monitor; only the 12-inch monitor requires modifed [sic] timing parameters to provide the 560-pixel display width.”

Ah, that’s the ticket! It’s only sending native 560x384 at 60Hz video if it knows it’s using the 12 inch RGB Display. Other monitors get what amounts to a bordered mode running at 67Hz. This even seems to be the case for the earlier LCs if you use a IIe card and a monitor other than the 12 inch RGB. mg.man on Tinker Different posted as much showing an LCD connected to an LC II running the Apple IIe mode at 640x480 at 67 Hz. I’m not sure how running at the 67 Hz refresh rate affects the speed or playability of the IIe card; someone who’s an expert on this could help fill me in.

What does this mean for modified Color Classics? It probably depends on how you modded your machine to run 640x480. If you did the traditional 60Hz VGA mod and used a LC 575 logic board you get the bordered screen behavior like the LC 575 as mentioned by JDW on the same TinkerDifferent thread. I don’t have a Mystic board to test but I’m curious if unmodified systems that run at 512x384 still switch to the native 560x384 60 Hz mode.

But how about users who upgraded a Color Classic II or installed an LC 520/550 logic board? I haven’t found any examples of anyone doing this on the web. My gut says if you do the 60 Hz VGA mod, you’ll still get the bordered display behavior. But if you did the 67 Hz high voltage mod—basically telling the logic board you’ve installed it in an LC 520/550 chassis—perhaps it would behave like the LC 520/550 and switch to the 560x384 60 Hz mode. I’d love to validate these scenarios and update this as necessary if I ever get a hold of such a machine.

The VCF East 2025 Report

July 22, 2025 by Dan Vincent in podcast

Welcome to the tech antiques road show! Here, in Userlandia, we’re back in New Jersey to explore every inch of Vintage Computer Festival East.

Where do you want to go today? If it’s a meatspace meetup with your favorite group of geeks, you’re in luck. Conventions for retro gaming and vintage computing fans are in vogue because we all know it’s more fun to experience this old tech in person. And with a new year comes a new edition of the venerable VCF East which is back with a fresh slate of exhibits, talks, and a consignment hall packed full of vintage treasure.

It’s hosted at the familiar confines of the InfoAge Science Center in Wall Township, New Jersey. A ticket buys your way into the VCF show as well as the associated InfoAge museums and exhibits. Single day and three-day passes are available, and like all things in life they’re a little more expensive these days. My pre-sale three day pass with fees totaled $103—$15 more than last year. That’s still less than the at-the-door price of $128, so it pays to pre-register. But I was less annoyed about a modest price increase than I was about the return of wristbands. The first 150 three-day pass holders to check in got a badge, but they were still required to wear a specific day’s wristband to get in the show. Boo-urns. If you’re ponying up three digits for a three day pass, you deserve a badge and the privileges therein.

After securing entry you’re free to explore five exhibit halls packed with dozens of unique computer exhibits. Your first stop in Hall A is Alex Jacock’s laptop extravaganza, where you could buy one of his many sensible ThinkPads for a sensible price. Flanking him was Happy Hardwear, offering an array of circuit board inspired jewelry crafted in a cute pixel style. Setonix Synth brought a sampling of their synthesizer kits, which you could buy to construct a rackmounted synth setup to create out of this world sound.

It wouldn’t be hall A without the folks from Tech Dungeon brightening things up—and they’re literally doing that with their new light-up marquees! Combine them with one of their joysticks to make your home computer feel a little more like a home arcade.

Long Island’s American Retro Shop is also back with their selection of light-up keyboards for many of your favorite retro computers. Or pick up a Retro Chip Tester Pro and finally figure out which chip is trapping your Commodore 64 at a black screen.

Do you like cute things? I love cute things, and these little mini Commodores made by Yankee 3D Forge amuse all my adorable senses. You could buy a C64 with 1702 monitor and 1541 disk drive to give your figures and dolls their own retro setup!

Back in its usual spot in a corner of hall B is the RCA Cosmac display. It’s a perennial feature and the hosts do a good job of keeping things fresh by showing off different games and materials.

Today’s game consoles share a lot of technology with PCs, but that’s not a new idea. Consolization shows three past attempts at porting computer architectures to a cost-reduced console format: the Commodore CD32, Fujitsu’s FM Towns Marty, the Apple Pippin, and Microsoft’s original Xbox. These machines didn’t dethrone Nintendo and Sony as the kings of console gaming, but they were cool in their own ways.

Eli’s software encyclopedia is back with more tables to hold his bounty of big box software. With a three-for-twenty dollar deal Eli found many willing adoptees of new old stock games and productivity packages. This year’s unique selections include this Felix the Cat Cartoon Toolbox which can turn your Mac into an animation studio!

What’s all that clatter? It’s John Lucas and the Baudot Basement… bau-dot? Correct me if I’m wrong, John! This fleet of teletypes hammered out news feeds throughout the weekend.

Are you bonkers about Betamax? Dave’s Retro Video Lab is back this year with Beta taking center stage. Tapes, decks, and cameras were on hand to tell the story of Sony’s groundbreaking video tape technology.

Pen plotters have seen a resurgence in the retro scene and they’re getting some excellent rep here at East. Paul Rickards is back with his plotters at The Plot Thickens, where his machines are drawing live at the show. And if you like what you see you can bring home prints of his colorful and quirky designs. Next door was Kate Szkotniki with Apple’s 410 plotter. People throw the word “rare” around too much for Apple gear but this is a legit rarity, given how few were sold and survived to this day. Luckily for you an Apple IIe was driving it throughout the weekend for lovely pen drawing action.

Business Computer: Japan is Quinn Quirble’s tribute to the NEC PC-98. These PCs from the land of the rising sun feature home-grown graphics and sound capabilities which made for unique business and gaming applications.

Retro music composers will be interested in Michael Weitman and Stefany Allaire’s Foenix 256 digital audio setup. Composers who want to make retro sound in a retro way can use both the Foenix’s built-in sound chip and external MIDI devices like this Moog synthesizer to create their perfect soundscape.

We take multiple displays for granted today, but there was a time when they were rare and expensive. The Forgotten King’s Power of Two Displays shows us practical applications for multiple monitors in the early PC era. One of the most common was AutoCAD, which used a monochrome display for its command interface and a CGA display for your drawings.

Are you hungry for HeathKits? In this corner of the hall there were two tables showing off Heath systems. Alex Bodnar brought boxes of programs to show all the things his H89 setup could do. And for more Heath goodness there’s Darrell and Evelyn Pelan, Joe Travis, and Glenn Roberts. Their booth was overflowing with multiple Heath and Zenith systems modded to the hilt, like systems with speech and sound synthesis. Or pull up a game of Galaga on an H8 with MSX capabilities!

It might’ve been gray outside but it was very colorful at EmmyBear Retro, with their combination of electronics and handcrafted cloth gear. After buying a ZuluSCSI or Greaseweasel why not pair it with one of their bags or mousepads in cool new retro patterns?

Our Dutch friends from the Home Computer Museum are back with a new sampler of European computing, featuring an Amiga CD32 and a Commodore PC1. Check out the Exidy Sorcerer, an American PC that found a lot of success in the Netherlands. And don’t forget to munch on some stroopwafels! I’ll make a trip out there some day.

Also, I need to give a nod to MacEffects. They had a table, but unfortunately by the time I went to go film them on Sunday they were already gone. Sorry I missed you, guys!

If the show hadn’t settled on the sound of retro as its theme it could’ve focused on the many microcomputer anniversaries taking place in 2025. The tour of the decades in Hall C starts at 1975 for a celebration of fifty years of Altair. Hosted by William Sudbrink, S-100 systems like the Altair might be known for their switches and terminals. But when equipped with the right combination of boards they could play sound and draw graphics.

Jump ahead ten years to 1985 at 40 Years of the Amiga and you’ll be amazed at the advancements made in a decade. Eight Amiga minds banded together to put the life and times of Commodore’s Amiga on display. You can track some mods, boing some balls, and even toast some video. Hot stuff.

Once you’re done marveling at the Amiga as it was, you can check out New Old Computers to see what it could be today. If your old Amiga needs a new motherboard then these modern recreations can bring your old friend back to life.

Of course the Amiga isn’t the only PC turning 40 years old in 2025, because its friendly rival the Atari ST is also celebrating its ruby anniversary. Peter Fletcher and Bill Lange’s booth set up 1040 and 520STs along with Mega STs and a rare Falcon 030. MIDI music, desktop publishing, and CAD drawing are just some of the applications you could try on these Atari desktops.

Our next stop on the timeline is ten years later in 1995 for Start Me Up: Thirty Years of Windows 95. This time capsule curated by Alexander Pierson and Katherine Alskog combines software and pop culture for a celebration of peak Microsoft. Take a seat and you can explore multimedia masterpieces and productivity apps powered by period appropriate beige hardware. But if that’s too corporate for you, don’t worry—you can go back to school at the Pittsburgh Classic Mac Lab with a row of LCs and compact Macs running edutainment titles and KidPix.

One of my favorite exhibits was Dan Forsythe’s Fuji Apples: Macs from Japan. The Macintosh was popular in Japan, and Apple and the clone manufacturers created specific machines to cater to the market, and many of them are on display here. But the cherry on top is this pair of Macintosh manga geared towards new users. I loved this particular one featuring new Mac buyer Lisa and her boyfriend Hoshio—that’s Japanese for Star—learning how to navigate the Mac desktop thanks to their magical bear friend Alto. Now that’s a mangaka who knows their GUI history. Had I known these existed when I was visiting Japan last year I would’ve scoured Bookoffs to find them! Thankfully a scanlation is in the works for us English readers.

Are you looking for a computer musical instrument? Then pick up a Commodore 64! Brigitte and Dan Laskowski combine the C64’s powerful SID chip with a MIDI-controlled Yamaha YM2149 to make some soulful synthesized songs.

In the blinkenlights department we have Experiencing the ENIAC presented by Brian Stuart. This miniature replica of the grandfather of digital computers features a live representation of its calculations dancing across a glowing orange matrix. It’s the perfect display to prepare you for the 80th anniversary of ENIAC’s debut next year.

Core64 is a regular exhibitor at VCFs across the country, and they’re back at East with their simulator kit. Pick up an interactive core memory module and test how computers remembered things in the mainframe era.

For some heavier metal, Maki Kato and David Griffith’s Motorola 88K exhibit returns this year with a new focus on the Data General AV5500 running DG/UX. And Chris Randall and Mike Rieker brought this PDP-8L back to life with a new FPGA based disk controller to make it sing some tunes for the crowd.

If you’re an exhibitor and the computer at your table goes pop, what do you do? I would head over to Amiga of Rochester for some expert advice. They were equipped with soldering stations to work on boards and troubleshoot issues all weekend long.

What was your first experience with computers? Justin Morgan’s Computerized Start wants you to think about your entry into technology by putting his first computer and other systems on display. After you’ve had your fill of classic Mac and Commodore action, try out this MSX synthesizer program by tapping the keys on a MIDI connected keyboard.

Video game music isn’t just a soundtrack, it’s essential to the interactive experience. A cavern full of consoles isn’t a requirement to check out game music you might’ve missed because the folks at The Sound of Emulation can show you how to explore the wide world of game sound without original hardware.

Brian Boellner and Ryan Schiff’s Early Microcomputers is here to remind you that PC history didn’t start with the Apple II or IBM PC. Check out interesting and influential machines like the Sol-20, a Kenbak-1, and the Mark 8 microcomputer. They also had a selection of restored machines for sale, like these recapped Mac IIci’s.

Now here’s a booth after my own heart: Ryan Burke and System Source’s IBM PS/2 & Micro Channel Mayhem. A monstrous PS/2 model 95 boasting a 66MHz Pentium complex will draw you in, but you’ll stick around to check out this cute Edquest all-in-one running Windows or Doom. Make sure to shuffle through some Micro Channel cards, like these Intel systems-on-boards intended for RISC workstations. But what’s this—a Model 80 with an authentic IBM CD-ROM drive bay bezel? It sure is! I’ve been working on something relating to replicas of these inserts—stay tuned to the channel for more—and Ryan had not one, but two different styles of these unobtanium beauty bezels for half-height drives on hand.

Over here is Mike Loewen’s “It’s not Unix, it’s Cromix!” Cromemco developed their own Unix-like multi-user OS to give their S-100 systems extra sophistication. Back in the day this setup could give you the flexibility of a mini for a lot less money.

It wouldn’t be VCF without the FujiNet folks, and they’re back this year with a new RS-232 variant for PCs. It joins the other versions in the stable which were on hand for demonstrations like the one on this Apple IIGS.

A short walk away over in section D is MDCon. The traveling Sony Minidisc fans brought a chunk of their collection for attendees to see rare and interesting players. Don’t forget all the docks and accessories, which are the spice that make this exhibit so authentic.

Communication is a beautiful thing, and this collection of Sun, SGI, and Windows boxes are all networking in harmony in the Y2K intranet set up by Stephen Mayo. A Cobalt Qube serves up files while this SmartBits tester monitors performance.

Finally we arrive at Hall E, the new exhibit expansion for 2025. In previous years this space was used as a bunk room for those wanting to stay overnight. Well, that’s not happening anymore, and this opened up room for five exhibits. My pals at Totally Normal Computing filled one corner of the room with the wackiest ways to use a Mac. Sean from Action Retro set up two compact Macs that couldn’t be more different. Ever thought about gaming in grayscale? It’s possible thanks to an M4 Mac Mini transplanted into this transparent SE! And this 512K running Laurie Spiegel's Music Mouse lets you improvise fun tunes by moving the cursor around the screen. Next is everybody’s pal Mike featuring two special G5s: a prototype iMac G5 with a card reader and a working liquid cooled Power Mac G5. And Steve from Mac84 brought not one, but two multiprocessor Mac clones. The dual UMAX SuperMac is a special computer, but even that machine must bow in the presence of the fabled DayStar Genesis quad. Both machines spent the weekend doing what they do best: 3D render demos and Photoshop bake-offs.

Hackers love stuffing a computer where it doesn’t belong, and Joe Marlin accepted that challenge. Here at It’s Just A PC he’s hacked an NES into an Apple II just because he could. It’s an interesting counterpoint to the comptuers-turned-consoles we saw earlier, which makes you wonder where to draw the line between computer and console.

Nick Mailloux’s luggable computer exhibit returns with an interesting new addition: a Minitel. Thanks to enterprising enthusiasts Minitels can be brought back online and you could experience old-school online services here at Nick’s table.

Video editing and playback are capabilities users take for granted today, but in the nineties most computers needed special hardware to handle video. John the BigBadBiologist featured four different computers capable of capturing video and television, including a Mac LC II with the SuperMac Video Spigot, the Macintosh TV, and a Performa with Apple’s TV/Video system card. But don’t count out Windows machines—a Compudyne 486 shows that Windows users could play director too.

You can always count on DosDude to deliver a unique exhibit, and this year it’s cellular phones. He set up a miniature cell network for folks to go hands on with handsets from Sony, Nokia, and Apple. And it wouldn’t be a DosDude exhibit without rare prototypes like an Intel Android phone and a Qualcomm Surf.

And last in Hall E is the massive Genericable booth hosting their ongoing efforts to preserve our memories of 1990s cable networks. Their collection of preview guides, descramblers, weatherstars, and cable modems is growing so much that they’ve expanded vertically. Talk about a tower of tubes!

Hall E certainly is a welcome expansion, bringing about 15% more exhibits to the show. But it wasn’t without it’s foibles. Exhibitors were tripping circuit breakers all weekend, likely because this dormitory room wasn’t wired to the same level as the other exhibit halls. I have a hunch that next year’s placements will be ones that require less power or get briefings on which outlets are on which circuit. And poor John the BigBadBiologist saw his gear crash to the floor when one of his tables spontaneously collapsed. Fortunately nothing broke and everything was back up and running within short order. Crisis averted!

Events and Panels

After you’ve exhaustively explored every exhibition at VCF East there’s still a whole ton of show left for you to see. If you’re into hardware the Computer Destruction Lab will be at the top of your list, with Glitchworks hosting three days of DIY makerspace. You could assemble the kits you purchased from exhibitors or pick up one of Glitchworks’ own kits. And if you don’t know what you’re doing, soldering classes will teach you how to put it all together without burning your boards.

When you’re finished hacking hardware you can head next door to the VCF classroom. Tandy’s Color Computer was the platform for this year’s sessions, with BASIC and assembler programming classes to teach you how to get the most from the CoCo. But unlike previous years there weren’t rows of CoCos set up for students to work on during class, and only two examples were set up at the edges of the room. The VCF class guide made it clear that attendees were expected to bring their own laptops to run an emulator, so I’m sure that the syllabus was planned around it. There are other classes that don’t cater to the headline platform, like ones for FujiNet and chiptune composition, and having clear tables might’ve been helpful to those sessions.

In addition to the VCF exhibits and classes there’s a wide range of InfoAge museum exhibits open during the show. Learn about the legacy of Camp Evans’ radar research at the newly renovated World War II radar exhibit. Or check out the New Jersey train museum, with scale models of lots and lots of trains! But the obvious attraction for VCF attendees is the VCF museum. It’s been reorganized into a new layout that snakes through a timeline of computer history. This is a great idea for the general public, because it gives them some perspective as computers evolved from mainframes to minis to micros to handhelds. The downside is that the narrow pathways were a bit crowded thanks to a surge of convention-goers. Still, I think the new layout is the right move for the museum’s day to day operations.

