In the course of a career, you may run up against projects that get cancelled, especially those that are interesting, but deemed unprofitable in the eyes of the corporate overlords. Most people would move, but [Ron Avitzur] just couldn’t let it go.
In 1993, in the midst of the transition to PowerPC, [Avitzur]’s employer let him go as the project they were contracted to perform for Apple was canceled. He had been working on a graphing calculator to show off the capabilities of the new system. Finding his badge still allowed him access to the building, he “just kept showing up.”
[Avitzur] continued working until Apple Facilities caught onto his use of an abandoned office with another former contractor, [Greg Robbins], and their badges were removed from the system. Not the type to give up, they tailgated other engineers into the building to a different empty office to continue their work. (If you’ve read Kevin Mitnick‘s Ghost in the Wires, you’ll remember this is one of the most effective ways to gain unauthorized access to a building.)
We’ll let [Avitzur] tell you the rest, but suffice it to say, this story has a number of twists and turns to it. We suspect it certainly isn’t the typical way a piece of software gets included on the device from the factory.
Even though Apple isn’t known for making the most pro-consumer devices ever (at least not since the Apple II), the trope that Apples aren’t upgradable, customizable, or otherwise hackable doesn’t really hold much weight. It does take more work to modify them or change how Apple wants them to behave, but it’s not completely impossible. Take this example of a ’94 Apple PowerBook which runs macOS Moneterey thanks largely to new internals from a 2015 MacBook Pro.
[Billy] originally intended for a Raspberry Pi to go inside this old PowerBook, but at the time, prices for ARM single-board computer (SBC) were astronomical. For around the same price as the Pi was at the time, he was able to pick up a retina display from an iPad and the internals from a broken MacBook Pro to outfit this retro case. There’s also a Teensy installed to get the trackpad working and a driver board for the display from Adafruit, and a number of case mods were needed to get everything to fit including the screen which was slightly larger than the original 9.5″ display the laptop would have shipped with.
This project took both inspiration and some of the actual code needed to get everything working from another project we featured a while ago where a Mac Mini was installed inside of a PowerBook case from 1993. Unlike projects that use smaller SBCs for retrocomputing, these builds are notable because the hardware on the inside makes them usable as daily driver computers even today, and might even be an upgrade if you’re using the internals from a MacBook Pro that would have originally had a butterfly keyboard.
Apple is fairly notorious for building devices that are difficult to repair, but with the right tools it’s often not completely impossible to circumvent some of their barriers. As they say, every lock has a key. [dosdude1] has wanted a specific M1 development board for a while now and has been slowly piecing together everything he needs to cobble one together, and finally got this unit running despite many roadblocks put in his way by Apple.
The development kit is a Developer Transition Kit or “DTK” meant for developers during Apple’s transition from Intel chips to their own in-house ARM-based M1 platform. This particular version is in a Mac Mini form factor but it has a few hurdles to clear before it powers on. First, the board was cut in a critical location that shorted out many of the PCB layers, so this had to be carefully filed down to remove the shorts. It was also missing a few tiny surface mount components and a NAND chip, but these were scavenged from other scrapped parts and assembled into a fully working machine.
There are a number of other non-physical problems to solve here as well, too. Apple coded their NAND chips to work with specific WiFi modules so if these aren’t programmed to work together the computer will get stuck in a boot loop. But with that and a few other details out of the way [dosdude1] finally has his DTK up and running in a 2018 Mac Mini chassis, right down to the working power LEDs. We’ve seen all kinds of PCB damage before (although not often quite this intricate) and even PCBs repaired that were snapped in half.
The Apple II was launched in 1977, a full 47 years ago. The Apple IIe came out six years later, with a higher level of integration and a raft of new useful features. Apple eventually ended production of the whole Apple II line in 1993, but that wasn’t the end. People like [James Lewis] are still riffing on the platform to this day. Even better, he came to Supercon 2023 to tell us all about his efforts!
[James]’s talk covers the construction of the Mega IIe, a portable machine of his own design. As the name suggests, the project was based on the Mega II chip, an ASIC for which he had little documentation. He wasn’t about to let a little detail like that stop him, though.
The journey of building the Mega IIe wasn’t supposed to be long or arduous; the initial plan was to “just wire this chip up” as [James] puts it. Things are rarely so simple, but he persevered nonetheless—and learned all about the Apple II architecture along the way.
