The Sol Prototype Project

[falter]

I think I'm almost ready to send this out to PCBWay or such. I always figured the last 1% would be 99% of the work - it's mostly just trying to get the board to resemble the original as closely as possible. It'll never be 100% accurate as the scans are poor quality and I struggled to get both sides of them to align correctly, but it'll be pretty darn close. KiCAD has had quite the learning curve, but I'm surprised at how close I've actually gotten. I could just print these off and do the toner transfer dance... but that means 900+ vias to connect from side to side.

Right now I'm just putting in smoother curves to match the hand-laid ones on the original, trying to align them as best I can with the original artwork. And I'm still finding the odd connection to various pads that I missed, because the gap is so small it's hard to see. Also, it's really hard sometimes to tell how thick a trace should be, because the scanning process produced noise that distorts the tracks and sometimes makes them look wider than they might have been.

I'm thinking of going with PCBWay to produce this, as they are the only fab that has been even willing to entertain my desire to have this etched on some vintage original PCB stock I have (after I do a prototype run), to further ape the look of the original.

Still working on the schematic too - that is going to take ages though as again there's a huge uphill learning curve for a non-EE, non-pro KiCAD user like me. I have to modify every IC symbol to match the schematic to have even a prayer of getting it right. I'm at a point where honestly I'd almost prefer to just print it out large size on my designjet and just do my best to work out the obscured stuff like IC numbers that were scanned badly. What I wouldn't give to find the guy who provided the originals to scan to see if we could get better results 20 years later!

Once this is done I will convert it to 4 layer, and assuming I figure out the missing connections, I will then note those on the schematic and try making them on the inner layers for neatness.

Big project.

 

[Hugo Holden]

I'm wondering about the 4 layer idea. It is better from the technical perspective in that really good ground and power supply connectins can be put in the two middle layers, and it simplifies the design of the top and bottom layers. But it does raise thecost of the board manufacture and middle layer errors in a prototype can be very awkward (as I once found out)

Since you are making a replica board, it would be advantageous to make it just like the original, and then you could use the original board pattern to check it in an image overlay scenario.

 

[wperko]

Hi,

Be sure to include "REPRODUCTION" on the board, also the Rev. too, so people are well assured it's not an original.


.

 

[falter]

I'm wondering about the 4 layer idea. It is better from the technical perspective in that really good ground and power supply connectins can be put in the two middle layers, and it simplifies the design of the top and bottom layers. But it does raise thecost of the board manufacture and middle layer errors in a prototype can be very awkward (as I once found out)

Yes I was reading up and I understand that's kind of normal practice with the middle 2 layers apparently. However if I did that, that would necessarily change the visual appearance of the board on the top and bottom layers, right? My goal is to make it look exactly like the original on the top and bottom side. Which, even Lee Felsenstein couldn't understand my devotion to.

 

[falter]

Hi,

Be sure to include "REPRODUCTION" on the board, also the Rev. too, so people are well assured it's not an original.


.

Yes I was thinking I should probably do that too. That's one of those ethical dilemmas, because I want my own personal one to look exactly like the original, but also don't want people being misled down the line.

 

[1944GPW]

I'm sure I've mentioned it here before but a years ago I saw a photo of a WW1 biplane being restored at the Smithsonian, it was a close-up of the tailplane with the fabric removed. Some of the wooden pieces glued into the structure had 'REPRODUCTION' stencilled on them. I was quite impressed by that.

 

[falter]

I suppose I could put the word Reproduction in a spot along the ground plane on the outside.. that way if someone tried to remove it, it would leave a noticeable 'hole' there..

 

[Eudimorphodon]

My goal is to make it look exactly like the original on the top and bottom side. Which, even Lee Felsenstein couldn't understand my devotion to.

If the goal is to look like the original then why would you want to use a four layer board to make the original "work right" instead of using the masses of jumper wires the original would have required to do the same? I guess scrolling back through the thread we kind of hashed this out already around page 8... but I guess I'm still a little confused. Is the goal here to have a board that looks like the (we have to be honest here, the train wreck of an unfinished project) that was the original published board, but has a working S-100 slot like the final SOL-PC without having to do all the manual patching?

 

[falter]

My main goal is to build one fairly exact replica (accepting compromises where I must) of the original Sol prototype that CHM has, for myself. I'm not really big into retro/recreations *except* in situations where I'm replicating something I can't buy otherwise (ie the TV Typewriter, before I lucked into a real one much later). I also really like the challenge of trying to figure out what materials they used, who made that orange can capacitor, etc. and getting it as close to a perfect copy as I can. It's like one big puzzle. Anyway, that's first goal.