You might find yourself a bit hungry after exploring these exhibits, and VCF brought in the Boardwalk Bites food truck to serve lunch, snacks, and drinks. While I thought OCEAN’s food quality was decent last year, Boardwalk Bites’ menu is considerably longer with new options like cheesesteaks, veggie sandwiches, pulled pork sandwiches, and onion rings. On Friday I ordered a cheesesteak, onion rings, and a lemonade which rang in at $21—thriftier attendees might stick to burgers and chips. I thought the quality of the food was good, and it saved me from the possibility of losing my parking spot if I left the campus to get lunch. Based on the lines I saw at lunchtime every day I think they did a pretty brisk business. My bet is that Boardwalk Bites will return next year.

After recharging your physical battery—and maybe your phone battery too—you’ll be ready to attend one of the many panels and talks. There’s twenty four sessions across three days with the majority centered on the show’s theme of vintage computer sound, with topics focusing on sound chips, sound programming, and computer music. Friend of the show Ian Scott chronicled the development challenges he faced when designing his multitalented PicoGUS retro sound card. Brendan Becker had two panels covering the different facets of chiptune music where he analyzed our favorite old synthesizers from technical and aesthetic perspectives. Rebecca Heineman took us on a tour of computer sound based on her extensive history in programming. But my favorite of the sound panels was the history of Ensoniq. Commodore historian Dave McMurtrie brought Al Charpentier, Bill Mauchly, and Joe Friel together to tell the story of Ensoniq from their breakup with Commodore to their sale to Creative Labs.

Other talks about interesting non-sound subjects were sprinkled into the schedule. One favorite of mine was Bart van der Akker’s panel detailing the recovery and restoration of the Aesthedes computer graphics workstation. Years before Photoshop and Illustrator these mammoth workstations brought digital drawing and design to people used to paste-up and airbrushing. Dean Notamicola and Justin Hopkins had back-to-back slots reserved for talking about how we curate and manage our collections of old tech. And I appreciate that VCF lets extremely specific subject matter like Alastair Hewitt’s bare metal booting or Frank O’Brien’s SAGE panel.

But this year’s headline events were two roundtable discussions celebrating the Commodore Amiga’s fortieth anniversary. The Saturday session was packed full of Amiga fans who wanted to steal a glimpse of the original breadboard prototypes and other engineering samples. Dave McMurtrie was joined by The Retro Hour host Dan Wood to chat with over a dozen Amiga Alumni in irreverent free-for-alls where they reminisced about the good and bad times forging one of computing’s crown jewels.

These panels and talks took place in the usual panel room which hasn’t changed much these past few years. New decorative LED display panels and ambient lighting brought some color to the space this year. VCF’s AV team always does good work here, as their projector, recording, and PA setup has been battle tested over years of shows. There’s only one weak spot, and it’s the handheld microphones given to guests and presenters. Holding a mic up to your mouth for an hour can get a bit tiring, and I noticed in a few panels that some speakers had trouble staying on mic. Audience members shouted “talk into the mic!” on multiple occasions because they couldn’t hear someone speak. Maybe some stands or lavaliers might be in order to help speakers stay on mic.

Also noticeably absent from this year’s programing lineup were any panels or group discussions with community content creators. It wasn’t for a lack of a host or guests—I certainly saw many qualified panelists at exhibits or just walking around the show. I’m not quite sure who organized those panels in previous years—did the show invite a host and guests, or did a host gather some guests and then propose the panel to the show? If it was the former, I could definitely see the Amiga roundtables taking up all that effort. If people though the show put it together they wouldn’t have suggested it themselves and by the time the schedule was published it was too late. That said, if the show had to choose between that and the Amiga’s fortieth anniversary, I think they made the right choice. I’m sure we’ll see the return of the content creator panels in the future.

Consignment

The business of vintage computing is booming, and there’s no better example than VCF East’s bustling consignment marketplace. Over the years the pile of things for sale and the crowd of people wanting to buy stuff has always outpaced the available space in the consignment hall. Last year’s move to the Monmouth County Fire Museum was a good idea on paper, but an unexpected floor collapse held it back from its full potential. For 2025 the VCF staff found a new home for consignment at Camp Evans’ former radar laboratory in building 9036. And after years of consignment struggling with tight quarters these new digs might finally solve the problem once and for all.

One peek inside the laboratory hall and you’ll forget all about those old cramped consignment halls. There’s more than twice the floor space of the fire museum, and it dwarfs the old cafeteria. Thanks to this abundance of square footage there’s plenty of room for all your stuff. I was able to keep all my items for sale together on one table for the first time—nice! Crowd management also benefitted from the extra space. Rope poles defined clear entrance and exit lanes to direct people in and out of the hall. Wider aisles between tables and shelves created plenty of breathing room between you and other shoppers. The only floor plan improvement I can think of is a dedicated queue for people to line up to check out. And buyers could test their potential purchases using these strategically placed power bars. Just plug and play to try before you buy. Excellent!

How about price and selection? Browsing consignment is like flipping through eBay or Craigslist in real life. I see a lot of people trying to downsize their collections, or rehome a few things they’ve rescued from estate sales and thrift shops. I can also sense when a seller “knows what they have” and lists something for an… aspirational price. Hey, if you think someone will spend the money, go for it—just don’t be too sad if you have to pick up unsold items at the end of the weekend. But I’d say hunting for whales is the exception and not the norm—most people selling their items in consignment know the general value of their gear and are pricing it accordingly. Toss in the show’s 18% commission—which hasn’t increased this year, thankfully—and you end up with a lot of prices in line with the market. Maybe that’s an indictment on what nostalgia-crazed buyers are willing to pay for this stuff, but there’s still plenty of bargains to be found.

But if a bargain is still too rich for your blood, you don’t need to spend money to take something home. Set up along the exit side wall was the free pile, overflowing with what could possibly be called trash or treasure depending on your point of view. It’s a smart location that was impossible to miss as people walked out. Boxes of books, old company swag, projects that people would rather forget—it’s all here. There’s also plenty of things that can be useful to somebody but are too much of a pain to sell, like still-working but not cutting edge networking gear. I can usually see a home for most of this stuff, but I have my doubts when it comes to things like these flat panel TVs that sat around all weekend. C’mon. They’re better left to a thrift store.

How about the checkout experience? Buyers brought their finds up to the registers where volunteers scanned barcode labels and settled transactions with cash or cards. This is built upon a NexoPOS system where sellers entered their inventory into a web app and printed out barcodes at home or at the show. I thought this setup worked pretty well last year, and I was curious what improvements would come this year.

As a returning seller I was able to use the my credentials from last year, and upon login I was greeted with last year’s inventory and sales reports. I tried deleting them to clean things up, but I was just left with a bunch of dependency errors. I reached out to the VCF staff, who said all items from last year had their quantities zero’d out and that new entries should be made for this year’s items. Fair enough, and I did just that. Fast forward to the show itself, and as I kept tabs on my sales during the weekend I noticed this year’s sales were mixed in with last year’s. I was able to track my sales with a little math, but I would’ve preferred to see only 2025’s sales in the report.

I asked the VCF consignment team why they didn’t wipe the database and start fresh, and their response was that they want to track historical sales data so they can see price trends and give guidance to sellers on how to price their items. It sounds like a great idea in theory, but as it stands I’m not sure how practical it’ll be. Am I really gonna see stuff I sold a few years ago clogging up my listings in the future? And how will they account for all the variances in listings and descriptions? I know I wasn’t the only one confused by the old items—some people reused listings from last year which didn’t play well with their sales estimates. Thankfully the system tracked when items were sold so it properly calculated everyone’s payout totals. If it were me I’d export out past data or flag it as historical so it doesn’t interfere with entering or tracking the current year’s items.

I also had a minor issue where items were shown as sold during one part of the weekend but disappeared off the list at another. When I alerted the register staff to the issue, they investigated it and found out a problem with how they were marked on the back end which they quickly resolved. So I give them a lot of credit in fixing the problem, and it’s in line with the good service I’ve received from the VCF volunteers over the years. And when it came time to settle up at the end of the show the line moved quickly as they handed out cash or PayPal receipts.

Another Year, Another Show

Vintage computing is such a weirdly wonderful thing. We willingly throw ourselves into the pursuit of the obsolete for reasons that might make sense only to ourselves. And that’s the spirit that draws us to these sometimes wild but always interesting events. When I talk to VCF East first-timers they usually respond with “It wasn’t what I thought it would be!” And that’s not a sign of disappointment. VCF East is more symposium than convention, and that goes a long way to make you feel like you’re part of history.

Naturally this places a limit on how much and how fast VCF East can grow. In the bubble era of retro gaming and computing there’s certainly a rush of both newcomers and old hats looking to see old hardware and software in person, and this is a unique, welcoming environment for them. VCF clearly values the community and its input, and this is evident in the post-show survey. You don’t have to be a writer like me to get your point across—just post your comments in the survey and they’ll be seen by the folks in charge. They take constructive feedback seriously.

I’ve mentioned the limits imposed on VCF East’s venue in past reports, and those aren’t really going to change. What makes it such a fun and pleasant experience does have some downsides. If it’s cold, rainy, and dreary like early April tends to be you’ll need to dress appropriately. You’ll get a workout with all the walking you’re doing between the various exhibit halls and other buildings. If you buy something heavy from consignment you’ll need to haul it all the way back to the parking lot unless you brought a wagon or managed to flag down a shared golf cart. But remember that giant hotel conference centers also have their drawbacks. I prefer VCF East working within the canvas of InfoAge so it can continue to grow while keeping its spirit intact. And they did exactly that with the upgrade to consignment, which was the area that needed improvement the most. Maybe they’ll have to move some day, but I don’t see it happening any time soon. So if your heart hungers for the beeps and boops of old computers, schedule a trip next spring to the sandy shores of New Jersey. We’ll see you there.

The VCF Midwest 2024 Report

January 07, 2025 by Dan Vincent in podcast

Hey, Chicago, whaddya say—are you ready for some old computers today? Here, in Userlandia, it’s time for a new and improved VCF Midwest.

Oh Vintage Computer Festival Midwest, how you’ve grown! After nineteen years of hard work you’re today’s premiere destination for connecting people with yesterday’s tech. But after last year’s blockbuster event a clear consensus emerged across multiple postmortem reports: the show had grown beyond the capacity of Elmhurst’s Waterford Banquet and Conference Center.

Of course, pointing out a problem is easier than solving it. But Chicago Classic Computing heard the feedback loud and clear, and on March 8th 2024 they announced VCF Midwest’s move to a new home: The Renaissance Hotel and Convention Center in Schaumburg, Illinois. To call this an upgrade is an understatement; it’s like replacing a 486 PC with a Pentium III. Now VCF Midwest is one of the many mid-sized conventions,

trade shows, and expositions that live in the Schaumburg Convention Center and its 97,000 square foot main convention hall. The Waterford's main ballroom was barely 12,000, and even when you add in 8,000 square feet of hallways and lobby space, that's still barely one fifth the size. And if that’s not enough there’s 50,000 more square feet available in the Schaumburg ballroom and other conference rooms.

Those are some big numbers, but would the new venue actually work? En route to Chicago, I pondered the consequences of this change—because there are always consequences. Sure, a larger venue can solve logistical problems like floor space, overcrowding, or parking. But what could go wrong? Did staff bite off more than they could chew? Would it change the character of the show? The only way to find out is to dive on in to the wonderful world of the new VCF Midwest.

Exhibition Excursion

Whether they were exhibiting, attending, or observing from afar, most everyone I talked to had one question: how would the new venue change the show? And when certain viewers with millions of subscribers say “Hey, I’m really looking forward to your video on this year’s show!” well… no pressure! But I love a challenge. So, here: what you've all been waiting for: exhaustively detailed coverage of a wonderfully exhausting event.

The obvious impact of all that extra floor space was a bigger list of exhibitors and vendors. This year’s lineup featured 197 distinct booths—over 60% more than last year. And all the available square footage meant many exhibitors could use multiple tables to set up some pretty impressive displays… that is, if they could afford them. VCF tables used to be free. Now they're fifty bucks each, with extra fees for four or more. I actually think this is good: VCF tables have costs in terms of setup, power delivery, and so on. Having to pay for it means people are less likely to no-show. It also keeps people from requesting more tables than they actually need.

But any attendee questions about all this newfound space were answered after stepping into the main exhibit hall. A seemingly endless ocean of tables covered in computers of all kinds created a bountiful bazaar of all things retro. At first glance it might seem like tables are assigned randomly, but after you walk the aisles and check the map, some patterns emerge. There were no hard and fast rules that, say, all platforms must be bunched together, but certain zones seemed to be a little more Commodore, Apple, or Atari. Celebrity YouTubers were in the southeast corner, and large vendors were mostly along the north wall. Last year, the big iron exhibits were all in one place. But arranging a convention hall gets tougher the bigger it is. Arranging the exhibitors to accommodate lines and minimize traffic jams is like a real-world game of Tetris… except the tables don't disappear. There’s practical needs like electricity and political ones like people who want their tables placed next to each other. The result was a bunch of C- and I-shaped islands. Crowds could gather at tables without obstructing foot traffic.

With a bigger venue comes more room for variety—and VCF had plenty of that to begin with. A CRT light show of epic proportions by Aron Hoekstra is powered by a Cromemco Dazzler, the first color graphics board for S-100 computers. If you found a Dazzler of your own and wanted to build a system around it, Jeffrey Wilson is here to help with his S100 Projects featuring an ATX S100 kit. A flight of retro PCs and Macs are brought together at Kokoscript’s system sampler, with this DoCoMo Post Pet giving me real D-Terminal vibes. Multiple first-of-their kind laptops from Kaypro through IBM are gathered at Ben Gennaria’s, with the coolest one being this working Atari Stacy. It’s rare to see one in such great shape. Did you know that Epson once made entire computers just as accessories for printers? It’s true! Steve Hatle had a QX-10 on display, and Brian Johnson had a QX-16 and even a portable PX-8.

Extreme specificity like that is catnip to me, and many exhibitors were more than happy to put extremely specific themes on display. MaidenAriana from RetroAlcove made a beautiful flowery pink table for her tribute to Wing Commander, featuring a playable period setup of the Sierra classic. CLIMagic boasts that anyone can become a UNIX wizard with a little guidance from a master mage. Quinn’s display of vintage logic analyzers lets you probe and prod with the tools of yesterday to see how hardware problems were diagnosed way back when. My award for the most specific niche goes to Isaac Z. Raske, whose Computing in the Dark display detailed a history of assistive interface technology. Braille keyboards, speaking assistants, and touch devices show how those without sight can interact with computers.

Kyle Gagnon’s collection of digital cameras spoke to me as an enthusiast of camera gear. Covering a decade of innovation in consumer digital imaging, these cameras represent a great period of experimentation. Of particular interest are the ones with tethered and pivoting lenses—a forgotten fancy for photographers freed from film. This exhibit paired well with its neighbor the MDCon Road Show and its traveling display of Sony MiniDisc players, media, and ephemera. You've probably heard of audio MDs, but the Roadshow has Data MDs and even Handycam Video MDs.

If PCI and ISA are too internal for you, Kevin Moonlight’s PCMCIA exhibit featured unusual PC Card expansions to let portable computers do things that were impractical for desktops. OS/2 And You presented by Joshua Conboy returns with more boxed OS/2 software and a new slate of PCs running IBM’s ill-fated operating system, including this IBM PC Server 310. It’s one of IBM’s rare transitionary platforms where you could option both PCI and Micro Channel.

Then there was section P, which was full of Atari goodness. The Suburban Chicago Atarians are back, taking advantage of more space. Next door was John Buell, with even more Atari gear. Though FujiNet is available on other platforms, they never forget their Atari roots. Scott K’s networked Atari Jaguars let you try multiplayer on a system you might’ve missed. Mixing it up is Slor’s Atari and Friends, showing that yes, Atari and Sinclair can be friends. If you’re interested in possible upgrades for your old Atari PCs, Stephen Anderson had several modified machines on hand. Other Atari fans could be seen across the hall, like the Atari Guy and his table full of Atari computers.

Although minicomputers and workstations weren’t as closely grouped together, there was plenty of old enterprise gear, like Hooloovoo’s DEC-powered plotting. You don’t often see HP minicomputers out in the wild, but Mike Loewen’s HP 2108 was punching paper tape souvenirs for everyone to take home. AT&T systems are another rarity, and John Orwin’s collection of AT&T boxes includes Unix PCs and a 6300WGS. There were fewer SGI and Sun setups than last year, but the two most popular Unix workstation makers still had some rep. Anthony Bolan returned with a colorful display of SGI workstations, emphasizing multi-seat applications running on one system. And if that’s too graphical for your tastes, Beehive Bit Bunker has actual Cold War-era nuclear control. This Nuclear Data Inc. spectral analysis system measured gamma radiation and other radioactive data. Local connection: NDI HQ was in Schaumburg! Or play some Zork on an AS/400 that's also running some businessware on the side thanks to Michael Mason and David Kudler. Just don’t tell their boss.

As always, Commodore commanded significant presence on the show floor—literally. Witness the mega-sized SX-64 Ultimax. Remember Tron? Jessica Petersen has a four-port joystick adapter for her Tron-esque Deluxe Light Cycle Game, and if I'd had more time, and three friends with me, I might’ve taken a turn myself. The Style demo group had a massive four-table end-cap with many Commodore machines running hardware-breaking demos and unique games. Steamed Hams, at this time of year, at this time of day, in this part of the country, running entirely on a VIC-20? Yes, and not only can you see it, you can try it! Cassettes fell off quickly here in the US as a primary storage medium but tape still had its day overseas, and Commodoreman’s collection of datasettes let you press play on all different kinds of tapes. A collection of history at Mike Shartiag’s table consisted of some interesting PETs and a Commodore cash register. Even more PETs were at Ethan Dicks’ corner running an enhanced version of the FLOPTRAN BASIC compiler. Hot-rod Amigas like Eric Wolfe’s PiStorm 1200 running AmiKit push the system’s limits from a hardware and software standpoint. And for a mixture of all kinds of Amiga ephemera, Mr Great & Booger Reborn brought demos.

