• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

Reproducing the Lisa 2 Card Stack: The CPU Card

warmech

Well-known member
On to unfortunately bigger things! Next up is the CPU card and, boy oh boy, is this going to suck. First and foremost, a huge thank you to user jamesdenton over at LisaList for providing me with a busted CPU card, as he's allowed me to depopulate it and scan it for reference for this task. I need to find someone/somewhere with a 1200dpi scanner that can manage this board, but it will be nice to have a high resolution image set to refer to.

The schematics have been replicated and I'll be starting in on the board layout once I have the scans in hand; all said and done, this was about a weeks worth of work in my free time so I'm happy it went as fast as it did. I got this done in about the same time it took me to do the RAM card's schematics, so I'm going to interpret that as having at least advanced my skills a little bit, lol. I'm not looking forward to laying down all the traces, as the number of them is substantially higher than the RAM card. It may be a couple of weeks before this is ready, but I'm looking forward to it being done.

Here's where I am at the moment - this should look better in a few days, lol...

Screen Shot 2022-08-08 at 1.24.21 AM.png
 

warmech

Well-known member
Okay, small update... well, less an update and more just providing some reference images. They're a bit large, but they're a fraction the size of their TIFF parents. These are unfortunately only 400dpi; I have access to another large format scanner that can go up to 12800dpi (lol, each side of the CPU card would have taken ~8 hours to complete), but no matter what the scans kept coming out fuzzy. I'm looking for something that will do at least 600, but would prefer 1200 or greater. When this all goes to GitHub, better scans will be available there.
 

Attachments

  • cpu_front_280mmw_x_233mmh_(244mm_card_edge).jpg
    cpu_front_280mmw_x_233mmh_(244mm_card_edge).jpg
    6.3 MB · Views: 48
  • cpu_rear_280mmw_x_233mmh_(244mm_card_edge).jpg
    cpu_rear_280mmw_x_233mmh_(244mm_card_edge).jpg
    5.7 MB · Views: 49

mg.man

Well-known member
I'm looking for something that will do at least 600, but would prefer 1200 or greater.
I have an A3 / 5000dpi (optical) flat bed. Will be happy to scan, but you'd need to get the board over to me here in the UK. 🤔
 

warmech

Well-known member
Going to have to take a few days of after-hours progress off on this to rapidly install a drainage system for my yard before round two of heavy rain hits the DFW region this coming weekend. Spent yesterday digging a trench in my back yard to prevent my shop from flooding, so we're just going to fix that now, lol.

In the mean time, here's a quick progress pic to keep y'all posted. I'd say this is around 40% done at the moment:

Screen Shot 2022-08-23 at 10.29.33 AM.png

Also, what the heck is this, Apple?

Screen Shot 2022-08-23 at 10.35.17 AM.png

There are several right-angle traces on this thing, which I always thought was a big no-no due to EMI/cross-talk. Oh well; their engineers probably know better than myself, an untrained hobbyist with no formal education in any of this, lol.

I have an A3 / 5000dpi (optical) flat bed. Will be happy to scan, but you'd need to get the board over to me here in the UK. 🤔

Lol, that might cost a bit! I actually got a lead on an A3 scanner about 100 miles away for dirt cheap, so I may see about picking that up this weekend. Thank you, though! :)
 

mg.man

Well-known member
that might cost a bit!
I don't think so... probably $25 or so I'd guess. Since the board has been stripped of components, it probably doesn't weigh that much. How much is 200 miles worth of fuel over there these days? Im sure it's still a fraction of what it costs here!! 😭
 

warmech

Well-known member
Well, almost two months later and here we are:

front.png
back.png

Not visible are the two inner layers form +5V and ground (well, DGND). This is effectively 100% complete, but I'm trying to get one last piece of cosmetic flair to work before calling it done. The software I'm working with (EasyEDA) does not make it easy to render the top and bottom substrates translucent by way of removing the masking/paste layers, so I'm trying to figure that out presently. As you can see with the front side, only the "1" is visible in the layer indicator, when it should look like this:

Screen Shot 2022-09-19 at 12.19.26 AM.png

Oh well, I'll get there eventually. Either way, my goal of finishing by the end of the weekend has been met, so I'm happy at the moment. In all, this card has over 1,800 traces that were reproduced by hand - no autorouting, just manually tracking each net by hand. Some license was taken when arranging the final traces but they are, in both count and connectivity, a 1:1 match to the original card; that is to say, this card is an electrically equivalent reproduction of the original PCB with a couple of minor updates. Which brings me to...