Bonus goals are - I'd like to actually produce a small run of these and maybe give some out to people here, my Youtube viewers and Patrons, etc. And in doing that, going off a suggestion I think was made earlier in this thread, I wouldn't mind releasing a 'corrected' version, that maintains the original look on the top and bottom sides, but has a 'secret' layer (or layers, I guess, since everything is done in twos), that makes all the connections the original designers didn't, so if someone wants to build one with a 'clean' look, like the second prototype board that was featured in the magazine by itself, they can. I also plan to release the KiCAD files and such and if people want to fix them up more, whatever they want to do they can. I doubt there's going to be that much interest given what it is, but I'm sure a dozen or two people might. I'd love to see someone who actually knows electronics find a way to give this thing more RAM and make it operate as 'the SOL computer that might have been'.

 

[wperko]

Hi,

I suggest using large font and put the word "REPRODUCTION" in the area that is normally covered by the Personality Module, or maybe in the blank space under the white label to the left of the Personality Module connector.

Take if from somebody that paid too much for a FAKE original on eBay.

 

[falter]

This version of the Sol motherboard doesn't have a personality module. Totally different board from the one that's in the Sol 20. But I'm definitely considering what you said.

 

[Eudimorphodon]

I'd love to see someone who actually knows electronics find a way to give this thing more RAM and make it operate as 'the SOL computer that might have been'.

I guess this is where I scratch my head again and wonder why the SOL-20 *isn't* "the SOL computer that might have been", given, you know, it's pretty normal for early prototypes to not match the final product.

Or to put it another way, it looks like the original firmware was basically a Potemkin Village, and the memory layout/architecture was really limited and inflexible compared to the later model. This machine won't run any existing SOL software, nor will it run CP/M or any other OS that requires RAM in the bottom page. The memory it does have is restricted to a 512 byte window wedged between ROM and the video memory at the 4K mark. You could certainly stick some more memory above the video window, but you'll have to write an entire library of software from scratch to use it, and you'll also need to write an actually useful monitor or tiny BASIC to substitute for the fake demo firmware, so... I dunno, I guess I'm a little fuzzy on what the "might have been" would have been here. To make this thing useful basically requires patching it to release SOL-PC specs (or something like) anyway. The best I can think of for an alternate universe version would be that you'd use the broken expansion connector to host a daughterboard that would have stuck more RAM and a BASIC ROM into upper memory to basically give you something like a Level I BASIC TRS-80, that's really all you can do without rejiggering the memory map.

 

[falter]

I guess this is where I scratch my head again and wonder why the SOL-20 *isn't* "the SOL computer that might have been", given, you know, it's pretty normal for early prototypes to not match the final product.

That's totally fair. The 'what if' scenario I'm imagining is not Proc Tech actually producing and supporting this thing, but a quick-off-the-mark hobbyist who got the plans early on, (like whoever provided the scans on sol20.org did) and presses ahead with building it, maybe decides not to drop more money on the new design and instead tinkers with the original and gets it to do.. I dunno.. something computerish? Just because they have the skills and can. I have a few machines in my collection like that.. either built from plans or totally homebrewed. No software for them.. or software they ported themselves to it (like BASIC). One machine I have just has a basic monitor ROM and that's it. Built for a purpose - be it programming EPROMs or just messing around. That's interesting enough for my purposes.

Or to put it another way, it looks like the original firmware was basically a Potemkin Village, and the memory layout/architecture was really limited and inflexible compared to the later model. This machine won't run any existing SOL software, nor will it run CP/M or any other OS that requires RAM in the bottom page. The memory it does have is restricted to a 512 byte window wedged between ROM and the video memory at the 4K mark. You could certainly stick some more memory above the video window, but you'll have to write an entire library of software from scratch to use it, and you'll also need to write an actually useful monitor or tiny BASIC to substitute for the fake demo firmware, so... I dunno, I guess I'm a little fuzzy on what the "might have been" would have been here. To make this thing useful basically requires patching it to release SOL-PC specs (or something like) anyway. The best I can think of for an alternate universe version would be that you'd use the broken expansion connector to host a daughterboard that would have stuck more RAM and a BASIC ROM into upper memory to basically give you something like a Level I BASIC TRS-80, that's really all you can do without rejiggering the memory map.

Yes, it's a real tragedy that third EPROM was destroyed. That would have answered a lot of questions. But I suspect you're right, I think they just whipped something up that could put up some demo screens (which we can see the ASCII remnants of in dumps of the surviving EPROMs), or maybe do some basic terminal communications. I think the entire article was kind of a Potemkin Village designed to get attention and drive people towards a more finished product. I did ask Lee for details about the functionality of the prototype, but after this long he didn't really have much memory of it. He did give me a name of the programmer at one point.