For some people VCF is the only opportunity to touch tech that wasn’t popular on American shores. A ZX Spectrum Next and other systems were running the Lantern text adventure engine at Evan Wright’s table. If you don’t know which side you would’ve picked in the British Micro Wars, Chris Roth can help you decide. You can try a BBC Micro, ZX Spectrum, or watch a 1982 episode of The Computer Programme to educate yourself about computing across the pond. For a different kind of British computer, check out John Ball’s excellent timeline of Apricot PCs. His array of Apricot’s interesting IBM in-compatible MS-DOS PCs—you heard that right—includes an Apricot Portable and the stylish compact F1. Plus, there’s a rotating lighted sign!

Japanese computing has a fascinating history, and Noah Burney’s MSX Around the World shows the universe of MSX computers in Asia and Europe. With eight models to try you’ll see just how much variation a standard could accommodate Across the way was NEC Retro, returning to the show with a new slate of Japanese domestic PCs and desktops with integrated Famicoms.

Sometimes a table is just a collection of computers doing interesting things. Where else can you play a game of Rogue or check out a Brookhaven Instruments digital correlator but at Evan Gildow’s booth? When there’s too many computers but not enough table space, why not copy Dillon Tracy and build some shelves to let you use five different computers at once. And don’t forget Andy Geppert’s Interactive Core Memory, with its new 6502-powered neon pixel core simulator.

If you love Texas Instruments and Tandy, VCF has plenty for you too. Jim Mazurak’s maxed-out TI-99 combines its many expansions to transform this micro computer into a mega computer. Trash-80s are treasure at Neil’s Computer Service. I went cuckoo for CoCos at Ken Waters’ table of all things Tandy. Get up close to the processor powering these Tandys and more at Chris and Gavin Tersteeg’s spotlight on the Z80. And Ken Van Mersbergen’s table had a little bit of everything, with Tandy PCs and software and a spectacular tribute to the Coleco Adam. Terminals, robotics, and other obscurities team up at the fearsome foursome of Forgotten Machines, Nevets01, Josh Bensadon, and Dave Runkle. When you’re done learning to program the COSMAC Elf, you can explore the RCA CPU that makes it tick. Flip some switches to control an Altair powered robot, then try to puzzle out the Convergent Technologies AWS. I don't know if this grouping was intentional or a coincidence, but it was excellent.

Several museums sent exhibits. Hailing from Columbus, Ohio, the Sprawl Technology Library preserves old media, which includes games like Balance of Power. System Source was back too, this time with their IBM 1130 from VCF East and their own tag sale of used equipment. Midwest Classic Video Game Museum returns with a feature presentation about the APF Imagination Machine and their AndroMan robot. And the Museum of Batch and Time-Share Computer History brought perhaps the biggest beast of the show: an IBM System/36 with a quad-pack hard drive. The transparent disk covers revealed the delicate dance of heads across platters big enough to carry an extra large tavern pizza.

Other conventions had tables to promote their own events. VCF East says that if you had a great time at midwest, try New Jersey for a springtime shindig. West Coast more your style? VCF Godfather Sellam Ismail represented California’s VCF scene with pieces of West Coast Computer culture. Midwest Gaming Classic says there’s more retro fun up in Milwaukee at their massive celebration in April. And members of the Wisconsin Computer Club brought some cave aged pieces ripe for the playing.

One thing I’d like to see more of is people doing archiving for old media. I know that’s tough to do live, but I appreciate folks like Chris Simmons giving it a go with his ESDI hard disk rescue setup. I know there were folks doing the same with VHS and LaserDisc setups—maybe people could work together at future events to explore more archiving opportunities.

We’re in a boom time for modern retro computers, whether they’re FPGA recreations of old machines or new platforms inspired by the classics. Mega65 systems were back at Dan Sanderson’s table, with the batch 3 revisions currently shipping worldwide. Tex-Elec and the 8-Bit Guy had demos of the Commander X16, while Foenix Retro Systems paraded the latest updates to their F256K platform. Next to the Foenix booth were two developers for the F256: Matt Massie discussing C game programming and Micah Bly showing off a port of NitrOS-9. Community support for these platforms is growing every day, and demoing at conventions is crucial to raising awareness by giving curious folks a taste before committing the hundreds of dollars it takes to buy one.

I appreciate when people create novel exhibits of commodity hardware that runs MS-DOS or CP/M. It keeps things from feeling too same-y. Sierra Back-OnLINE has a library of classic Sierra games ready for you to play on different PC clones. These machines weren’t special back in the day, except that Compaq LCD portable, but attrition has turned them into collectibles. We joke about the Osborne effect, but Adam Osborne’s computer company got some love from several tables, like this one called Osborne Computer Group! Other luggables joined in at Nicholas Mailloux’s table, where I basked in the amber glory of an IBM 5155 Portable PC. Steve Maves’ timeline of laptop evolution chronicled the progress of shrinking computers. Interested in the early attempts to bring graphical interfaces to the PC? Regret_the_van’s Rise of the GUIs shows what might have been with a buffet of visual interfaces for sampling different interpretations for windows, icons, and pointers.

Apple fans would love this year’s crop of classic Macintosh, Apple II, and Lisa exhibits. Scott Barret and the Pittsburgh Classic Mac Lab recreated an early ‘90s classroom computer lab full of KidPix, Print Shop, and Oregon Trail. DanaDoesStuff’s monstrous Apple Network Server was live all weekend, flanked by old towers and PowerBooks. Video conferencing and collaborative drawing were excellent multi-user activities powered by Peter R’s network of classic Macs. Ryan Burke’s 40 Years of Macintosh returns with a bigger lineup of classic Macs, but the star is this mountain of a Macintosh Plus with an astounding quadruple SCSI hard drive chain. Who says you couldn’t buy a tower Mac back then? Exhibiting for the first time at Midwest is Kate the Macintosh Librarian, and her mascot Maccy made friends with everybody stopping by. Mac Enthusiast friends of the show were around, starting with Action Retro and his lineup of greatest hits. Play a round of Creepy Castle on Ron’s Computer Vids’ breakout Macintosh SE powered by his boards he had for sale. Experience the latest in HyperCard stacks with Eric’s Edge and his home-brew Adventure stack. And Steve of Mac84 brought a very special guest: his color LCD SE/30. Luckily his mousepads arrived just in time to be put on sale. SIT happens!

Many popular exhibits were multimedia marvels that demonstrated how computers were behind some of our favorite memories. Frank Palazzo and Evan Allen had restored a Cyberamic control system, Those ran the animatronic stage shows at Chuck-E-Cheese. These tapes and cards turned a cigar-chomping rat into a rock-n-roll star. Beefy Betamax cameras were the bread and butter of the broadcast industry, and you could try them yourself at Kyle’s Digital Lab. Sorry about catching you when your batteries were low, Kyle—hope you felt better! Genericable added new service to their homegrown television network, like the ability to call in an emergency weather alert. An interactive Delorean time stack at Ted N’s table lets your inner Marty McFly go back in time.

Vintage synthesizers and computer audio are big hits at VCF, and the opening act is The OPL Archive. A display of PCs, sound cards, and MIDI boxes teamed up with classic synths to form an FM synthesizer supergroup. Next door was Skye Janis with more FM history, including a Mac SE driving Yamaha synths to produce some incredible music. Avery Grade’s IIGS stack is ready to jam with some sampled digital audio, and their next door neighbor Bea Thurman at the Sampling Apple rocked a Greengate DS:3 Apple II synthesizer. But something that was new to me was this Blaster PC at NightWolfX3’s table. It’s a TigerDirect barebones PC building kit with a motherboard that integrated a SoundBlaster Live. Tiger didn’t sell these for very long, which make them a rare bounty for Creative collectors. The folks displaying it lamented that they were missing drivers for things like the remote control, but they were in luck because a random passerby mentioned that he had a driver disk! When he got home, he uploaded the contents to archive.org. Now the remote is fully operational. Only at VCF.

These computer conventions are great for networking, and I’m not just talking about the human kind. Computer networks, phone networks, and radio networks are tons of fun in person. Leading the charge was the VCF Midwest BBS, returning this year with more terminal types to show how different platforms display the same board. ProtoWeb and Darren Young’s Nabu Retroweb let those who weren’t around for the ancient web or TV computers get a taste of old-timey networks. Also returning are the Atari BBS Gurus, whose Atari-centric presentation lets you peek into the life of a terminally online ST user. Ronald Coon Junior’s Never Land BBS is still running after 35 years, and you could post live from the show floor. Do you miss QuantumLink? Check out the Commodore precursor to America Online at QLReloaded. Six modems stuffed in an Apple IIe powered Steve K’s recreation of DiversiDial, a dial-in chat service that worked like CompuServe chat for a fraction of the cost. If you’re interested in networking that’s a little more local, Old World Computing had a classic AirPort setup featuring all the colors of the iBook rainbow. And John Mark Mobley’s TI Silent 700 terminal challenges you to level up your Linux skills by connecting to a remote bash shell using a paper teletype.

ShadyTel Midwest’s central office serves a key role at VCF Midwest by providing phone and network services to exhibitors. But this year brings a new connection: cell phones! A collection of vintage Motorola cell phones lets you relive the glory days of yuppie status symbols. Other telephony tricks include Thomas Major’s Time-Division Multiplexing System, a small scale implementation of a long-distance serial link. And if you don’t need wires where you’re going, ByteShift can show you how to transmit your data over long distances using Ham radio packet communication.

More tables this year featured people promoting their YouTube channels, blogs, and podcasts. Maybe some day I’ll be among them! There’s something fun about finding recommendations in person, especially when they’re backed by interesting exhibits. CityXen put up a demo of the C64 game Whackadoodle, with some controllers from RetroGameBoyz that were utterly unique. This Heroes of Might and Magic Necropolis model isn’t just for show—the Canadian Computer Collector built an entire PC inside of it. I gave five dollars to SavvySage for a copy of QuarkImmedia—now that’s a purchase I’ll never regret! If you wanted a live podcast experience the hosts of the ANTICS and Floppy Days podcasts were available for some back and forth. And June’s Nybbles & Bytes promoted the Commodore 128's multi-display power with a psychedelic game of Drunken Snake.

Now after this whirlwind tour you might be thinking “wow, that’s an amazing amount of stuff,” but that was only two thirds of the tables! I haven’t even gotten to the vendors and VIP placements yet. But having all of this stuff doesn’t mean squat if you can’t see or reach any of it. And thankfully the greatly expanded expo center meant lots of space for crowds to flow. Even the VIP tables had lots of room for folks wanting a selfie or face time with a high profile YouTuber without interfering with general traffic. I’m not sure how things would’ve worked if LGR was there—he had the longest lines in the past—but there was enough room to plan around such things if he was.

One area that could use improvement for both exhibitors and vendors is the differentiation between tables. Some exhibitors seemed to blend together and it could be hard to tell where one exhibit ends and the other begins, especially when similar themes are grouped together. The best way for an exhibitor to differentiate their table is a custom tablecloth and a stand-up banner—wise investments if you’re regularly tabling at shows like these. First-timers might try the thrifty option of tabletop sign holders from an office supply store. But the event could help too. Alternating tablecloth colors or adding small gaps between each exhibit wouldn’t hurt. Those tiny green table tents with names and numbers were hard to see or sometimes missing entirely. Many shows hang simple foam core signs with the booth’s name and number on the front of the table. That’d definitely help see who’s who in the ocean of tables.

And if you didn’t see your exhibit in this video, please accept my deepest apologies. This show was so massive that I spent most of Saturday filming as many booths as I could, and I know I missed a few. Some I overlooked because I misidentified two or three adjacent tables as a single exhibit. And then there were the ones I missed because I thought “Oh, I’ll get them tomorrow” and it turned out that they sold out of all their product or they bailed early on Sunday. And most shameful of all I accidentally deleted a subfolder of clips of some VIP tables. What’s worse is that I did it while sorting and backing up other footage! I wanted to highlight that really cool fansubbing setup by Retrobits, Adrian’s C64s, and B-roll of the other guest tables, and now I can’t. I’ve learnsed a lesson about immediately duplicating footage SSDs that I won’t soons forgets. Can you ever forgive me?

Venue Variables

Outgrowing the Waterford’s ballroom wasn’t the only reason VCF needed a new venue. Highway access, parking, accommodations, and food service matter to attendees, and what worked for a small show—or smaller, anyway; this isn't the first time they've needed to find a new home—no longer satisfied a growing VCF. Fortunately, the Renaissance is a more suitable location. It’s located twenty minutes away from O’Hare airport at the crossroads of I-90 and 290. And you'll definitely have an easier time parking, too—the Ren has more than four times as many parking spaces than the Waterford. And that’s with the Ren hosting a very swanky wedding on the same weekend—though I did notice some wedding guests hunting for spots around lunchtime. Sorry, folks; hope the honeymoon went great. You might’ve had a long trek from the farther-flung lots, but it beats last year’s overflow arrangements.

The Ren is also a massive improvement in terms of lodging. I didn't stay there myself—I stayed at my buddy Mark’s place—but I have no qualms booking a room there because I've stayed in other Rens in the past, and it's a consistent brand. The general ambiance outside the panel room and expo hall is modern four-star business hotel. This is in contrast to the Waterford and Clarion Inn, whose decor… well… It's a little too on the nose for an exposition of computer hardware from decades past. Now, the Waterford’s retro style was part of the charm of VCFs past, but it's hard to argue with a better class of service. I know several people who were jumping with joy when they saw elevators to the guest rooms—or who would’ve been if they hadn’t been carrying so much stuff, anyway. Exhibitors wheeling stuff through cramped hallways is now a thing of the past thanks to loading docks! And if you couldn’t secure a room at the Ren there were official overflow hotels and other lodging in the area.

A better class of hotel also brings a better class of food service. The Waterford’s basic cafe and rolling snack cart was inadequate for feeding hundreds of hungry hungry hackers. The Ren does have a cafe for snacks and sandwiches, but for something more substantial the Schaumburg Public House offers actual dinner at hotel restaurant prices. When you want to kick back with some friends for a post-show drink there’s a full service bar and lounge. And if none of that tickles your tastebuds or fits in your budget there’s plenty of restaurants and fast food joints nearby. You don’t even need to drive—the Village of Schaumburg shuttle will bring you to the Woodfield Mall and other local destinations. And for those who can’t wander too far from their tables, there's a concession stand in the expo hall - though I wouldn’t exactly call it gourmet. It’s your standard convention center fare—burgers, hotdogs, chicken tenders, and personal pizzas—with convention center prices to match.

Another subtle but smart idea was a room set aside for chill space. Nestled between the expo hall and the main panel room it was perfectly positioned for a break from the show floor. Pull up a seat to recharge your batteries—literal or figurative—and keep the chill vibes flowing. Given the size of the crowd it’s good to have some getaway space. Hard and fast numbers aren't available—VCF is free, so tracking attendance is difficult—but, but I think it’s safe to say that the show’s year-over-year growth continued in 2024. And thanks to all the extra space the presumably larger crowd wasn’t as stressful as last year. It was refreshing to walk around the hallways and lobbies of the Ren without needing to dodge and weave around endless waves of bodies. Inside the expo hall the wide aisles and buffer spaces meant you could cruise from one table to another without bumping into people trying to buy something or playing a demo. There’s finally enough room for the crowds of today and tomorrow.

Snags and glitches cropped up, of course. A group of tables lost electricity on Saturday morning, which was an issue for people like Genericable where every minute without power is a minute that visitors couldn't interact with old WeatherStars. I think they were down for an hour or two before things were fully working again. The only crowd control problem I observed was the initial rush for T-shirts and merch at the VCF show tables. Just like last year the show tables were located by the main entrance, and the line snaked right through the doors and around the main hallway. If you didn’t know better, you’d think this was the line to get in, when in reality you could just walk by it if you didn’t want some merch—just like last year. It did clear up after a few hours, and having the merch booth by the main entrance is a boost to visibility, but there ought to be a proper queue set up.

And I might be a minority on this but I’d still like some form of printed information guide. There was a giant sign by the main entrance with a map and table directory, and I’m sure people were expected to take pictures of it with their phones. But sometimes phones are busy doing things! I’m not asking for 27 8x10 color glossy photographs with circles and arrows and a paragraph on the back of each one. A simple black-and-white folded sheet with a map on the inside and a legend on the back would be super helpful. If there was one and I missed it, I offer my sincerest apologies to the show. And yes, I know the website’s a thing, and while that faux-QuantumLink design looks cool it’s not very phone friendly. Combine that with occasional signal issues in a busy hall and you can understand why a low-tech option might be useful.

Lastly, the hallway between the panel room and the expo hall felt a bit bare at times. Aside from the classified whiteboards and the VCF banner at one door you’d be hard pressed to tell what kind of show was going on. I obviously prefer the new venue, but part of VCF Midwest’s charm is that it feels like a big party. When the show monopolized the Waterford that vibe was easy to see because people were everywhere and little diversions like the LAN were situated in hallways and lobbies. At the Ren the party was still going on; it just moved to lounges, lobbies, and restaurants. Things are just… a little more spread out. Depending on the show’s agreements with the venue they might not have the ability to add more set-dressing to the open areas. But if they could find room in the budget for bigger, bolder signage for the alternate entrances to the expo hall or the conference rooms that’d help liven things up. But really, that'd just be a nice extra. Overall the new venue is such a massive improvement that it’s hard to find nits to pick. Well done.