Updates! After speaking with AlexTheCat123 over at LisaList, I decided to keep this as close to the original hardware as possible save for a couple quality of life improvements:
  • Connections requiring factory bodge wires have been corrected; the original broken traces are absent from the card and their bodged corrections have been routed as proper traces.
  • The +5V jumper from the CPU section of the schematics has been made more apparent, with a J1 designation of the front of the card.
I didn't keep close track of my time but I'd estimate that there's between 80-100 hours in this thing by now, all of which I would say are worth it. This is a cool piece of Apple history that is now preserved at the informational level and, for that, I am very excited. For now, though, I'm going to set this thing aside for a couple of days and clear my head. Mid-week I'll come back with fresh eyes and take another very close look at everything to check for issues; just because there are no DRC errors doesn't mean it's ready to set sail.

If anyone has experience with EasyEDA and can point me in the right direction with respect to rendering a section of a top or bottom layer translucent, hit me up!

Will,
 

warmech

Well-known member
This is really nice work. Well done.
Thanks!

Quick update for anyone following. Version 1.0 of the PCB is going to get a second set of eyes doing one last sweep and then it's off to fabrication. With the help of the person working on the I/O card, we (they, really) were able to figure out how the layer indicator transparency needed to work, so that's all settled out. If all turns out well and no further revisions are required, this might very well be a done deal by the end of next weekend! If so, expect a link here to GitHub with high-res scans of RAM and CPU cards; Gerber files; exports of the EasyEDA projects (if anyone wants to tool around with the designs); and documentation for the cards.

Screen Shot 2022-09-22 at 2.00.10 PM.pngScreen Shot 2022-09-22 at 2.00.34 PM.png
 

warmech

Well-known member
They have arrived...

IMG_4250.jpg
IMG_4251.jpg

...and it's a perfect fit in the card cage. Should have one assembled by this weekend; wish me luck - this thing has over a thousand pads to solder, lol.

@mg.man Eventually, yes! I'm working with AlexTheCat123 over at LisaList on some collective first revision stuff (not just this card); once that's done, and we've either kinda ruled out any egregious faults (read: mistakes) or determined what major fixes are needed, we'll produce a second revision if necessary and go from there. I uhh... only have the one Lisa 2 and it's currently dead pending the fruits of this and Alex's work, so once the bugs are stomped out, I'll probably want to have Alex and one or two other folks test this out on a couple of different card stacks/ROM revisions/etc. to test for any compatibility issues. I'll keep you posted!
 

mg.man

Well-known member
Eventually, yes! I'm working with AlexTheCat123 over at LisaList
Lookin' nice!! FYI, I've ordered 5 x of Alex's 2/5 motherboards. I believe I mentioned, I have a small collection of Lisa parts - enough to put together one fully working 2/10 so far. I do have a 2nd 2/10 motherboard, but that motherboard has some serious CPU | I/O card edge connector rot - hence ordering some of Alex's 'boards. I have also now realised the 2/10 motherboard (and I/O card for that matter!) won't work in a 2/5 -- the other Lisa 'parts' machines are one 2/10 and another 2/5. So... I need one, likely two CPU cards -- and will be happy to be a guinea pig early tester to see if I can get another Lisa [2/5 more than likely] going... :)
 

CC_333

Well-known member
This is fun! If I can't have an original Lisa all at once (too expensive!), maybe I can build a modern reproduction one piece at a time instead (probably still expensive, but likely not the stupid money originals sell for).

Now to make a complete Lisa clone, all someone has to do is reproduce the case and inner chassis, the keyboard, mouse, and power supply.

c
 

warmech

Well-known member
This is fun! If I can't have an original Lisa all at once (too expensive!), maybe I can build a modern reproduction one piece at a time instead (probably still expensive, but likely not the stupid money originals sell for).

Now to make a complete Lisa clone, all someone has to do is reproduce the case and inner chassis, the keyboard, mouse, and power supply.

c
This is kinda where I saw this going originally - not just a means of replacing dead hardware, but of also building a "new" Lisa altogether if one were inclined. The form factor really binds you to using a pre-existing Lisa chassis, so after all this is said and done I'd like to try to integrate a lot of the Lisa hardware into a smaller, more unified boardset. Using surface mount components and SIMM-based RAM, the CPU, I/O, and RAM cards could probably be integrated into one PCB fairly easily (I'll let auto-routing handle that beast, lol). As the number of expansion options was pretty small, the major expansion cards could also probably be integrated. SCSI control and additional parallel interfaces were kind of the big ones; if anyone feels like sending me or Alex one of the ungodly rare/expensive DAC or NIC cards, those could also, presumably, be added. The video output is a weird TTL format whose horizontal frequency doesn't quite match up to a run-of-the-mill TTL display, so it would need to either be modified (a challenge I'm not up to meeting, lol) or you'd need an RGB2HDMI type device either integrated or running external to the system.

