Reproducing the Lisa 2 Card Stack: The CPU Card

[warmech]

@stepleton Instead of IDE cables, how do you think these may fair as far as avoiding interference? The CPU riser is pretty simple, but the I/O riser inverts the connector to allow the card to have its component side face up for easier diagnostic access. I know that's a lot of traces to invert that connector, but do you think it might fare better than 120 wires running a foot or so? The power rails on both are .254mm as well; think that might cause some issues with draw, or would that be more negligible than I'm thinking?

Edit: Earlier I said these would have to be used with the fiberglass cage removed but, after some testing, these should drop into the motherboard's CPU and I/O connectors with the cage attached. This will allow them to maintain a good deal of stability by using the cage's rails to hold them upright. There's enough space at the back of the cage for both of these to seat into the motherboard and connect both cards (with about a +/- 1mm shortfall for the CPU card - not enough to matter, thankfully). The amount of space at the back of the cage is oddly close to perfect for these, which makes me wonder if Apple had a similar setup for diagnosing/repairing Lisa cards. I'll probably add some space between the topmost trace on the I/O riser and the edge of the card; it looks a bit close at the moment.

Edit 2: Because I overthink everything, the I/O card can obviously be worked on in the cage be default without need for anything like this. If you wanted to work on the CPU card, however, you'd have to fold the I/O card over in order to use the CPU riser.

 

[stepleton]

Hmm... I'm no electrical engineer, but I've occasionally seen a few out in the wild.

My instinct is that bus extenders ought to strive to be (a) as short as possible and (b) as straight as possible. Short because you want to minimise extra signal propagation time, increased resistance, and other electrical effects (wires have intrinsic inductance; wires adjacent to one another are capacitors, and so on). Straight because you want your extension to minimise the differences between the signal propagation characteristics of each wire --- in the plug-to-socket adaptor boards you have up above, the traces on the outside of the "half racetrack" are nearly twice as long as those on the inside.

In addition to this, I have the folk instinct to avoid designs that look like loops, which I understand to be a good way to make certain kinds of antennas. Consider what NFC antennas look like: not too dissimilar to your racetracks, although of course an NFC antenna is a single coiled conductor, not a bunch of traces in parallel.

Once again I have to repeat that I'm not an EE and so am only making guesses here. I don't really have any idea what I'm talking about. I would be interested to know what a real engineer would have to say here.

But without that input, if I were browsing in a shop and I saw a design like this one next to another that uses short IDE cables, I'd probably go for the IDE cables.

(Is there a way to avoid the racetrack and flip the straight-line extender board over? You could just put that socket connector on the other side, maybe?)

 

[warmech]

That all makes complete sense, and I can't believe I didn't catch the resemblance to an NFC antenna. The only main reason I'd inverted the pins was so the PCB could be viewed component side up but, after thinking about it some more, that's really kind of unnecessary. If one needs to work on the I/O card, they can in the stock config; this is really just intended for the CPU card. Bearing that in mind, I just set it back to the way I originally had it setup and now the I/O card faces down as normal.

 

[warmech]

Well, it may turn out that the riser cards are unnecessary for what I was making them for. I'll still have a run made for future cases, but I think I've found the issue, and completely by mistake at that. I was reviewing some stuff to get ready to upload to GitHub and noticed something... off.


Yep.



Somehow I missed U5A's power and ground connections. I checked one of the other blank PCBs I have and, sure enough, they're just floating. Alex is going to take another look in a couple of days when he has some free time, but, damn if that isn't frustrating.

 

[warmech]

Okay! After getting some readme stuff written up, here be the Lisa hardware repo on my end. Alex's is here, and he's got all the other PCBs. I need to wrap up the RAM cards and get them posted, which I may try to do this weekend over the holiday. The RAM additions will include three separate PCBs: the stock, OEM card with 512K of parity RAM over 72 4164 DRAM ICs; an LRambo-modded OEM card that's equivalent to a 2MB AST RAMStack card; and a 2MB parity-based SIMM version.

Once those are up and tested, there are a couple minor boards left (the LisaLite and the video board) and the Lisa hardware can be considered complete. After the LOS source release at the end of January by the CHM, the Lisa will have effectively become an open-source machine. I cannot express how happy I am that that is now the case. Alex and I have one more trick up our sleeves, though... more on that later.

 

[stepleton]

one more trick up our sleeves

Exciting --- I'll put some hashes here of things I think this could be.

9242afb411d5030c01321d037d62b3f4222f39132b79c70a21adfdc79bacdf9a
d20c7b6af10c352da144389467d62f76dbd90672e2bf47576e50622738fee06c
566c83258ba324ce51d9ca6e2729756a6e9a8da9c4bc30ee0dc3e9d9e905f80b

After the LOS source release at the end of January by the CHM, the Lisa will have effectively become an open-source machine.

Worth saying though that LOS probably won't be released open-source if other CHM releases are anything to go by. Here's the CHM's license for PostScript, for example. You can see the code, but there are limitations to what you can do with it.

Congrats on completing so much!

 

[warmech]

Exciting --- I'll put some hashes here of things I think this could be.