Vendors, Makers, and Traders

VCF Midwest is a good place for a temperature check of the retro and vintage computing economy. With the expansion in floor space there’s even more ways to trade your hard-earned cash for a touch of technological treasure. Need a specific system or part? The many professional resellers and amateur collectors unloading surplus or refurbished equipment might have what you need. Or you could hit up the numerous makers and creators selling and demonstrating new hardware and software solutions.

At the Waterford, the vendors were mostly positioned in the big hallway and lobby outside the ballroom. Note, ‘mostly’—there were always some mixed in with the exhibits. But at the Ren’s expo hall those divisions are gone. Now vendors of all sizes are mixed in amongst the regular exhibitors. The closest equivalent to the old hallway might’ve been the north wall where returning large-scale vendors set up shop. Jeff’s Vintage Electronics, whose parts table is a familiar sight to many VCF attendees, anchored the row. They were joined by Bonus Life Computers’ array of refurbished PCs and accessories. Ecotech Computer Solutions covered their tables with software that you could buy along with a PC to use it on. Rounding out the wall was Ecotronix E-Waste, selling recovered PCs and parts to new owners to keep them out of the scrap heap.

Long lines of tables along walls weren’t the only way to accommodate sellers with lots of wares. All that extra space allowed for numerous end cap and corner spots. Take Eric Moore—AKA The Happy Computer Guy—and his gargantuan IBM system shop. His five-table corner was covered with IBM keyboards, terminals, and the only DataMaster I’ve seen in person. I don’t know if anyone took that PC precursor home, but they should’ve! And the record for most CRT displays goes to the E-Waste Mates, where after a round of Smash on N64 you can bask in the glow of a five-by-five TV wall. Digital Thrift’s numerous shelves towered over their end cap full of Apple gear, big box games, and their own share of CRTs. Those hunting for parts for vintage builds would do well at Uncle Mike Retro’s acres of video cards, chips, and motherboards. The prices at Crazy Aron’s PC Parts are so low he’s practically giving this stuff away!

Vendors big and small can set themselves apart by combining their sales with some kind of exhibition. It’s nice to show that the machines you’re selling are functional, but a separate demo PC or a game console is a great way to attract customers. Quarex’s table of TI-99 goodness leads you right to buckets of big-box games, with lots of PC classics that were snatched up quickly. Same with the Windy City TI-99 Club, who piled their table high with items for Texas Instruments computers. Matt’s Cool Old Stuff employed a similar strategy—have a Power Mac G3 and other interactives on one table, and a pile of parts and cards on the other. When you were finished shopping at Joseph Turner’s massive booth you could enjoy a round of Mega Man on a Lightning McQueen TV. If you were hungry for a Sun pizza box, Paul Rak had a stack hot and ready for take out. Then again, a static display could be just as interesting—this massive old hard drive at Ben Armstrong’s table certainly attracted its share of admirers.

Embracing a theme is another way to distinguish your table. It’s all IBM all the time at Sam Mijal’s corner, with multiple PS/2s and ATs for the discerning Big Blue buyer. A bushel of compact Macs were ready for picking at 68K Rescue, with a stack of Mac II desktops also looking for forever homes. Logic boards, keyboards, RAM, and other bits were available to help refurbish your adopted Mac. MACNician offered refurbished PowerPC and Intel Macs along with edutainment software and teaching materials. But the biggest beige bounty was at Corbin Johnson’s corner table, where piles of Performas, LCs, and Quadras were available for reasonable prices. Most worked, but if you wanted to save a few bucks and also give yourself a project for a rainy afternoon, you could pick one that almost worked. And why not buy a NuBus network card while you’re there? Amiga fans were coveting these A3000 and A4000 setups at Peter K Shireman’s booth. Boxed Amiga software, rare peripherals, and documentation were just some of the odds and ends he brought, along with this very beige Amiga jacket.

FreeGeek Chicago is back with their ongoing mission to connect people with affordable technology, and they brought some of their inventory of refurbished PCs and parts to sell. If you missed them, they recently opened a new retail store where they accept donations and volunteers to help restore old machines to get them into the hands of people who need them.

Dozens of makers and creators attended this year hoping to get products they built or designed into your hands. Want to play around with an Altair but don’t want to pay obscene prices for a real one? Adwater and Stir’s Altair-Duino project lets you build an affordable replica of the famous MITS microcomputer. Add a bit of color to your classic computer with a light-up keyboard from Long Island’s American RetroShop, who made their way to Midwest for the first time. The Retro Chip Test Pro was also available to help you troubleshoot RAM, ROM, and more.

Custom cards and adapters for Apple, Commodore, and Atari were available at 8bitdevices, like these Apple II video cards for connecting modern monitors. New from MacEffects this year is their wonderful iMac inner bezel replacement which saved many iMacs from cracked and shattered CRT bezels. It joins their regular lineup of clear and colorful cases for Macs and Apple IIs. Also on sale were upgrades and expansions to make your Mac as colorful on the inside as the outside. Next door was 8bittees selling a wardrobe of pixel shirts to upgrade your style along with useful items like Apple IIc USB power supply adapters. CayMac Vintage returns with TechStep clones, PDS boards, and enough ROM SIMMs to make every SE/30 at the show 32-bit clean. The ever-popular BlueSCSI hard drive emulators were flying off the shelves at Joe’s Computer Museum—I bought four! Have an old broken iMac that you want to upcycle into something useful? Juicy Crumb Systems’ DockLite can breathe new life into old all-in-ones by converting them to external monitors. Same goes for Sapient Technologies’ recreation Lisa boards, which can revive temperamental Apple Lisas.

Commodore users were well served by folks looking to sell modern solutions for old systems. RetroInnovations brought not one, but two SX-64s to round out their demo display of various Commodore products for sale. After buying a C64, C128, or Amiga be sure to visit Kevin Ottum and Jim Peters to pick up a NuBrick power supply and save yourself from the danger of those old potted bricks. You’ll want a disk drive too, and over at BitBinders these modern recreations of 1581 disk drives can finally give you the 3.5 inch drive you always wanted for your Commodore. There’s even dual drive mechanisms for those who need to copy a lot of floppies. And don’t forget to pick up a joystick at Tech Dungeon to complete your new system.

One of my favorite products from last year is back: Scott Swaze and the Wifi Retromodem. Hayes versions are still available, but he showed off a prototype of a version for US Robotics Sportsters! These modems are more plentiful than the Hayes, so it’s more affordable than ever to make old modems useful again. Friend of the show Ian Scott’s PicoGUS sound card practically sells itself with an excellent display of its audio capabilities. It keeps getting better all the time—the latest update enables simultaneous SoundBlaster and MPU-401 output!

I’m also a fan of tables that don’t fit the traditional mold of VCF vendor. Inverse Phase is back with more record albums and now they’re promoting the Bloop Museum. Also several artists entered the expo hall this year to test the market for selling retro-themed art. Posters, postcards, buttons, and stickers featuring paintings of classic computers and consoles were all over Melissa Baron’s booth. As a fellow professional con artist I approve of this setup with fully-stocked shelves, clear pricing, and lovely watercolors. For those looking for something more cartoony, Your Sinclair was selling manga-style characters based on classic computers and operating systems. And these retro lithoplates at Lige Hensley’s booth project a color image on these reliefs to create a cool three-dimensional image from your favorite games.

Throughout the show a seemingly inexhaustible mob of buyers swarmed tables to tell vendors to shut up and take their money. Bigger tables and more space at each booth meant sellers could bring more merch to sell. Even with more inventory some tables were picked clean halfway through the weekend. Most vendors managed to keep things contained, but I did notice some booths with merch spread out on the floor. Usually floor items were tucked under a table, but in rare cases I noticed boxes strewn in front of booths. This happened at corners of the C-shaped islands and therefore the boxes weren't blocking traffic in the aisles. But a few were fairly large, like Dan Arbaugh’s cornucopia of boxes from 25 years of collecting rare and interesting Apple documentation, software, and accessories. Dan, thank you for remembering to keep space clear so that people could visit adjacent exhibits. I hope all your materials found good homes. Like the awnings in the parking lot last year, one or two instances isn’t a big problem. But the show might need to tweak their surveys about the size of collections for sale to make sure people are placed optimally in the hall and that vendors request the right amount of tables.

From a reviewer’s standpoint there’s a lot to like about this vibrant variety of vendors. And when talking to the people behind the tables I heard few complaints about their experience at the show. Yes, it was exhausting and exhilarating, especially on a busy Saturday. Granted, there were sometimes issues with cell reception when processing cards or Venmo. The most common request was for stricter vetting of people in the hall during setup—just like last year there were non-exhibitors floating around before opening hours. I don’t think people are worried about shoplifters; they just don’t want excited early birds interrupting their setup time. Personally I’d like to see the more flea market-type sellers grouped together, but given the complexities of arranging this space I think placements were balanced overall.

Anchoring all this was VCF central command where the show accepted donations of both cash and merchandise to sell in the garage sale to replenish its coffers. Volunteers were busy all weekend sorting through contributions, selling shirts, and helping attendees. It was hard to keep tabs on items moving through the garage sale, but I’d say the array was on par with previous years. I think the only thing it needs is some bigger signs or banners, because at first glance it sort of blends in with the other vendors that were nearby. Still, if the shrinking pile of stuff behind the counter was any indicator, this was a very successful fundraiser.

After picking over the garage sale you could pick up official VCF merch, like these delightfully ‘70s-style T-shirts and tote bags. I’m told the design’s an homage to Illinois Bell Phone Center stores, which is obviously lost on a New Englander like me. But you don’t have to be a local to appreciate good typography. I also appreciate that it was offered in a rich shade of malaise-era chocolate brown. But by the time the lines died down and I got a chance to shop I had missed out on medium-sized browns. Black isn’t quite as fun, but it's better than no shirt at all.

And lastly, the renowned Free Pile settled into its new home at the northeast corner of the hall. With more tables than last year it was easier to donate and browse items without fighting the crowd. Getting a subjective feel of items up for grabs was tough, given that I didn't stay by the pile the whole time. As I see it, this year’s ratio of junk to treasure leaned slightly towards junk, but who am I to call something junk if it’s useful to somebody else? Stuff moved so quickly that I saw completely different things every time I walked by on Saturday, which is a good sign in my book. I contributed some Zip drives—one working, and one for parts. Sunday was slower, with some of the undesirable stuff hanging around longer before finally being picked up. It was a better experience all-around than last year, even if I didn’t personally take anything. Kudos both to the staff for the improvements and to the community for their spirit of generosity.

People, Panels, and Events

Though exhibits and sales tend to take up most of the oxygen around VCF events these days, the truth is that these are social events meant for us to share our love for the obscure and obsolete. The VCF website promised a fun-filled weekend of panels, roundtable discussions, LAN events, and DIY building. And with a new venue comes the potential for bigger and better versions of the events we all know and love.

The LAN parties—that’s right, plural—were a great example. Last year’s LAN was a cozy little diversion tucked away in the Clarion Inn connector. This year it was upgraded to a major attraction in the expo hall with twenty stations available for pick-up PC gaming. And if that wasn’t enough, a nearby mini-party powered by the LANCommander digital distribution system offered more multiplayer madness. I think there’s enough potential for future LANs to expand into one of the venue’s many conference rooms, should budget and logistics allow. For now having them on the expo hall floor makes it easy to draw players in from the crowd. And when they were done gaming they could hop over to the DIY build area anchored by Build-a-Blinkie. Its footprint has grown significantly to allow more people to try soldering for the first time. With over forty workstations and an expanded variety of ready-to-build kits anyone could learn the basics of building their own electronics in a safe and supervised environment.

New to the show this year was an arcade and pinball zone brought by AVS Home Arcade. One might say “Hey, these guys are vendors!" and I’m sure they made a few quadruple-digit business deals. But in my experience you’re not getting any kind of arcade or pinball machines at shows like these without a company or a local museum that’s willing to loan some inventory. These aren’t vintage machines; rather they’re modern recreations or multi-systems targeted towards the well-heeled enthusiast. But it’s still within the spirit of the show to let people get hands on with some retro arcade action. If I had money to blow I’d be very tempted to buy this recreation of one of the creepiest, kookiest tables of all time: The Addams Family.

Once you were satisfied with your soldering and finished with fragging, you could pop in on a panel at the Schaumburg west conference rooms. The setup for this year’s talks was similar to prior events: a single main conference hall hosting all the panels, with subject matter ranging from extremely specific presentations to free-form roundtable discussions. The closest thing to a keynote at Midwest is the annual YouTube personality roundtable. Hosted as always by the genial and gregarious Jim Leonard, this year’s iteration featured a star-studded lineup of retro computing personalities with nearly two and a half million subscribers [2,494,930] between them. A standing-room only crowd hung on their every word as Jim peppered the panel with questions about their favorite platforms and the state of the retro community.

A new event which I sadly missed was the first VCF Midwest concert. Sean from Action Retro, Taylor & Amy from their self-named show, and Veronica from Veronica Explains combined their musical talents to form The Stop Bits for a truly outrageous show. I don’t believe it was recorded, but when I asked people that saw it they said they had a great time. Musical acts aren’t unusual at conventions, and since it went well I’m betting we’ll see more music at VCF in the future.

Most of the panels were presentations hosted by passionate members of the community. Peter Shireman got a little meta with his presentation about creating digital slideshows with Amigas in the early ‘90s. Those looking to save retro treasure from e-waste scrappers should listen to Chris Skeeles of Boardsort.com gave hints on ingratiating yourself with local e-waste facilities. For the game historians, Ken van Mersbergen took a look at Nexa Corporation and the games it published before merging with Spectrum Holobyte. The annual Mac collecting panel hosted by Ron and Steve concentrated on the clone era, one of my favorite periods of Apple history.

Many panels focused on creating hardware and software. Veteran hardware designer Jeffrey Wilson shared forty years of lessons learned while designing circuit boards and FPGAs. The Commander X16 update by the 8-Bit Guy and TexElec showed off new audiovisual demos, games, and software for their modern retro computer. A deep dive by Daniel Baslow into the Area 5150 demo shows how skilled hackers squeezed every bit of power out of CGA video and the 8088 CPU. Old MS-DOS fractal generators get a second life thanks to Richard Thomson’s open source efforts. And a roundtable about designing new hardware and software for vintage machines brought some of the finest makers and creators together on stage for the first time. Useful products don’t just appear from thin air, and the group shared their setbacks and successes in a frank discussion about home-brew product development.

Creating this schedule couldn’t have been easy. Over three times as many panels were submitted as were selected, and some hard calls had to be made when choosing the candidates. And while the venue does have enough space for a second panel track, the show likely couldn’t afford it this year from a financial, technical, and volunteer standpoint. That could change in the future, but for this year I think the staff made reasonable trade-offs in terms of scheduling the events.

From my attendee’s perspective the panels seemed to run like clockwork. There were no hiccups like last year’s technical difficulties that obliterated footage from the YouTube roundtable. It was thanks to Veronica from Veronica Explains filming from the audience that there was a full version of 2023's roundtable! You can't always count on that kind of luck, so the show took extra steps to prevent it from happening again. The only real flaw was the panel room’s inability to accommodate the crowd at the biggest events. I’m sure the show wanted more space, but their hands were tied due to sharing the venue. The wedding booked the majority of the Schaumburg Ballroom’s floor space, leaving one quarter slice for VCF. Only 203 seats fit in the room, a fifteen percent increase over last year’s 175-ish. It was enough space for the majority of panels, but not for the auction or the YouTube roundtable. Hopefully there's an opportunity next year for a larger seating area, logistics and finances permitting.

Speaking of the auction, VCF Midwest wouldn’t be complete without its fabulous fundraiser. Jason Timmons returned to the stage in full suit and tie to resume his role as auctioneer in an epic battle of the bids. The pace of last year’s auction suffered from having all the items upstairs in the Waterford lobby while all the participants were downstairs in the panel room. Bringing the items down just wasn't feasible, so VCF set up a remote feed for the audience. No such hack was needed this year since shelves of items could be rolled in from the expo hall. Jason kept things going nice and smooth, calling out bids and moving items along. Early Apple II logic boards, Motorola core memory, interesting old PCs, and weirdly cool printers were all vying for your generous monetary donation. There was even what looked like an old payphone saved from the scrappers.

Though the auction itself went off without a hitch, I noticed that a list of auction items was missing from the website. In past years VCF staff updated the listing page during the show to give you an idea of what was up for bids. Either a technical or logistical problem prevented these updates, so unless you checked out the racks in the expo hall you’d have no idea what was in store. But thanks to an advance posting on VCF’s Facebook page I knew of one very interesting entry: an IBM ThinkPad 701. That’s right: a butterfly keyboard laptop. They don’t show up locally very often and even parts machines tend to go for hundreds of dollars on eBay. But I had room in my budget for one unreasonable purchase, and I knew that hot butterfly had to be it.

Said budget was in the range of $3- to $400, but I wasn’t confident that would be enough. A rare item like this could gavel for $500 or more. When it came up for bids it looked to be in decent physical shape and the keyboard mechanism unfolded perfectly. But there was no power supply and therefore nobody tested it to see if it turned on. Just before bidding an audience member asked a crucial question: what kind of screen did it have? The stage manager checked the bezel and announced model number 701C… s. That pesky s drew a chorus of groans from the crowd—unlike the 701C-no-s and its nice active matrix screen, the 701C-with-s has an inferior passive matrix screen! Oh well—something something, beggars choosers. Still, the keyboard was the star, and if it worked—and that’s a pretty big if—the winner would walk away with one of the most collectible ThinkPads.