The keyboard/mouse controls, however... that's where things get stupid. The COP421 has been dumped and decompiled; heck, I have a copy of the assembly code for it sitting in my GitHub. If you were going to reproduce the hardware in a meaningful manner, though, that MCU has to go. It would need to be replaced by a modern offering, like a PIC or something, but that would require the code be ported and I have no idea who (if anyone) would be capable of that. That stupid IC is the only irreplaceable thing in the whole card stack; it's technically reproduceable, but requires you to have a maskless COP400-series MCU that is attached to an EPROM with the COP421 ROM dump on it. Again, not impossible, but those maskless MCUs aren't going to get easier to find over time. I've been tempted to put a bounty on this project, but I honestly don't know that it would ever be met. That's a pretty niche piece of hardware for a pretty niche corner of a pretty niche hobby. Never say never, though...
 

stepleton

Well-known member
one of the ungodly rare/expensive DAC or NIC cards
If the NIC is the unusual AppleNet card, you could start from my schematics to replicate it. That said, it's got some unusual parts. Also, I haven't been successful in setting up an AppleNet, and I'm not sure anyone else has.

If there were eighteen of me, then the seventeenth would work on the COP421 replacement. It would extend the protocol by which it talks to the rest of the Lisa by including a new command and response for dates beyond 1995. It would probably also include a CR2032-backed RTC.
 
The keyboard/mouse controls, however... that's where things get stupid. The COP421 has been dumped and decompiled; heck, I have a copy of the assembly code for it sitting in my GitHub. If you were going to reproduce the hardware in a meaningful manner, though, that MCU has to go. It would need to be replaced by a modern offering, like a PIC or something, but that would require the code be ported and I have no idea who (if anyone) would be capable of that.
I would expect the code to be extremely dependent on code executing timing of the COP421 (software-timed loops etc. – didn't check the code). Nevertheless, a COP421 emulator on a fast modern microcontroller might be an option?

(Edit) There's also a VHDL implementation of the COP400 at opencores.
 
Last edited:

warmech

Well-known member
Very minor update time!

Not long after this came in I got it populated and started testing. Ran into some issues initially and started diving into them, but ended up hitting some roadblocks due to a lack of time and tools, as well as a still in-progress understanding of Lisa hardware that just wasn't where it needed to be to probe much farther. Alex has been helping me understand the Lisa hardware more and more, though, so I'm picking up more knowledge as I go. He agreed to take a look, as things were getting stupid busy at work for me, so off it went to him for the moment. One of the things we immediately realized was a royal pain when it came to diagnosing Lisa hardware in-depth was the inability to have the card cage outside of the Lisa when testing. So, I whipped these up and am sending them of to have them fabbed:

Screen Shot 2022-11-24 at 2.26.42 AM.png

Screen Shot 2022-11-24 at 2.09.27 PM.png

As their names imply, these are extension harness PCBs that use IDE cables to extend the motherboard outside of the chassis and the CPU and I/O cards away from the motherboard for more extensive testing. I need to reduce the number of "perforations", but they are meant to just snap apart and wire up in order to reduce the number of fabbed boards and save on cost. These were a quick and dirty solution, but they should make diagnosing Lisa hardware issues a bit easier in the future.
 

stepleton

Well-known member
I'd love an update on how it works --- adding more length to the bus always runs some risk of introducing signal issues. But being able to have the card cage outside of the computer would be extremely convenient!
 

warmech

Well-known member
I'd love an update on how it works --- adding more length to the bus always runs some risk of introducing signal issues. But being able to have the card cage outside of the computer would be extremely convenient!
That's my biggest concern. The card cage extender should be fine; it's just floppy/keyboard/video/power, and those already have ribbon cables or wires comprising the internal harness anyway, so I'm not terribly worried. The bus, though... that's the tricky one. If it doesn't cooperate well enough, I may see about designing a right angle adapter that just folds the cards over with minimal addition of length. You'd have to remove the fiberglass cage for it to work, but I don't see why it wouldn't.

Actually, I'm, just going to make that now; wouldn't be hard at all and would be a decent backup if these don't work. I'd really prefer not to have to reorder just a couple boards; shipping ain't cheap, unfortunately.
 
Top