 

[Eudimorphodon]

That's totally fair. The 'what if' scenario I'm imagining is not Proc Tech actually producing and supporting this thing, but a quick-off-the-mark hobbyist who got the plans early on, (like whoever provided the scans on sol20.org did) and presses ahead with building it, maybe decides not to drop more money on the new design and instead tinkers with the original and gets it to do.. I dunno.. something computerish? Just because they have the skills and can. I have a few machines in my collection like that.. either built from plans or totally homebrewed. No software for them.. or software they ported themselves to it (like BASIC). One machine I have just has a basic monitor ROM and that's it. Built for a purpose - be it programming EPROMs or just messing around. That's interesting enough for my purposes.

The source code for Le-Chin Wang’s Palo Alto Tiny Basic is out there, you might be able to get that to run on this in place of the original (not useful) ROM code. The extremely limited amount of RAM is still a bummer, but… I suppose if I had one of these boards that’s probably the angle I’d take. I would give up on trying to make the half-implemented expansion slot be anything like a full S100 and just tinker up a ROM/RAM extension card. Assuming the cassette interface can actually be made to work then with Tiny Basic and a simple monitor you’d have something at least as useful as an Apple I, I guess.

 

[Hugo Holden]

It is interesting the idea of building a prototype that is so early in the production that it was still covered in expierimental hookup wires and mods. I have never done that myself, partly because the examples of these early iterations were never available for study. It certainly is much easier to replicate a vintage board when 99% of the bugs were already out of it. Though in the case of the Dazzler, there were two generations of boards and the later ones were improved. So it was better to stick to that.

The SOL-20 itself is so vintage now that the timeline difference between most of the productions boards and that hand modified prototype must be getting smaller and smaller. It would be much less of a headache and more accurate to replicate a production board, than the one with the wire mods. But, I can see the value in trying to recreate that piece of history. Though, pcb's wthout wire mods would be much more sellable to other parties.

In the past I made an exact replica of the original personality modules with the MM5204's, I had to build Martin Eberhard;s programmer. I also used these modules in an S-100 card project to interface them with the bus in the "Three Brains" project:

https://www.worldphaco.com/uploads/THE_SOL_WITH_THREE_BRAINS.pdf

 

[falter]

Yes I am dreading trying to figure out how to program the EPROMs for this thing heh!

This is a bit quixotic I guess. It's in the vein of 'because I can'. If the artwork hadn't been published somewhere then this would never have happened, but it has, so we're on our way.

Hopefully by the end of next year I will have three boards replicated: the Sol Prototype, the original JOLT, and one or two of the main Sphere boards.

I'm surprised no one has made a replica of the production Sol-20 board. I guess maybe because the originals are still fairly available, albeit expensive. I've never seen photos of a bare board nor the original artwork for it.. I'm not sure if that was ever published? Would be an interesting project though.

 

[Hugo Holden]

Yes I am dreading trying to figure out how to program the EPROMs for this thing heh!

This is a bit quixotic I guess. It's in the vein of 'because I can'. If the artwork hadn't been published somewhere then this would never have happened, but it has, so we're on our way.

Hopefully by the end of next year I will have three boards replicated: the Sol Prototype, the original JOLT, and one or two of the main Sphere boards.

I'm surprised no one has made a replica of the production Sol-20 board. I guess maybe because the originals are still fairly available, albeit expensive. I've never seen photos of a bare board nor the original artwork for it.. I'm not sure if that was ever published? Would be an interesting project though.

Kilobaud magazine did a 5204 Eprom programmer in 1977.

Kilobaud Magazine (September 1977) : Free Download, Borrow, and Streaming : Internet Archive

archive.org

Ian Reynolds in the UK made a 5204 programmer in 2024, it was on the UK's Vintage radio website, it looks very good to me. See post #23, apparently inspired by an Australian article:

Programming NSC MM5204Q EPROM - Page 2 - UK Vintage Radio Repair and Restoration Discussion Forum

Programming NSC MM5204Q EPROM Vintage Computers

www.vintage-radio.net

Though it is probably difficult to beat Martin's programmer, he did a really professional job on it.

 

[GlueLogic]

Good luck with this project! After you've recreated a few boards in KiCad, it becomes a lot easier to use.

 

[falter]

I think we are pretty much there now. Here's a 'print' of the layout for the top side. I've printed it out on an actual printer at full size (I really need to get a large format printer - sigh). I think it looks pretty awesome considering who did it, and the level of experience they *did not* have. This was 2 years worth of work.

I'm only having a minor debate about the vias and what are appropriate hole sizes for those. I studied Lee Felsenstein's prototype PCB very carefully and the vias for the 25 pin serial connectors in particular look to have larger holes with a very thin rim around them. I'm hoping the hole sizes I've chosen for all of them are appropriate.

 

[ClassicHasClass]

Nice work!