Bidding started at $100 and entries were coming in hot and fast as the price climbed over $200. Bidders dropped one by one as the price reached $300 until it was down to me and one other person. At $360 it went once… twice… sold, to the guy with the beard and the t-shirt advertising his YouTube channel! And it was still within my budget, too. I picked it up from VCF central command and immediately played with my new toy. Keyboard goes out, keyboard goes in! Keyboard goes out, keyboard goes in! But enough about the trick keyboard—does it actually work?

After watching the last few panels Mark and I headed back to his place to test the Butterfly. After removing the corroded battery and plugging in a spare ThinkPad power supply we held our breaths and pressed the power switch. The laptop powered on! Yes, it powered on with a ‘missing operating system’ error, but that's not nothing! To quote one of the great luminaries of our time: it freakin’ works! After we restored the IBM factory hard drive image it booted right into IBM’s dual-OS selector for Windows and OS/2. We loved the introductory video of a besuited IBM flack welcoming us to the ThinkPad family and how a ThinkPad embodies IBM’s “spirit of excellence.” And that passive matrix screen? It’s not so bad after all. A magical folding keyboard makes up for a lot, and I have a hunch it’ll show up in a future video.

The Winds of Change

I have to tip my cap to the VCF Midwest staff. In last year’s review I speculated on ways they could work around the Waterford’s limitations. While I thought a new venue was necessary I didn’t think moving was in their plans. Well, they proved me wrong by making the tough call to spend the money and effort to relocate to a larger venue. And it worked out! Their new home filed off many of the rough edges and the result is a better, less chaotic VCF Midwest. That’s not a bad thing per se—there comes a point in every show’s life when they’re forced to decide if they’re a glorified meetup or a capital-C Convention. VCF Midwest isn’t a scrappy little get-together anymore. It’s a big show now, and it’s acting like it.

I’m sure some people are grumpy about the loss of intimacy, or how it’s impossible to get a table or panel. And they're not necessarily wrong, but those are gripes about any growing convention. Now the show’s ongoing concern will be sustaining itself from a financial and logistical standpoint. This is the case for all conventions, of course, but for VCF it's more important than ever, since the Ren is more expensive but attendance is still free. I understand the desire to keep admission free as in beer because the staff aren’t in this for profit. But if I was treasurer I’d be exploring additional revenue streams to help finance expansions like a second panel track. Creating sponsor packages to attract larger donations from individual attendees might motivate some people to give more to the show. Or it could be a boondoggle that costs more than it brings in. But given how the staff handled the venue change I think they’d be able to figure it out.

Something else that needs to evolve is my coverage of the show. One benefit of the new venue solving a bunch of problems is that I could spend more time talking about the actual content on the show floor. And the downside is that there’s so much more to see that I struggled with how to present it all. It’s tough writing blurbs without devolving into “here’s this table, and there’s that table.” I care about presenting the show honestly, and it’s daunting to weave a satisfying narrative when you’re staring down over five hours of recorded footage. I have a feeling I might split next year’s coverage into two videos: one for an overall review, and another one for in-depth coverage of all the tables. We’ll see what happens next year.

And for all I’ve said about venues they’re just a means to gather the amazing community that shows up to put on fantastic exhibits, find old machines new homes, and talk about their particular passions. The Midwest staff knows this and when the size of the Waterford became a barrier to that goal they moved to new digs. We don't go to a show to see the venue, we go to see the show! Vintage and retro computer fandom is powered by a sometimes irrational desire to keep ancient and forgotten tech alive. It’s infectious, especially up close when people get to try the hardware and software of their dreams. I always encourage viewers like you to check out vintage computing events, especially local ones. And if you decide to make the big trip out to VCF Midwest next year, it’ll be ready for you.

The Apple IIGS Megahertz Myth

August 04, 2024 by Dan Vincent in podcast

There’s many legends in computer history. But a legend is nothing but a story. Someone tells it, someone else remembers it, and everybody passes it on. And the Apple IIGS has a legend all its own. Here, in Userlandia, we’re going to bust some megahertz myths.

[A side note before proceeding… I see all you Hacker News people filing in. I haven’t had the time recently to properly format this with the numerous source links, citations, and footnotes that exist in the video version. I’ll try to get those filled in here ASAP. For now, you might want to experience it all in the video version.]

I love the Apple IIGS. It’s the fabulous home computer you’d have to be crazy to hate. One look at its spec sheet will tell you why. The Ensoniq synthesizer chip brings 32 voices of polyphonic power to the desktop. Apple’s Video Graphics Controller paints beautiful on-screen pictures from a palette of thousands of colors. Seven slots and seven ports provide plenty of potential for powerful peripherals. These ingredients make a great recipe for a succulent home computer. But you can’t forget the most central ingredient: the central processing unit. It’s a GTE 65SC816 clocked at 2.8 MHz—about 2.72 times faster than an Apple IIe. When the IIGS launched in September 1986 its contemporaries were systems like the Atari 1040ST, the Commodore Amiga 1000, and of course Apple’s own Macintosh Plus. These machines all sported a Motorola 68000 clocked between 7 and 8 MHz. If I know anything about which number is bigger than the other number, I’d say that Motorola’s CPU is faster.

“Now hold on there,” you say! “Megahertz is just the clock speed of the chip—it says nothing about how many instructions are actually executed during those cycles, let alone the time spent reading and writing to RAM!” And you know what, that’s true! The Apple II and Commodore 64 with their 6502 and 6510 CPUs clocked at 1 MHz could trade blows with Z80 powered computers running at three times the clock speed. And the IIGS had the 6502’s 16-bit descendant: the 65C816. Steve Wozniak thought Western Design Center had something special with that chip. In a famous interview in the January 1985 issue of Byte magazine, Woz said,

“[the 65816] should be available soon in an 8 MHz version that will beat the pants off the 68000 in most applications, and in graphics applications it comes pretty close.” End quote. That’s already high praise, but he continues further: “An 8 MHz 65816 is about equivalent to a 16 MHz 68000 in speed, and a 16 MHz 68000 doesn’t exist.”

If you read this in January of 1985 you’d have little reason to doubt Woz’s words. He built the Apple I in a bedroom from a box of scraps and when given real resources followed it up with the Apple II. Even when faced with doubters, he’s got the confidence that comes from engineering the impossible.

“Some of the Macintosh people might disagree with me, but there are ways around most of the problems they see.”

But that “should” in “should be available” was doing a lot of work. Eighteen months later when the IIGS finally shipped, there was no 8 MHz ‘816. It was as nonexistent as Woz’s imaginary 16MHz 68000. 8MHz chips were barely available three years later. What happened? Woz promised us 8 MHz of 68000-crushing glory!

If you poll a random vintage computer user they might posit the idea that Apple held the IIGS’ processor speed back during its development so it wouldn’t compete with the Macintosh. There’s an unsourced claim on Wikipedia that limiting the CPU speed to 2.8MHz was a deliberate decision followed by a note—which is sourced, thank you very much—that the original CPU was certified to run at 4MHz. And that’s true—there’s many IIGSes that have CPUs labeled for 4MHz. This idea’s made its way through various newsgroups and webpages for decades, so it’s no surprise that it made its way into a Wiki article too.

But this theory never sat well with me. People making the claim that Apple restrained the IIGS’ CPU speed for marketing purposes rarely provide sources to back it up. I understand their logic—Apple spent the majority of its marketing and monetary might making the Macintosh the machine of the future. Because the Mac was Steve Jobs’ baby, you end up with declarations like “Steve Jobs hobbled the IIGS so it wouldn’t compete with the Mac.” It’s a common take, especially because it plays into a lot of people’s dislike of Jobs. But there’s one major problem with it: all of Steve Jobs’ power at Apple was stripped away in May of 1985 after the months of executive turmoil that led to the company's first quarterly loss. The IIGS didn't launch until 16 months later.

So why were IIGSes with chips rated at 4 MHz not running them at that speed? Why 2.8 MHz? Isn't that… weirdly specific? Did an 8 MHz machine really get put on ice due to executive meddling? To solve these mysteries I descended into the depths of Usenet, usergroup newsletters, magazines, and interviews. My journey took me through a world of development Hell, problematic yields, and CPU cycle quirks. And this walk down the Apple chip road starts with the wonderful wizard named Woz.

The Apple IIx

It’s the summer of 1983 and Steve Wozniak has just wrapped up a two year leave of absence from Apple Computer. It all started with a six week stint in the hospital after crashing his Beechcraft Bonanza airplane on February 7, 1981. Amnesia from a hard concussion short-circuited Wozniak’s short-term memory. After his recovery he took a leave from Apple to go back to UC Berkeley and finish his degree in computer science and electrical engineering… and run a few rock festivals. By June of 1983 Woz felt he was ready to return to work, and asked Dave Paterson, the head of Apple II engineering, for a job—but this time, in the trenches.

His timing was excellent. Even though the Lisa was taking headlines and the Macintosh was shaking up R&D, the Apple II was making all the money. Brisk sales of the IIe along with the imminent launch of the IIc meant the Apple II division was busier than ever even if they weren’t getting all the glory. And while Steve Jobs was heralding the Mac as the Next Big Thing, designing a next-generation Apple II as a contingency plan was just good business.

At the heart of this proposed machine was a brand new CPU. Bill Mensch’s Western Design Center was busy developing the 65816, a 16-bit update to the venerable 6502 architecture. This chip would bring 16-bit computing to the Apple II while promising backwards compatibility. Users wouldn’t lose their investment in applications, add-in cards, or accessories. Alongside the new CPU was a special coprocessor slot that allowed the user to install a board with a 68000 or 8088. The goal was to build a bridge between the 8- and 16-bit world, so the project had code names like Brooklyn and Golden Gate.

This project would be known publicly as the IIx thanks to Wozniak discussing it on CompuServe or at user groups. But as late ’83 rolled into early ’84 the IIx project stumbled over multiple roadblocks. The coprocessor slot added layers of complexity that distracted from the mission of architecting a new Apple II. But a major complication was the 65816 itself. Apple expected engineering samples in November 1983, but didn’t actually get them until February 1984. What’s worse, those late chips were buggy, unstable, and ultimately unusable. WDC delivered a second batch of chips a few weeks later, but they were no more reliable than the first.

Even if Apple abandoned the coprocessor slot, the project couldn’t go forward without a CPU, and Apple cancelled the IIx project in March of 1984. Now before you aim your ire at Steve Jobs, A+ Magazine, in their IIGS development history, says it was the team leads who suggested canning the troubled machine. With no managerial appetite for a next-generation machine, the Apple II team pivoted from a moonshot to something more achievable. Dan Hillman and Jay Rickard started a project to consolidate most of the discrete chips of an Apple II into a single chip called the Mega II. When they finished the design six months later they weren’t quite sure what to do with it. Would they reduce the cost of the IIe or reduce the size of the IIc?

Imagine their surprise when Woz suggested a second shot at a 16-bit Apple II. The conditions seemed right to give it another go. Less expensive 16-bit computers like the Atari ST were looming on the horizon and the Mac’s hot start was slowing down. By October 1984 Apple finally had a good supply of working 65816 CPUs to design a system. And the Mega II would free up a lot of board space to add new graphics and sound chips. But just as important as chips and specs was a new, focused mission statement. This computer would be an Apple II by Apple II fans for Apple II fans. Woz, Hillman, Rickard, Harvey Leitman, and Lee Collings spent the rest of 1984 hammering out the specifications and solving hard problems like how to speed up memory access without breaking existing software.

Now we’re finally back to that Woz interview I quoted earlier. Byte published it in two parts across the December ’84 and January ’85 issues, and based on the average time to press I reckon it took place in October 1984. By this point the IIx is dead and buried and he’s talking about the new 16-bit Apple II, now codenamed Phoenix. His excitement about an 8MHz ‘816 is palpable, but, again, Woz was careful to say it “should be available soon.” Woz left Apple in February 1985 when the ink for this interview was barely dry. He had a dramatic fight with John Sculley after the Apple II was snubbed at the annual Apple shareholder’s meeting in January 1985. Apple II sales supplied seventy percent of Apple’s revenue in 1984 and Woz’s Apple II compatriots felt they weren’t getting their due. Steve Jobs may not have dialed back the IIGS’ clock speed, but he did shovel endless piles of money towards his pet projects at the expense of the Apple II. Even if Woz had stuck around the 8 MHz ‘816 of his dreams was years away. The IIGS wouldn’t sniff 8 MHz until Applied Engineering released the 7 MHz TransWarp GS accelerator four years later in 1989.

The Need for Speed

If you go looking online for photos of Apple IIGS logic boards, there’s a decent chance you’ll see a 4MHz GTE G65SC816 CPU. Most IIGSes had 4 or 3 MHz CPUs running at 2.8 MHz regardless of the chip’s rated speed. Why?

First, we must understand where that clock signal comes from. The IIGS, like many computers of its era, derives its base clock from a crystal oscillator. The one in the IIGS vibrates 28.636 million times per second, or megahertz. The VGC divides that 28.636 megahertz in half, and 14.318 MHz is then supplied along with a stretch signal to other parts of the system. I bet you've already noticed that these frequencies are multiples of the NTSC colorburst frequency of 3.58MHz. Keep that in mind, it’ll be relevant later.

This 14.318 MHz clock travels to a special ASIC which—depending on your board revision—is called the Fast Processor Interface or Control Your Apple chip. One of its responsibilities is dynamically controlling the CPU’s clock speed. The IIGS normally runs in Fast mode at 2.8 MHz, but the user can switch to Normal mode in the Control Panel which reduces the speed to 1.023 MHz. It’s like the turbo switch on PCs, except controlled by the FPI. This lets the IIGS run speed-sensitive software at the same speed of an original Apple II. But even in fast mode there are times the CPU needs to slow down to access other parts of the system.

The CPU, ROM, and Fast RAM are on the 2.8 MHz, or fast side, while the Mega II chip, slots, sound, video, and so on are on the 1MHz, or slow side. When the CPU is running in fast mode and needs to talk to something on the slow side the FPI dynamically throttles the clock signal to a 1 MHz cycle time to let the CPU synchronize with the Mega II side. This is usually invisible to the user because the system still executes the majority of its cycles at the faster speed, but it means the CPU is not always executing as fast as it could. I haven’t even touched on the eight percent performance penalty from the cycles the FPI spends refreshing RAM.

There’s nothing inherent to this design that limits the system to 2.8 MHz. The Apple IIGS Hardware Reference, Wayne Lowry’s Cortland Custom IC Notes, and Daniel Kruszyna’s KansasFest 2011 presentation about the FPI lay it out clearly that the IIGS’s fast mode could support higher speeds. In principle a redesigned FPI or CYA could use a divider other than 1/5 for the clock signal. A 1/4 divider of 14.318 MHz yields a 3.58 MHz clock speed, which should be well within the capabilities of a 4 MHz chip. And once again, that “should” is doing a lot of work. So why didn’t it run at that speed?

The Birth of the 65C816

The 65C816 and IIGS are inextricably linked, and the ‘816’s story starts in 1981 when GTE’s Desmond Sheahan approached Bill Mensch and Western Design Center about designing a CPU. GTE wanted to manufacture their own chips for their key phone systems, so they licensed both a CMOS manufacturing process and a 6802 CPU design from Mitel. But Mitel’s CPU design team wasn’t up to the task, so GTE asked Mensch to step in. Mensch offered two options: a two-chip 6802 solution or a 6502-based microcontroller. Either one could be manufactured on the CMOS process GTE licensed from Mitel. Sheahan convinced the GTE brass that the one-chip solution was the way to go, and the 65C02 was born.

GTE became the first licensee of the 65C02 thanks to WDC designing their 65SC150 telephony microcontroller. Eventually WDC would license the core to other companies, like NCR, Synertek, and Rockwell. The result was WDC netting a royalty every time one of these licensees sold a standalone CPU or microcontroller with its 65C02 core. Functional 65C02 silicon was available in 1982, and the revenues from licensing deals were finally filling up WDC’s coffers. This is when—according to Mensch’s Computer History Museum oral history and other sources—Apple approached him about a 16-bit successor to the 6502.

The prospect of a big player like Apple being the launch client for a new CPU was hard to resist. Further information on the early days of the ‘816 is fairly elusive, but looking at both historic and contemporary interviews with Mensch and Woz reveals Apple’s influence on the chip’s design. One influence was compatibility. When designing the ‘816 Mensch improved the 6502 bus architecture to eliminate little weirdnesses like false read cycles on store accumulators. This consequently broke Steve Wozniak’s Disk II controller, which had been designed to rely specifically on that exact weirdness.

Now there’s two ways to solve this problem: redesign the floppy controller, or add the weirdnesses back in. Apple tried the first one; redesigning its floppy controller into a single chip called the Integrated Woz Machine. This chip was made independently for the Apple IIc to control its built-in floppy drive. Among its many improvements was eliminating the false read cycle requirement. The Apple IIx could have used an IWM, but the UniDisk drives that took advantage of it wouldn’t be out until 1985. Therefore existing Disk II interfaces and drives still had to work with the IIx. If you were to claim Apple II compatibility, but not be able to work with one of the most popular boards, well, there’d be protests lining the sidewalks of De Anza Boulevard. Other cards might also depend on specific 6502 timing behaviors. Mensch eventually honored Apple’s request that the 65C816 be fully compatible with the NMOS 6502’s timings.

Apple wouldn’t be Mensch’s only licensee for the ‘816—the core can be found in countless embedded systems and microcontrollers. Licensees could extend the core to customize it into a microcontroller or a system-on-chip, or add features specific to their needs. A great example is Ricoh, who licensed the ‘816 core and added several functions, like dedicated multiply and divide units along with special DMA functions. All these add-ons were at the request of a pretty famous customer: Nintendo. The result was the Ricoh 5A22, the CPU inside the Super NES.