Simply Mega
For the unfamiliar, the Mega II contained most of Apple IIe, but condensed down to fit on to one single 84-pin ASIC. It first appeared in 1986, and was used in the Apple IIGS. The Mega II chip doesn’t contain a CPU, but does contain a lot of the supporting hardware that makes the Apple IIe what it is. The IOU, MMU, video and keyboard ROMs, and keyboard and mouse controller are all there.
[James] began his work with breadboards and plenty of flywires.In theory, you could wire it up with a CPU, some RAM, and a ROM, and you’d have an Apple II. That sounded about right to [James], but as he would soon find out, the reality is more complicated. Had real documentation been available, he might have learned this sooner, but he was flying blind and stuck at it anyway.
[James]’s talk starts by dispelling some myths. The first is that the Mega II is a single-chip Apple II. His own build makes it obvious that’s not the case. Beyond that, he also notes that the Apple IIGS didn’t use the Mega II for backwards compatibility. It was instead used as a memory controller and for working with classic Apple II video modes and the expansion slots. He also states that the “Gemini” chip from the Macintosh LC Apple II card is not a Mega II, based on significant differences in registers and general design. He wanted to check on a rumor, too, that the chip was never finished, hence why it was just used as a basic controller in the Apple IIGS. To figure that out, he wanted to prove that you could build a functional Apple II using the Mega II chip salvaged from the IIGS.
Getting the boot sequence working at all was a challenge.
Early work went well. After spinning up a basic board with the Mega II and the necessary hardware, [James] was able to get a text video mode working. It spat out a lot of gibberish, but it was evidence that something was going on. The string “Apple ][” even appeared in the output!
It wasn’t smooth sailing from there, though. Months of debugging ensued, with a logic analyzer showing the “computer” he’d built was making seemingly random jumps all over the place rather than going through a normal boot process. He suspected it was because the chip wasn’t finished, but that wasn’t the case. Eventually, a friend suggested he check if one of the data bus lines was stuck. It turned out a stuck wire fragment on a solder pad was causing all of his problems. With that fixed, he made a huge leap forward. He saw the message appear—”KERNEL OK”. Success!
However, he still wasn’t at a prompt. He was stuck in a test routine, and didn’t have a fully fledged computer. Nonetheless, he knew this meant he was on the right path. If the test routine was behaving appropriately, it suggested the Mega II really did have most of an Apple II living inside it.
From there, the talk goes through the arduous work required to get this thing fully operational. He faced crash after crash as he tried to get a full boot to happen. Once he got that far, he had to try and go from a rat’s nest of wires and breadboards to something that wouldn’t fall apart every time he breathed on it. His final design has some neat features—like the way he built his own motherboard with a connector that let him plug his logic analyzer right in with a single connector.
Revision 2 used a multi-board design with headers used as backplane connectors.
By revision 2, he had the thing built up on a bunch of PCBs with header connectors, and it was even booting from floppy disks. The amount of work it took to get even this far was phenomenal. [James] didn’t just have to whip up a motherboard, he had to do all kinds of work to give the thing a working keyboard interface and the ability to read floppy disks. By this point, he’d realized that the supposed single chip solution that was the Mega II anything but. Beyond RAM, ROM, and a CPU, it also needed another custom ASIC called SlotMaker, and a keyboard controller. This chip is for handling expansion slots, but it also handles one important signal for the keyboard that makes it essential to the computer’s proper functioning. Frustrating stuff indeed. Other work involved producing video output over VGA and creating a working sound board.
The revision 3 board was neat and compact.
His third revision involved consolidating his multi-PCB solution into a working computer on a single PCB. A few mistakes cropped up in the board design, but it wasn’t enough to stop [James]. With a little creative bodging, he got the thing operational. He’d managed to build a working computer based around the Mega II ASIC.
He may not have been the first to build a Mega II into a working computer; [James] admits that the prototype of the Tiger Learning Computer from Tiger Electronics does serve as an existing example that the Mega II was usable as a fully-fledged machine. But he did manage to do it without any support or documentation from Apple at all. That’s no mean feat.
It’ll even run Prince of Persia!
Watching the talk, it’s easy to understand why the project took [James] several years to complete. It’s always neat to see someone set themself a difficult goal and follow through to completion. It’s also rad that [James] was able to teach us so much more about the Mega II that we never knew before. That’s worthy of celebration!
Conectar