9242afb411d5030c01321d037d62b3f4222f39132b79c70a21adfdc79bacdf9a
d20c7b6af10c352da144389467d62f76dbd90672e2bf47576e50622738fee06c
566c83258ba324ce51d9ca6e2729756a6e9a8da9c4bc30ee0dc3e9d9e905f80b

Worth saying though that LOS probably won't be released open-source if other CHM releases are anything to go by. Here's the CHM's license for PostScript, for example. You can see the code, but there are limitations to what you can do with it.

Congrats on completing so much!

Thank you! Alex and I are pretty darn happy we got to this point.

As far as the LOS source code goes, even the PostScript license isn't (to me, at least) terrible; I'm more excited for being able to just review the code and see what insight folks can get out of it. Even if the code itself can't be moved around or modified for distribution, I read that license as at least allowing for the creation of patches or instructions to produce patches that can (so long as you avoid distributing original code) be distributed. I would assume something like Galoob v. Nintendo would cover that, but I'm not a lawyer and my opinion is worth about as much as a wooden nickel.

Last, those hashes - they don't match up with anything I've got my hands on... any chance I could get a hint via DM what you're thinking?

Edit: For the PS source license, I would understand a "derivative work" to be a copy of the patched OS and not the patch itself and, therefore, the license would allow for patches to be distributed.

 

[cheesestraws]

Here's the CHM's license for PostScript, for example. You can see the code, but there are limitations to what you can do with it.

I find the phrase 'trade secrets of the computer history museum' unreasonably amusing

 

[al kossow]

Worth saying though that LOS probably won't be released open-source

It won't, no matter what a random reporter claimed from quoting me out
of context from a message on a private mailing list five years ago.
Getting Apple to release the code means we abide by their rules.

"I find the phrase 'trade secrets of the computer history museum' unreasonably amusing"

Guess you'd laugh your ass off at the concept of a "dark archive" then.

CHM has things we can't release. We then spend years trying to get companies to let us do it.
Ha Ha Ha

 

[cheesestraws]

Guess you'd laugh your ass off at the concept of a "dark archive" then.

When I said it was the phrase that amused me, I meant precisely the phrase. The words. No more, no less. I come from a family of librarians and historians, statistically speaking.

Every time you post on here you look down your nose at people and are obvioously dismissive. I was quite excited about the release of Lisa-related source code. Your way of speaking about it has totally killed any enthusiasim I had on the subject. If this is how your museum does outreach, it's unorthodox, I'll give you that. (Not, of course, that I have anything unique to offer; this is hardly a loss to the Lisa community)

 

[Tarkie]

@warmech do you have a BOM for the CPU card as mine are arriving today!

 

[warmech]

@warmech do you have a BOM for the CPU card as mine are arriving today!

I'm actually in the middle of assembling a BOM for someone else, so I'll send it over once complete (and post it to the GitHub page). Some of the ICs aren't going to be as easy to find as going to Mouser or DigiKey; 74ALS1002s can be found on eBay and some places that carry NOS ICs, but they've been EOL for ages. There isn't really an equivalent IC for this, as it combines a 74ALS02 (a very fast quad dual-input NOR gate) with a set of buffers on the outputs. I've been meaning to test some kind of replacement adapter board that potentially uses widely available 74HC02 and 07 ICs (the speed should be only a couple of nanoseconds slower) to make up for these, but I'm just now catching back up on things after work and haven't had a chance to test this.

 

[warmech]

Et voici!

Almost everything can be found at Mouser: Mouser Project
For the four ICs not available there, there's DigiKey: DigiKey List

Unfortunately, there are a few things that either aren't generally available anymore, or I couldn't pin down. In the oscillator circuit, Q1 and Q2 are PN4258 BJTs that are EOL and not stocked anywhere mainline; the same goes for the uber-specific crystal at 20.37504MHz. L1's inductance I do not know, nor do I have a way to measure for, so it's a bit of a mystery to me; mine was salvaged off a dead board. Those should be unnecessary, however, as you can either get a programmable oscillator set for the Lisa's 20.37504MHz frequency or use the 20MHz oscillator in the Mouser project link and drop either into the four pins situated just above U1E, which should allow you to ignore the entire oscillator section's worth of parts altogether. Allegedly, a 20MHz oscillator should work fine, though I have yet to test that myself; the video may need a little tweaking, but that's about the only thing I can think of that needs adjustment.

Lastly, the two 74ALS1002s... Again, you should be able to find these on eBay and the like, but they've been out of production for years. Like I said, I need to test out whether you can just replace them with a combination of more generally available parts. I had to check, but the 74ALS02 (the non-buffered version of the 1002) and the 74ALS1034 (the ALS version of the non-inverting hex buffer) are still manufactured, so it may be worth a look on my part to make a quick and dirty adapter for these.

Hope these serve you well!

PS - The board silkscreen says to use 74S109s (both mine and the originals), but both of my Lisas have 74F109s for all seven ICs; this change is reflected in the Mouser link.

Also, congratulations on building a Lisa! There have been a few folks who've done so already, so welcome to a fairly small but very awesome club - owners of the first newly manufactured Lisas in 30+ years. DosFox on Mastodon and TinkerDifferent has already built one to completion so, if you run into snags somewhere along the line, you might try pinging them as well.

 

[Tarkie]

Epic! Many thanks

 

[calvinchu1999]

@warmech Thank you for sharing. Could you please provide the photo image of the finishing board for a reference?