One of the famous tales of the ‘816’s development is how it was laid out without computer aided design. As told by Bill Mensch, he designed the ‘816 on a card table at WDC’s humble headquarters in Mesa, Arizona. His sister Katherine did the layout on mylar sheets. Many semiconductor designers had moved on to computer aided design tools to help design and lay out their circuits. Mensch is rightly proud of this feat, but the decision to lay it out by hand with mylar and rubylith wasn’t without consequences.

Let’s circle back to that interview with Woz. GTE published data sheets in 1985 for the G65SC816 which detailed 2, 4, 6, and 8 MHz variants of the chip. Apple, as the prime customer, would’ve had these data sheets far in advance. This would be consistent with Woz’s belief that faster variants were on the way, but for purposes of designing the IIGS they had to settle for 4MHz chips. Multiple photographs of IIGS prototype logic boards with 4MHz chips are on the web, and 4MHz parts shipped in production IIGS computers. But the promise of a faster CPU is very tantalizing to a computer designer, and I’m sure it wasn’t just Woz who was excited about future faster CPUs.

But when would GTE actually produce those faster variants? That’s the question. One source is a 1986 interview with Mensch published in the Fall/Winter issue of COMPUTE!’s Apple Applications magazine. This interview took place before the IIGS announcement, likely sometime in the summer of ’86. Mensch states their average part on a 3 micron process should be 4MHz, and an upcoming 2.4 micron process would yield 6MHz parts. I’d wager that the 8 MHz parts would rely on a 2 micron process at that reduction rate. Some 6MHz parts did exist, and Bill Mensch’s Cortland-badged prototype IIGS has one. A photo of it was shown at his 2021 VCF East panel, and I will note that it’s a WDC branded chip bearing a date code of the first week of February 1987. Whether Mensch put this chip in there himself or it was a sample installed by an Apple engineer is not explained. Nor is it known if its FPI chip actually drives it at a higher speed. But this shows that faster ‘816s did exist. So what was getting in the way of these faster chips actually being used?

Yields, Bugs, and Errata

This brings us to the 65816’s next quandary: yields. This might be a surprise to you, given that the CPU is found in embedded devices worldwide. But a common thread through many of the reports I’ve read about the early days of the ‘816 is that WDC and its fabrication partners struggled to deliver 4MHz and faster chips on time, in volume, and at spec.

Mensch positioned the 4MHz chip as the ‘816’s volume product and said as much in that earlier interview with COMPUTE!.

“Our typical product is 4MHz. We sell 2MHz into the retrofit market, but our typical run-of-the-mill is 4MHz.”

But in reality the IIGS shipped with a mixture of 3 and 4 MHz parts which actually ran at 2.8 MHz in fast mode. Which brings us back to the question of why a machine designed around a 4MHz part would ship with an almost 30% haircut in clock speed. Conspiracy theories aside, could there be a technical reason why the IIGS would run slower than its CPU’s rated speed?

In the same Fall/Winter 1986 issue of COMPUTE!’s Apple Applications where Mensch talked about future plans for the ‘816, David Thornburg interviewed Steve Wozniak about his role in developing the IIGS. The subject of the 65816 came up and Woz delved into some details about its clock speed.

“Our early ideas for the computer had it running around 8MHz. Soon we found we had to back off to about 5.5MHz, and then to 2MHz for that version of the processor. In the end the product came out around 3MHz, which is a good compromise.”

This is consistent with his comments about the desire for 8MHz more than a year earlier in the Byte interview. Woz doesn’t mention what factors made them back off on the clock speed, but during my research I learned a lot about the troubles the ‘816 faced in terms of scaling and yields.

One problem was GTE’s ability to supply chips—not just to Apple, but to other customers. The IIGS would be shipping by the tens of thousands when it launched, and this necessitated a good quantity of chips on hand. Dave Haynie—yes, the Amiga’s Dave Haynie—had trouble in 1985 sourcing 4 MHz 65816s for a potential Commodore 128 successor. He posted about this experience on Usenet in March of 1990.

“At that time, they had fully specced 8MHz parts, yet in ’85, GTE (the only company actually MAKING 65816s) had all they could do to make enough 4MHz parts. Rumor is that Apple managed get enough for the IIGS by actually having a special 2.8MHz version tested.”

He further elaborates with:

“When the GS came out, the only company making '816s was GTE. The main reason I couldn't get any 4MHz ‘816s in quantity was that Apple bought them all. They could make a real deal on 2MHz parts, since the yield on 4MHz was so low, they had more of those than they knew what to do with.”

Haynie also comments in other posts about how early samples of the ‘816 were delivered at 500KHz—yes, kilohertz—and maybe that’s a clue as to why Apple was unhappy in the Apple IIx timeframe.

Yields are a common woe in semiconductor manufacturing and Haynie’s comments about yields line up with what we see in the real world. GTE’s three micron process apparently had problems producing enough 4 MHz chips in volumes to satisfy its customers. Many IIGSes have a 3 MHz G65SC816 despite this speed rating not showing up in GTE’s data sheets. My guess—I can’t find any confirmation for this, but it's what makes the most sense—is that these were chips that couldn't be binned as 4 MHz, so GTE labeled them as 3MHz. Insisting on 4MHz would have meant shipping fewer computers, and the IIGS was delayed enough as it is. While VLSI supplied some 4MHz 65C816 CPUs later in the IIGS’ life, the vast majority of CPUs found in these computers were made by GTE—or, eventually, by California Micro Devices, which purchased GTE’s Microcircuits division in September 1987 after GTE decided to get out of the business. Apple was also negotiating with NCR as a second source, but according to Mensch and others, the deal fell apart before the IIGS shipped.

But let’s say for the sake of argument that GTE was able to produce as many 4 MHz chips as Apple or anyone else wanted to buy. Based on the FPI’s clock divider mechanism and a 14.318 MHz source clock, Apple had a logical clock speed target of 3.58 MHz using a 1/4 divider. That’d still be a compromise over Woz’s dream machine, but it’d be faster than what we got. And if (or when) faster processors became available, the FPI clock divider could be adjusted for them.

Yet those faster processors weren’t materializing; at least, not in any volume. Yields were a factor, yes, but the faster speeds revealed other problems. Trying to learn more about this took me down a rabbit hole of Usenet posts, Applied Engineering documentation, AppleFest event reports, and 6502 development forums. All of them pointed to a common factor: the REP and SEP opcodes. When designing the new 16-bit native mode for the ‘816, Bill Mensch added many new opcodes to enable new features and capabilities for programmers. Two of these new opcodes—called SEP for Set Status Bits and REP for Reset Status Bits—control flag bits for the processor’s status registers. These are crucial to the dual 8/16 bit nature of the ‘816 and how it can switch between 8 and 16 bit operations on the fly.

Unfortunately these opcodes proved problematic at higher speeds. Multiple posts relay statements from WDC or programming guides that timing problems with the layout mask prevented these instructions from completing in their allowed cycle times. These problems only got worse as they tried to shrink the mask down to smaller process nodes. I’m loath to take what amounts to second and sometimes even third-hand accounts from newsgroups and forums at face value—they don't call it the Net of a Million Lies for nothing, after all. But I’m willing to believe the overall theory based on other corroborating evidence (like this WDC data sheet note from 1991). If you look at an Apple IIGS accelerator like the TransWarp GS or the ZipGSX, you’ll notice that they’re not just a CPU and some cache memory. The TransWarp GS has a bunch of support chips and gate array logic, while the ZipGSX has a full-blown ASIC on board.

The GALs for the TransWarp GS were reverse engineered long ago, and Reactive Micro lays it out plainly: GAL3 handles opcode detection and speed control. This matches up with posts relaying comments from Applied Engineering about stretching clocks to handle problematic REP and SEP opcode timings.

Analyzing these posts also reveals the temperature of the Apple II community circa 1990. Apple announced a revised IIGS at San Francisco just before AppleFest in September 1989, and the reaction from attendees was muted at best. Unfortunately there was no CPU speed increase, but base memory was increased to 1MB and the new ROM added more toolbox routines and some feature enhancements. There was someone whose reaction was anything but muted, though, and it was one William David Mensch.

According to multiple accounts of the event, during his keynote address Jean-Louis Gassée said that there would be no speed increase for the revised IIGS because of difficulties securing faster 65816s. Mensch was in attendance and was understandably displeased with this statement. He approached one of the crowd mics and said that he was the designer of the CPU and had in his hand a bag of 12 MHz ‘816s. He proclaimed that if Apple ordered the 12 MHz chips, he would ship them. Jean-Louis reportedly made a comment about wanting a reliable production run, and the two men got into a heated back-and-forth before Gassée left and Mensch was escorted out. No recordings of this keynote exist, or if they do they’re locked away in the Apple Archives at Stanford University.

The version of the story Mensch recounts at his 2021 panel at VCF East largely follows what I’ve read in contemporary reports, except with one difference. He includes an anecdote about how he got Jean-Louis’ Apple coffee cup. He mentions running into Gassée after the keynote, and says that Gassée was very upset and threatened to, quote, “kick [Mensch’s] butt.” No punches were actually thrown, and no butts were actually kicked, and the story peters out without really explaining how Mensch got the coffee cup, but this story shows just how bad Apple and WDC's relationship had become.

Now, you’re a smart viewer, so I bet you know how chickens and eggs work. A company like Apple doesn’t just buy bags of chips; they buy Frito-Lay’s entire annual output. Mensch could have samples of silicon, but WDC wasn’t the one making the chips in volume; its licensees like GTE were. If the fab (say, GTE/CMD or VLSI) can’t guarantee a production run of, say, 50,000 chips, the order doesn’t happen. The evidence in front of us—the delays during IIx development, the inability to deliver 4MHz parts at volume, and the opcode issues that prevented scaling—would certainly justify skepticism of WDC’s ability to work with a fab to deliver faster chips at volume. There were still possibilities for a faster IIGS, though, and these would play right into an Apple II fan’s belief that Apple held it back.

Accelerators, ASIC Incorporated, and Mark Twain

But let’s say you weren’t buying tens of thousands of chips like Apple was; maybe you only needed a thousand or two. Were smaller customers being taken care of? Ray Settle of the Washington Apple Pi Journal was also at the fall 1989 AppleFest, where he reported on the confrontation between Mensch and Gassée. Afterwards, he mentioned a visit with an Applied Engineering representative. Settle still hadn’t received his TransWarp GS, and the AE rep pinned the blame on unreliable 7 MHz CPUs. Another attendee report posted on Usenet by Michael Steele featured similar comments. Keep in mind that the TransWarp was first announced in 1988, didn’t ship until the fall of 1989, and struggled with speed and supply restrictions through 1990. This is further supported by A2-Central’s accelerator reviews in February 1991, where it’s mentioned that AE resorted to offering 6.25 MHz versions of the accelerator because of supply issues—and reliability issues—with 7 MHz chips. Zip Technologies also took a while to ship their 7MHz ZipGSX accelerator, which finally appeared in October 1990, almost a year after the TransWarp GS.

But we don’t have to settle for second-hand reports. Enter Applied Engineering employee John Stephen III. In an October 1989 Usenet post he mentions the problems with getting 7MHz chips, and alludes to the REP/SEP timing issues. But the other interesting thing he mentions is that getting high speed 10 MHz ‘816s to run at their rated speeds required boosting their input voltage well beyond the standard 5 volts. This made the chips run hotter and often resulted in crashes. And I see all you overclockers in the audience nodding your heads.

Scaling problems aren’t unusual when you move to a smaller process node, and sometime a layout needs fixes—or even a complete redesign. The original mask laid out by Katherine Mensch on the WDC card table had reached its limits. Redoing the mask wouldn’t be easy or cheap, especially when higher speed ‘816s were a smaller piece of the revenue pie. Royalties from the 65C02, 65802, and slower embedded ‘816s were paying the bills. Mensch was also busy working on a 32-bit iteration of the architecture: the 65832. But this proposal never made the jump from datasheet to shipping dock.

Interestingly, this is where a new player steps in. While researching the yield problems I encountered numerous posts on comp.sys.apple2 about an “ASIC 65816.” This tripped me up at first, because there are numerous application-specific integrated circuits that contain an ‘816 core. But no, it turns out that a company named ASIC was creating a new 65816 processor using gate arrays. And they promised that this redesigned ‘816 would reach speeds of 20MHz and beyond.

Imagine my surprise when I saw the name Anthony Fadell mentioned in these posts. Could that be the same Anthony Fadell—Tony Fadell—who was in charge of the iPod? Sure enough, I found an excerpt of Tony’s book, Build, where he talks about designing a 65816 for his startup company, ASIC Incorporated! Now we’re on to something. This gave me enough clues to dig up the November/December 1989 issue of The Michigan Alumnus magazine, where an article tells the tale of the fateful summer of 1989. Fadell was an undergrad student at the University of Michigan, and he spent his summer internship designing nuclear control computers at Ronan Engineering. When he met William Hayes the two hit it off immediately. Fadell was a big Apple II fanboy and yearned for more power. Hayes knew chip design and had connections with other chip designers. The two decided that they would design a better version of the 65816 using a sea-of-gates array. Fadell borrowed $5,000 from his uncle, Hayes leveraged his connections to get cut-rate pricing on supercomputer time, and the two reverse engineered a 65816 design in six weeks. After finding a fab willing to manufacture their design, Fadell was prepared to pitch his prototype to potential patrons.

The Michigan Alumnus article is coy about who the chip was for, but it mentions Fadell flying to Texas in September 1989 to meet with a manufacturer of Apple IIGS accelerator boards. There he negotiated a purchase order and a two year contract, catapulting him to the role of CPU vendor at the age of twenty. With these clues we can deduce that Fadell’s customer was Applied Engineering. If all had gone according to plan, ASIC's chips should have started production in early 1990, and formed the basis of the next generation of TransWarp accelerators. There’s that word again—should.

Ultimately, no TransWarp GS or ZipGSX accelerators ever shipped with ASIC’s design. The chips did exist—multiple trip reports from 1990 AppleFests mention Fadell demonstrating his CPUs in TransWarp accelerators. And in 2022, Fadell posted a photo on Twitter of an ASIC 65816 which he claims would still work today—I’m guessing this is one of the 17 MHz chips used in the AppleFest demos. But posts about ASIC fizzled out after the spring of 1991, which coincides with Fadell graduating from Michigan. The ultimate fate of the chip isn’t really known—did Fadell face legal challenges from WDC? Did Applied Engineering give up on him? Or was it because—as described in an interview with WIRED—he was just too busy roaming the halls of General Magic trying to score a job with former Apple superstars Andy Hertzfeld and Bill Atkinson? My guess is the latter, since General Magic hired him later that year.

In the same tweet as the photo of the chip, Fadell said that he, quote: “sold some to Apple for a new Apple IIGS revision before they canceled it!” Many IIGS superfans have heard tell of the cancelled final revision of the IIGS code named Mark Twain. Featuring an internal floppy drive, a hard disk, and a redesigned logic board the Mark Twain is what many thought the ROM 03 IIGS should have been. It’s entirely probable that Apple fitted some prototypes with faster CPUs. But when media outlets like InCider/A+ magazine were given a top secret demo a mere month before its cancellation, the clock speed was still the same. And the few Mark Twain prototypes that escaped Apple’s development dungeons were equipped with the usual 4MHz chip. This is where the business side and the technical side butt heads.

The Mark Twain was under development in late 1990 into 1991 and then mysteriously no longer under development as of June 1991. Rumors of Apple killing the entire Apple II line hung over the product like a dark cloud, and the dev team had hoped to prove they were greatly exaggerated. Despite John Sculley’s statements promising Apple’s full support for the over five million strong installed base, the lack of marketing and technical improvements to the Apple II over the years meant his words rang hollow. Apple had just introduced the Macintosh LC as their low-cost color Mac, and IBM compatible PCs were getting cheaper and more capable. If Apple released the Mark Twain without any CPU speed boosts, it would’ve appealed mostly to the Apple II’s cost-conscious institutional education market. Would existing IIGS owners buy one instead of just getting an accelerator card for a fraction of the price? And how many new users would it bring to the platform? The Mark Twain would’ve been like the Amiga 1200: a decent improvement, but ultimately too little and too late. The March 1991 release of the Apple IIe card for the Mac LC also put another nail in Mark Twain’s coffin, because many users—especially educators—bought a IIGS for backwards compatibility with classic Apple II software. If you’re the number cruncher who had to choose between spinning up a run of 50,000 Mark Twains that cost a lot more to build than 50,000 IIe cards for Mac LCs that are already in production, already in a warehouse, or already sold to customers, which would you pick?

Now this is where you can rightfully criticize Apple for holding back the Apple IIGS. A Mark Twain with even a 12MHz CPU from ASIC would’ve been a legitimately powerful, well equipped computer for its price. But that would’ve been one last campaign in a war long since lost. Maybe ASIC could have helped, but the good ship Apple was sailing straight into a storm called the beleaguered era. Costs, staff, and projects were starting to spiral out of control, and the Mark Twain would’ve only delayed the inevitable death of the Apple II.

Sanyo, Nintendo, and the End of Acceleration

Even if Apple didn’t see fit to release a faster IIGS, accelerator cards kept the enthusiast market alive for a few more years. Upgrade guides for the TransWarp gave tips on how to install higher-rated ‘816s to get even more speed. This usually required buying samples of higher speed processors from WDC, changing crystals, upgrading the cache, and acquiring new GALs. Once you hot-rodded your card you’d often need to supply more juice over the 5V rail to keep things stable.

All this hackery was finally put to bed in 1992 when new 65C816s rated at 14 MHz hit the market. These chips took Usenet posters by surprise, especially after the ASIC saga. Fabbed by Sanyo, the 14 MHz models could run cool and stable at 5V and apparently solved the issues with the REP and SEP opcodes. Sanyo achieved this by abandoning the die laid out by Katherine Mensch and creating a new layout from scratch. Why Sanyo chose to invest in this is unclear—I found a lot of speculation that they wanted to build a PDA based on the ‘816. Sanyo’s a giant manufacturer, so I’m sure they found a use for it. Maybe WDC felt the heat from ASIC, or maybe they saw ARM and 68K pivoting to the embedded market and finally bit the bullet to stay competitive.

Existing accelerators could be upgraded by changing out their CPU and clock crystal, but by this point the IIGS was but a fraction of the ‘816s out in the wild. Optimistic estimates of the number of IIGSes shipped hover around 1.25 million. The other ‘816 system that most people know—the Super NES—sold 1.3 million units in 1990 just in Japan according to numbers compiled by NeoGAF user Aquamarine, who claims to have gotten them directly from Nintendo's Kyoto offices. The system sold nearly 50 million units worldwide in its lifetime. Mensch is very proud of the SNES using the 65816 and speaks highly of working with Ricoh, the manufacturer of Nintendo’s chips. Seeing as the 5A22 was a custom job by Ricoh, I wondered if they fixed the SEP and REP problem during its design. It wouldn’t be beyond them; they did tweak the 6502 to dodge patents when designing the NES’ 2A03. I haven’t seen the SNES assembler community complain about issues with REP and SEP with the basic 3.56 MHz part, but that doesn’t mean problems don’t exist. Same with emulation and FPGA groups, though I’d still defer to experts in that field. Checking a high-resolution decapped scan of a 5A22, the CPU core looks very similar to the first-generation layout seen on the Programming the 65816 book—I’m guessing it still has the REP/SEP flaw.

And like the IIGS, the SNES could take advantage of accelerators thanks to add-on processors in cartridges. The most well known is the SuperFX, but Ricoh also made a faster 65816. Better known as the Super Accelerator-1, the Ricoh 5A123 runs at a 10.74 MHz clock speed—three times faster than a 5A22. You’ll find it in fan-favorites like Super Mario RPG and Kirby Super Star. SA-1 games started shipping in 1995, which is late in the Super NES’ lifecycle. I haven’t found a decapped scan of the 5A123, but nintendo was shipping 1-Chip SNES consoles around the same time. The 5A122, which combined the CPU along with the SNES’ custom PPU chips, formed the heart of the 1-Chip console. Decapped scans of that chip exist, and its layout looks very similar to the post-Sanyo core. I’d bet that the 5A123 has that same core, and doesn’t suffer from REP and SEP issues.

ARM, Möbius, and The Road Not Taken

Apple discontinued the IIGS in December 1992. Even though the IIe hung on for another year or so, the platform truly died that day. Even in the timeline where WDC had 8MHz chips ready in 1983, and Apple put the GUI on the IIx in 1984, I still think the Apple II as we knew it would’ve died eventually. There’s several limitations that would necessitate an eventual migration to a new architecture.

The primary roadblock is the Mega II side of the machine. This is supposed to be what makes it an Apple II, but it drags the rest of the machine down with it. Depending on the video system for clock generation and timing was a practical engineering choice that became hard technical debt for almost all 1980s computer architectures, especially with ones that traced their roots to the 1970s. The GS is an excellent example of trying to maintain compatibility while adding capability, but it had an obvious ceiling.

But something that gets lost in the 65816 versus 68000 arguments is that CPU architectures have qualities beyond raw speed. You might care about power consumption, memory mapping, or ease of acquisition. And all these factors are a balancing act depending on your application. The 68K’s massive flat memory space was a big point in its favor, as well as its native support for user and supervisor separation. These don’t matter as much to someone who’s writing hand-crafted single-tasking assembly language apps, but they very much matter to someone building a multitasking UNIX workstation.

And that’s not to say that 68K isn’t good for assembly applications. It’s got a great reputation among assembly programmers. But as the world turned to compilers and development environments the 68K architecture was primed for the rising popularity of languages like C. More people were getting into programming, and it’s unrealistic to expect them all to have the potential to become machine language maestros like Rebecca Heineman or Nasir Gebelli. C compilers exist for 6502 and 65816, but it's fair to say that these architectures weren't particularly suited to the intricacies of C.

Another sticking point for high-speed 65816s is the speed of memory. Assuming you threw the entire Mega II side of the IIGS away and built a new ‘816 computer from scratch, a 14 MHz example would need very fast main memory to operate without a cache. In the early 90s, that kind of memory was barely available. How Woz would have built his 8 MHz IIGS in 1985 while affording a decent amount of memory is a rather inconvenient question.

Apple wasn’t the only company facing the limitations of their early designs. Commodore and Atari decided to pivot to the 68000, like Apple did with the Macintosh. Tech debt eventually caught up with them too, especially Commodore, but the problems weren’t unsolvable—the Amiga tried to migrate to PowerPC, after all. Even the Macintosh and IBM PCs had eventual reckonings with fundamental planks of their platforms. Another company was facing the same conundrum: Acorn Computers. The similarities to Apple are there: they were dependent on a 6502-based architecture and had a crucial education market with a significant installed base. Acorn did ship a computer with the 65816—the Acorn Communicator—but when Sophie Wilson visited WDC in 1983 and saw Mensch and crew laying out the 65816 by hand, it struck her: if this motley crew on a card table could design a CPU, so could Acorn. Thus Wilson and Steve Furber forged their own CPU: the Acorn RISC Machine.

Though today’s Arm and WDC sell very different CPUs, they do have something in common: they’re fabless semiconductor designers that license their cores and architectures to other companies. Apple is of course Arm’s most famous partner: they joined Acorn and VLSI to form Advanced RISC Machines in 1990 to create the ARM610 CPU to power the Newton. But what you might not know is that Apple’s involvement with ARM originates with a desire to replace the CPU in the Apple II. The little known Möbius project helmed by Paul Gavarini and Tom Pittard in the Advanced Technology Group was an ARM2-based computer that could emulate 6502, 65816, and 68000 code. Gavarini and Pittard started the project in 1986 and were demoing and compiling benchmarks in 1987—right on the heels of Acorn releasing the Archimedes! There’s little information about this on the web, with Tom Pittard’s bio and Art Sobel’s ARM pages being some of the surviving hints to its existence.

Based on my knowledge of how ARM2 works, I believe the emulation performance of Möbius is wholly derived from the ARM2’s memory system. ARM’s designers were inspired by the 6502’s efficient memory access and optimized the 8 MHz ARM2 to wring as much performance out of the available memory as possible. By using 4 MHz 32-bit-wide fast page memory, pipelining, and special load-store instructions, ARM2 could perform burst transactions at twice the speed of random ones. With a theoretical maximum of 32 MB/s bandwidth in fast page mode, this was eight times the maximum bandwidth of an 8 MHz 68K shackled to 2 MHz DRAM. This strategy would peter out eventually because memory speed couldn’t keep up with CPU speed, but hey—that’s what cache is for!

I’m not sure if Möbius would have been the Apple II’s savior—Acorn’s Archimedes wasn’t 100% backwards compatible despite including a BBC Basic interpreter and eventually a BBC Micro emulator. But with EcoNet network adapters and expansion podules to connect old Beeb peripherals the Arc could ingratiate itself with Acorn’s existing installed base. Could the Apple II have been reborn with an ARM CPU? Maybe. Nobody mentions how well Möbius integrated the rest of the Apple II architecture like slots or video or sound. And say what you will about Apple’s troubles with Western Design Center; ARM was barely a blip on the radar in 1988. Apple wasn’t going to upturn their cart for an unproven architecture from a competitor. Möbius was a skunkworks project; it could’ve taken years to turn its demo into a shipping product and the 68040 was already on the drawing board in 1988. But it was still worth it: Möbius’ benchmarks convinced Larry Tesler that ARM could save the Newton from the disastrous AT&T Hobbit processor. And hey—without ARM, Apple wouldn’t have the iPod, iPhone, and Apple Silicon today. So it worked out in the end.

The End of an Odyssey

What a ride, huh? Thanks for making it this far down a fifty-plus minute rabbit hole. I can’t claim that this is the final take on the subject—so many of the players aren’t on the record, but I’m pretty confident in saying that Apple did not artificially limit the IIGS’ clock speed during its development for marketing purposes. Now, I’m not a fool—I know Apple didn’t push the IIGS as hard as it could, and it was very much neglected towards the end of its run. If the REP/SEP flaws hadn’t existed and GTE could’ve shipped stable 4 MHz chips in volume, I’m sure Apple would’ve clocked them as fast as possible in 1986.

I’ll admit that I initially started this deep dive out of spite. The idea that “Steve Jobs deliberately held the IIGS back, bleh bleh” is everywhere, but it's all just people saying things in an endless game of telephone, with no actual evidence. That’s enough to grind anybody’s gears, but what’s worse are people who I respect uncritically repeating these things in videos and blog posts. It hurts me to see videos with millions of views repeating old Internet urban legends pushed by partisans filtered through Wikipedia articles with bad citations.

But that spite subsided quickly once I started untangling the messy web of people and circumstances that wrapped around this story. I realized that what I wanted wasn’t to prove anybody wrong. What I wanted was to get people to think about these stories and why they became legends. Of course, you could flip that right back around at me and say “who made you the guardian of love and justice?” And that’s a fair point. But my goal here isn’t to push an agenda, but to get a better of understanding of how things happened and why history went the way that it did. I’ve provided my evidence, and it’s up to you to judge if my argument is compelling enough.

And even then, one of those people who needed a refresher on computer legends was yours truly. I’ve done my share of repeating things based on bad sources, just because I had a fanboy desire to defend something. Or because I just thought a story was neat and didn’t look too deeply into it. It’s not so much about somebody who’s being wrong on the internet; it was the realization that I could be somebody who’s being wrong on the internet! It was humbling, really. As the years go on I’ve realized that there’s so much out there to learn, even if I thought I already was an expert. A perfect example is that Bill Gates BASIC easter egg, where I got tripped up by an oft-repeated legend until I actually dug into it. And as vintage and retro tech enthusiasts, are we not people of science? Are not our minds open to new ideas? We’re into this because we enjoy the history and personal connections, and we should all be excited about digging deep and not just repeating the same old story.

Even though the IIGS may not have been able to unleash its full potential, it’s still an amazing machine even at its base speed. If you haven’t had the chance to play with one, give it a try. And turn on accelerator mode in your emulator to get a feel for what could have been.

Installing OS/2 on a PS/2 Model 80: 5000 Subscriber Special

July 02, 2024 by Dan Vincent in podcast

Hi, I’m Dan. You may remember this PS/2 Model 80 from such videos as the IBM 386 Tower of Power. Back in January I threatened you with the good time of installing OS/2 on this beast, and today’s your lucky day. Here, In Userlandia, we’re getting warped.

OS/2. Other than some hands-on time at vintage computer festivals in the past few years I’m a complete noob when it comes to the quote “Better Windows than Windows.” I never saw a live Warp 3 system during the Nineties, and I touched Warp 4 once or twice in the Aughts. But I do have a fascination with operating systems and how we interface with computers. And when I found this Ten Minute Guide to OS/2 Warp at the thrift store a few weeks ago it reminded me that yes, I promised you an installation of OS/2 on this mighty Model 80. And I’m a nerd of their word, so let’s cash in that IOU and take this machine to Warp.

Today I’ll be installing OS/2 Warp 3 Connect Blue Spine with the OS/2 Bonus Pack. Sorry to disappoint anyone who wanted to see me swap forty-plus floppies—I’m not that much of a masochist to risk one bad floppy torpedoing the entire install. What sets the Blue Spine edition apart from this boxed Red Spine edition is that Blue Spine comes with Windows 3.1. Red Spine was pitched to users who already owned Windows, and ergo save IBM and that customer the cost of a Windows license. Using a Blue Spine CD-ROM will save me from finagling more floppies, and Connect includes local area networking. I’ll also need to apply Fixpacks at some point, but I’m leaving that for another day.

Returning viewers might recall that this Model 80’s equipped with a 90MB hard disk and no CD-ROM drive. Instead of messing around with other disks and drives I’m using a BlueSCSI to emulate a hard disk and CD-ROM. This requires booting from the reference disk and sitting through minutes of clunks and chirps while it auto-configures. But my patience was rewarded as it successfully identified the new drives. Huzzah! Next, installing from CD requires booting up from a pair of floppies. After I chose my desired installation type—I’m an advanced user, and I need an advanced installation—the installer partitioned the hard drive image and forced the first of many restarts.

Another reason to choose Advanced Install is to set up IBM’s Boot Manager, which lets you boot multiple operating systems from various drives or partitions. It’s fairly robust and easy to use boot loader for the era, You can also pick the file system for your hard drive: MS-DOS FAT or OS/2’s High Performance File System. HPFS provides nice features like long filenames and large volume support, so I’ll choose HPFS.

I’ve got some time to kill while the installer installs, so let’s chat about OS/2. What I won’t be doing is an exhaustive history of OS/2 in this video. If you’re interested in good modern takes on OS/2’s turbulent history you should check out the excellent retrospectives by Another Boring Topic and RetroBytes. This is more about exploring what it’s like to run OS/2 on a 386. I missed out on OS/2 back in the day, mostly because Microsoft and Apple consumed so much air during the mid-Nineties in the US. When I started reading mainstream computer mags in 1995 the hype train for Windows 95 was full steam ahead. I saw OS/2 ads in PC/Computing but the actual editorial content barely discussed it. A three-way comparison between OS/2 Warp, Windows 95, and Windows 3.11 in the August 95 issue was about the most exposure I got to Warp’s headline features.

What little I read about OS/2 was more than I’d seen in person. OS/2 was completely absent from my schools, even in the tech labs that ran powerful software like CAD programs. I grew up in Digital country, where we used DEC terminals to search library card catalogs that ran on VMS. Accessing email at school required telnetting into a UNIX shell to run Pine. Maybe the ATMs for our local banks used OS/2, but I never knew for sure. Then again, I had the same lack of experience with other OSes like NextStep, Solaris or IRIX.

So far the installation is going well, which wasn’t always the case. Warp’s installer was a major improvement over 2.0’s, but I’ve still seen several sorrowful stories of setups gone awry. To be fair, installing on a PS/2 is basically doing this on easy mode. Users on clone systems might not be so lucky depending on their BIOS and cards. Straying outside of IBM’s hardware garden in early versions of OS/2 was a path to pain, although it did get better in Warp 3 and 4 because each version included more drivers. I don’t think Warp completely shook OS/2’s reputation for brittle installers, but I’d love to hear the tales of your installation successes or failures.

Let’s check back in on the installer. The first phase is nearly complete, and after a reboot it’ll boot from the hard drive to start the second phase. This is where I learned a quirk about the MCA Tribble SCSI adapters—apparently they can only boot from drives that are less than a gigabyte. This and other imitations can be lifted if you swap in newer ROMs from a Spock adapter. I mention this because I originally started with a 1GB disk image and found myself stuck at a blinking cursor after the first phase completed. I was more successful after switching to an 800MB disk image and re-running the installer.

At the start of phase two I’m asked to confirm hardware devices like graphics, sound, printers, and ports. Once again installing on a PS/2 made things easy—it’s all detected and ready to go. The next step is selecting OS/2 software components, and I’ll select them all for a complete install. Lastly, since I’m installing Connect I must choose my networking stacks. Might as well YOLO and install them all! Interestingly, it didn’t auto-detect my Western Digital ethernet card even though it’s configured in the PS/2 BIOS. I thought I’d have to load a driver from a floppy, but after a quick Google I learned that the bundled Standard Microsystems EtherPlus MCA driver was compatible. I selected it and the installer carried on.

While phase 2 continues let’s investigate what’s inside an OS/2 Warp box. This Red Spine edition—along with most of the boxed operating systems and office suites you see on my shelves—was part of a software collection I acquired from somebody decluttering their garage. As far as I can tell this is a complete copy that’s never been used. The top layer of this box cake is a sub-package for the OS/2 Bonus Pack. Inside is a license agreement, a thin guidebook, and an old disclaimer scroll warning you about the dangers of the internet. After thumbing through the inserts you’re left with sixteen floppy disks, which is almost as many as the OS itself. I don’t envy whoever drew the short straw to babysit floppy installs back in the day. I’ll babysit a CD-ROM Bonus Pack instead.

Next is a 400 page-plus OS/2 Warp User’s Guide. Printed in glorious black, white, and Pantone 286 blue, it explains the basics of installing and using OS/2. It’s pretty thorough, with multiple chapters on troubleshooting and fixing compatibility problems. OS/2 Warp included an interactive tutorial, so you might not need most of the getting started part of the manual, but it’s there in case you’d like some light evening reading.

If that’s not enough reading material, then check out the Fall/Winter 1994 edition of IBM Sources and Solutions. This tome cataloged OS/2 compatible hardware and software along with qualified vendors and consultants. Some fun IBM ads are sprinkled throughout, but the most entertaining bit was finding all the defunct stores and consultants in my area. RIP to Egghead and Software Etcetera. And I have to call out the copy editor for letting misspellings of Worcester, Massachusetts escape their red pen. For shame, for shame!

It wouldn’t be a software box without registration and support cards. The OS/2 registration card is pretty straightforward—fill out the fields and and drop it into a mailbox to tell Big Brother… er, Big Blue all of your demographic info and usage habits. You can rest easy knowing that you’ve sold your personal data to a major corporation for absolutely nothing in return. There’s also a technical support card entitling you to sixty days of telephone support starting from your first call. I dialed the 800 number just for laughs and it goes to what I presume is a scam to get seniors to sign up for a knockoff Life Alert device. Bummer.

IBM also saw fit to include ads for several third party software products. Golden ComPass is an assistant for CompuServe that downloads emails, news, forum posts, and so on for offline consumption. Computer Associates’s CA-Realizer is a graphical BASIC development environment. Watcom VX-REXX is another visual development tool, this time for IBM’s REXX scripting language. And Adobe wanted you to buy some fun font packs to leverage OS/2s built-in Adobe Type Manager. $80 for 64 fonts doesn’t sound too bad for the time, though I’d classify most of these typefaces as gimmicks.

Lastly, there’s the requisite legal paperwork and last-minute technical bulletins. The license agreement isn’t as interesting as this warning about XDF-formatted diskettes, or a tip sheet for installing OS/2 on top of an existing Windows setup. But the most fascinating one is this folded sheet asking you to call IBM if you’re installing OS/2 on an IBM Aptiva. Based on the Aptiva’s launch date my hunch is these boxed copies might be missing some necessary drivers.

After digging through all this paperwork we’ve reached the treasure hidden at the bottom of this beastly box: a cache of 21 floppy disks. For those keeping score, that’s eight more than the retail RTM version of Windows 95 and one fewer than Windows NT 3.1. And that number’s actually lower than it could be thanks to XDF formatted floppies cramming 1.86MB of data onto standard 1.44MB high density disks. Combining these floppies with the bonus pack makes for a whopping 37 disks, and a Blue Spine box would have even more for Windows support.

With the lengthy phase two finally complete, we’re on to the even lengthier third and final phase: OS/2 Warp Connect. It gets its own special installation display and partway through it switched to a video mode that none of my LCDs liked. I had to dig out a CRT to track its progress, and whatever mode it used was more unfilmable than usual. So in lieu of that, there’ll be some Bonus Pack installation footage instead. Check out that happy little PC!

And while that installs, this wouldn’t be a 5,000 subscriber special without a 5,000 subscriber special channel update. I’m humbled by the positive reception to my videos, and I genuinely appreciate the time you spend watching, discussing, and sharing my work. I resisted YouTube for a long time because I’ve always had a rocky relationship with recording and editing video. That’s a bit ironic considering that in 2007 it was Slowbeef, Maxwell Adams, and yours truly who respectively made the first, second, and third video Let’s Plays on the SomethingAwful forums. I hung around as one of the original LP superstars until I bowed out around 2012 to concentrate on my convention artist alley tables.

I’ve been writing about computers here and there for almost twenty years, though most of that material has vanished after various websites went offline. If you’ve been listening since the Icon Garden days you might remember my solo Macinography podcasts, some of which might be adapted into future videos. Although my style’s changed a bit since then, what I launched as Userlandia in 2021 is broadly similar to that work. But after a year and change of Userlandia not getting much traction as a blog/podcast I conceded to the reality that the audience for what I’m doing is on YouTube.

I made the switch to video with what amounted to a glorified slideshow for the 2022 VCF Midwest podcast. It was also my first time using DaVinci Resolve. My prior video editing experience was with Camtasia or iMovie, and the one time I tried Premiere didn’t leave me with a good impression. But I picked up Resolve pretty quickly—as it turns out my skills as an audio editor translate pretty well to editing video. But as time went on I realized I needed to take my production to the next level and record actual footage. Thus began my crash course in motion picture filming. My video quality has improved compared to my early attempts, but I’ve still got a ways to go.

Most of my work starts with a script. For something like a Computers of Significant History video I start by researching a subject, taking notes, and writing an outline. Scriptwriting can take weeks, depending on the complexity of the subject and whether I need to run things by an editor friend for a double-check. Once I’m happy with a script I record and edit audio, which takes around three to four hours for a half-hour track. Then I compile a shot list to figure out what needs to be filmed and what gaps can be filled with stills, clips, or gag reels. My camera moves are fairly basic—pans, tilts, slides, zooms, and focus pulls. All the footage and resources are cut together to form a video, and then I polish it up with some background music. This process is largely the same for my trip report videos, except the footage is recorded before scriptwriting. And this video is a bit of a hybrid—I wrote some script ahead of time, recorded some footage, and then wrote some more during and after the installation.

Sticking to a script has its pros and cons, but it’s probably the best for me. I have a lot of respect for people who can speak extemporaneously to a camera, or narrate off-the-cuff over footage they’ve edited together. Being on camera is difficult for me, and until recently I didn’t have enough room to really try it. Over the past month I’ve picked up a wireless microphone and rearranged my filming room to create more space, so I may try being on camera in the future. Another complication is my filming camera: a Sony a99ii. It’s great for stills photography, and its mediocre video capabilities weren’t really a concern when I upgraded from my a99 mark 1 in 2018. But now that I’m recording more video I’m crashing right into its limitations. It overheats when recording long 4K clips and its 1080p output is plagued with moire on screens and textures. There’s also many restrictions on video autofocus, and tracking moving objects or my face is a real challenge. At some point this year I’ll migrate to E-Mount—probably around my trip to Japan in November—but I’m making do for now.

All these factors influence my production style. My goal is to produce long-form narrative stories, like telling my personal history of personal computing with Computers of Significant History. Or examining a specific machine like the SE/30 or this Model 80 from an angle you might not have seen before. I don’t think I’d bring my best material if I filmed my unscripted thoughts. And I don’t think that kind of video is bad—far from it! What I am saying is that it doesn’t play to my strengths. There’s tons of talented people making great walkthrough and talk-through material, and they all deserve your attention.

But I believe one of the best pieces of advice for anyone in a creative field—whether you’re an artist, writer, videographer, whatever—is to make the things you want to see. I was raised on a healthy diet of The New Yankee Workshop, Shadetree Mechanic, and The History Channel. YouTube is full of people making shows like these but for computers, which spoils me for choice. But I love watching people who know what they’re talking about sharing their expertise and enthusiasm with others while showing they can learn and grow. That’s what I want to bring to the table.

LGR remains the gold standard in this space for a reason: he’s constantly tweaking and improving his production while consistently releasing new material. Every video might not be to every viewer’s taste, but his frankly astonishing ability as a one-man band to publish polished pieces on a near-weekly basis means if you’re not feeling this week’s subject the odds are pretty good you might vibe with next week's. Clint makes what he wants to make and even when things go sideways he still commits to the bit. This is on top of the fact that he’s constantly iterating his production in small but noticeable ways. He never assumes he knows everything and he's open to suggestions and constructive criticism towards improving his work. I was really flattered when I got some comments on the Model 80 video that compared me favorably to LGR. We share a lot of common influences and tastes, after all. But I’m not trying to be LGR, because he’s already doing a great job of being LGR. I want to do my own thing.

Lastly, LGR is one of many creators who’re willing to spotlight other people’s channels and products or lend their expertise when somebody’s in a bind. Folks helping folks succeed is honestly one of the best aspects of this community. This stuff isn’t zero-sum—helping somebody with their production or shining a spotlight on a new creator won’t detract from the success of your own work. Keeping the community’s enthusiasm high will lift the boats of view counts and watch time for everybody.

I also want to share some plans about future projects I have coming down the pike. The biggest one is a nearly hour-long video with a working title of “The Apple IIGS Megahertz Myth.” It’s about the development of the 65C816 CPU and the IIGS leading to a fraught relationship between Western Design Center, Apple, and the Apple II community. I’m pretty happy with the state of the final draft—I finished the script in March right before my trip to England—but I need to record the audio and assemble all the video coverage. One reason I’ve put off recording it is because I’ve been using Invisalign for the past ten months. My treatment will be done by the time you hear this, but it’s put a real damper on recording sessions. You’re not supposed to take out your aligners for more than an hour at a time, and recording with trays on my teeth was a no-go. Hopefully I’ll get video production going by the end of June and maybe release in late July or early August. Aside from that mega-project I’ll aim to produce something a little lighter to fit in-between that and VCF Midwest. Yes, I’ll be there and I’m planning my usual coverage.

How about Computers of Significant History? That’s a project with a definite end point, though I don’t see myself getting there for a while. I have three machines in the queue: HP Minitowers of the year 2000, the Titanium PowerBook G4, the Blue and White G3, and my first self-built computer. I’ve already acquired the hardware and the next step is writing the scripts and producing the videos. My plan is to use them as a springboard to discuss online communities and consumer operating system transitions. Obviously I’ll run out of PCs for this timeline, but I’m planning a series about software which I’ll probably call Software of Considerable Importance. Just in time for the thirtieth anniversary of Windows 95!

The last bit of literal and figurative business I’d like to bring up is Patreon. People have been asking me about ways to support the show, and I said I would consider something like Patreon after passing 5,000 subscribers. Well, here we are. I’m under no delusions that I’ll be able to quit my day job and live off this hobby’s revenue. But what I’d like to do is get this channel on the path to self-sufficiency, so Patreon funds will be invested back into video production. Computers and parts cost money, and so does investing in cameras, hardware, lights, and music licenses. My initial focus for the launch will be early video access and occasional bonus content. Maybe I could post work in progress scripts. If you’re one of the people that’s been bugging me about it, you can find it at patreon.com/userlandia.

And now, three installation phases later, we’ve arrived at OS/2’s desktop: the Workplace Shell. The subjective experience on this Model 80 with a 20MHz 386DX, 387 FPU, 16MB of RAM, and an XGA-2 graphics card is a slow one. It takes minutes to bootstrap—which is mostly my fault for loading up on network services—and once you get to the desktop you should let it sit until the disk activity light calms down. It’s not unusable, but you’ll encounter some slowdowns and stopwatches. Mac OS 10.0 Panther feels snappy by comparison.

Disk access times aren't too bad thanks to BlueSCSI, but throughput is held back by the 386 and Tribble card. Some folders take forever to open if they’re full of objects with lots of metadata, like a template folder, because the system has to parse and interpret every object’s properties. If all you asked of this machine was to serve files or host BBS nodes its performance would probably be acceptable. I’ll give it credit for being performant enough to play the video equivalent of canyon.mid: macaw.avi. IBM loved touting OS/2’s multimedia bona-fides, though much like the rest of OS/2 Warp these multimedia features would be happier running on a 486.

But what I lost in speed I gained in reliability. OS/2 had a reputation for stability thanks to preemptive multitasking and memory protection. This foundation wasn’t just for OS/2 native apps—it enabled OS/2’s Windows subsystem to run Windows programs in an isolated process. When your monster Excel spreadsheet inevitably crashes it won't take the rest of OS/2 down with it. If you had lots of RAM you could even run each Windows program in its own instance, so that same Excel crash would only crash Excel and not your other Windows apps. OS/2, DOS, and Windows programs could run side-by-side without stepping on each others’ toes. Yes, there is the synchronous input queue problem, which could cause the user interface to lock up while the underpinnings chugged along. But that was… mostly addressed in Warp 4 and back ported to Warp 3 via fix packs.

If you want to try some native OS/2 apps the Bonus Pack’s got you covered. IBM Works is a light productivity suite featuring a basic word processor, spreadsheet, database, and personal information manager. And while they’re not on the level of Lotus SmartSuite they are fully-featured OS/2 apps that leverage IBM’s System Object Model. You can easily embed an auto-recalculating spreadsheet chart in a word processor document, which was all the rage at the time. When you need a bit of downtime you can switch over to OS/2 Solitaire, which presumably is sixteen more bits better than Windows Solitaire. Too bad those extra bits can’t buy whimsy because Windows still has the edge when it comes to art design.

The Bonus Pack also includes an IBM internet connection kit, which bundles online services and ISP connections along with dial-up networking. But the most interesting piece is IBM WebExplorer, an in-house web browser developed specifically for OS/2. The history of this browser could easily be its own video, but it’s probably best known for letting websites customize the throbber animation. It was also remarkably fast at downloading and rendering webpages for its era. But WebExplorer was doomed to fail. The egalitarian ideal that anyone could make their own web browser was no match for the reality that developers and users would gravitate to one or two popular choices. IBM would abandon WebExplorer in favor of Netscape Navigator for OS/2.

The GUI glue that binds all these programs together is the Workplace Shell, and its user experience is… reasonable. IBM got religion about objects in the 90s and launched the System Object Model, or SOM, with OS/2 version 2. Icons are objects, printers are objects, drives are objects—I had a hard time finding something that wasn’t an object! You can install new object classes like the light table slide objects from the Bonus Pack’s Multimedia Kit which created thumbnail previews for photos and videos. Third parties could leverage SOM too—Stardock’s flagship product is called Object Desktop because it started out on OS/2.

SOM turned out to be a dead end—forcing every user interface problem into an object metaphor was a bit of a stretch. There’s other UI shortcomings of the time, like the absence of a visual task switcher and the directory browser tree windows. But there’s a lot to like about Workplace Shell if your alternatives were the Windows Program and File Managers. It’s very good at remembering the positions of your icons and windows. Everything is right-clickable with robust and well-organized context menus. The system can restore your programs and workspaces on a restart. Shadows and Program Objects behave like aliases on the Mac and Shortcuts on Windows. You can group objects together into a Work Area folder which when double clicked will automatically open all of the included documents and programs. The LaunchPad isn’t as good as a taskbar in Windows or a Dock in NextStep but it’s leagues ahead of Windows 3.1’s nonexistent launcher.

Exploring all these features made me feel, oh, 20% more productive. If Mac OS is a quirky artist and Amiga OS is the auteur filmmaker, then OS/2 is the business causal middle manager. So many elements of this system feel corporate, for better or worse. The UI is called the Workplace Shell. The deletion object is called “the shredder,” and not because the designers were disciples of Oroku Saki. IBM’s marketing and training material all focused around typical business tasks. Games were scarce. Multitasking was pitched as a way to more things done and still have time to break for lunch. Networking was about servers and directories and meetings.

Pitching a system that was work-focused wasn’t unheard of in an era when “workstation” was a thriving market segment. Workstations were serious business, and they ran OSes with higher system requirements for tasks where crashes could prove costly. Domain OS, VMS, Solaris, AIX, IRIX, WinNT, the list goes on. And it’s not like microcomputer users couldn’t use the benefits of preemptive multitasking, protected memory, and networking. Amiga users had preemptive multitasking out of the gate! But doing all three in the limited memory and CPU power of most 1980s PCs was too great of an ask.

But consumer PCs in 1995 finally had enough memory and CPU power to multitask, and average users who tried multitasking in Windows 3.1 were met with constant crashes and freezes. In that context OS/2 Warp could’ve been a contender. Its system requirements weren’t all that different from Windows 95’s, plus it had strong compatibility with existing DOS and Windows programs. It also beat Win95 to release by nine months. An opportunity was there for Warp to overtake Windows. The collapse of IBM’s Personal Systems division sealed its fate despite OS/2 Warp 4.0’s launch in September 1996. How did IBM squander so much money, talent, and technology?

A constant source of trouble for OS/2 was its reputation as a memory hog. In the late 80s and early 90s the US-Japan DRAM tariff war meant average PC users couldn’t afford enough RAM to run OS/2 without performance-degrading virtual memory. By the time computers caught up to OS/2’s requirements IBM’s position in the market had drastically changed. Users who made the commitment to OS/2 found it difficult to find native programs to exploit its power and stability. Big Blue had a hard time convincing developers to stick with OS/2 due to expensive developer tools and slow market growth. Microsoft was dominating with Windows 3.x, and used its power to forge aggressive license agreements to freeze other OSes out of the OEM market. And we’re not even touching IBM’s misguided microkernel misadventures which resulted in the abortive OS/2 Warp for PowerPC.

Yet even if OS/2 was preinstalled on more PCs there were other factors that would doom Warp. IBM had about zero skill in pitching OS/2 to the broader consumer market. Its Get Warped advertising campaign was a tired, out of touch exercise that excited nobody. The Fiesta Bowl sponsorship cost IBM millions of dollars with nothing much to show for it but broadcasters cracking jokes about OS/2. And their plans for a Warp Speed-themed launch event were thwarted when negotiations with Paramount’s legal department broke down. Compare that to Microsoft, who pulled out all the stops to bring the Rolling Stones on board to start up Windows 95’s marketing juggernaut. But let’s say IBM managed to market OS/2 better and captured ten or twenty percent more marketshare. It wouldn’t have survived Lou Gerstner’s radical reorganization of IBM which saw the eventual divestment of its consumer PC business.

Yet despite all the bungling and false starts IBM did manage to produce in Warp a stable, powerful OS that was beloved by the users that found it. These fans—and slow-moving conservative businesses with sunk costs—kept its spirit alive even after IBM pulled the plug. Maybe I would have been one of those fans if I had used it back in the day. Yes, there were several things about it that frustrated me, but overall I found OS/2 Warp 3 to be remarkably charming. Warp 4 fixes a lot of those frustrations while keeping its foundation as a powerful, reliable system. I can see a path where it evolved into something that satisfied both the home and business user in the same way Windows NT became the Windows we begrudgingly tolerate today. That is, if IBM had paid any attention to markets other than office productivity. My gut tells me that they wouldn’t have prioritized development of an API like DirectX. But I’ll set aside that argument for another time.

I suppose I should call it a day for my adventures with OS/2. If you made it this far, I’d like to thank you for sticking with me and indulging this 5,000 subscriber celebration. I’m amazed the installation went as well as it did. There weren’t any real problems aside from that 1GB drive limitation, which wasn’t OS/2’s fault. Installing a different OS on a weekend afternoon is something I liked to do back when I was a computer obsessed teen thumbing through stacks of computer books. If you enjoyed this, please let me know and consider subscribing if you’d like to see more. Thank you for reading, and I’ll catch you next time here in Userlandia.

Video Version of Compaq ProLinea Computers of Significant History Now Available

May 27, 2024 by Dan Vincent in text-only

A video version of the Compaq ProLinea 4/33 Computers of Significant History episode is now available. I made a few edits and updates as well. Enjoy!