• Please review our updated Terms and Rules here

Confused about 2101 RAM

falter

falter

Veteran Member
Joined
Jan 22, 2011
Messages
6,573
Location
Vancouver, BC
I'm just trying to get a fix on how much RAM the original Sol prototype had. It physically had 4 2101 256x4 bit chips. I always get confused about how to read that, but I thought 256x4 meant nibbles, and if you had two of them you had 256 bytes. I'm sure I'm getting that wrong. The article says the original machine had 1280 bytes of RAM total. I'm not sure how you get that from 4 2101s?
 
You're not hallucinating. 2 2101s gives you 256 bytes. Maybe there's a separate memory card? 1280 would mean that there's 256+ 1024 bytes of memory, perhaps on the VDM-1?
 
Thanks Chuck. I'm wondering if there's an error in the article. This is what they say:

"On-card memory: 512 bytes PROM (expandable to 2048 bytes), 1280 bytes RAM (expandable to 1560 bytes)"

They do mention it can be expanded to 64 externally, but that'd be via the S100 connector onboard, and they didn't demo that with the prototype because the first design was so messy.

This is a picture of the motherboard - I only see four 2101s there, which implied to me the prototype only had a half kilobyte of RAM total. I don't see any other RAM in the picture, although it's a bit blurry.
 
I found a copy of the magazine schematics here:

Sol Newsletters, Articles and Ads

www.sol20.org

And if you pick through them you'll find that on the "Memory" sheet there are sockets for four more 2101s (separate from the 1k dedicated to video RAM), IC 58-61, on the same bus segment as the PROMs. The video memory was ICs 18-21, 29-32. So in total there would be a minimum of 10 2101s in a SOL. * (The second 256 bytes of CPU ram were optional.)

If you were using the SOL solely as a terminal the 256-512 bytes that fit on the motherboard might be adequate, but obviously most "computer" applications would need a memory card.

* see correction next post
 
Last edited:
... correction to the above: The video memory isn't 2101s, it's 2102s. (It's misprinted in the magazine article's parts list but the schematics are correct.) That's why you only see the four 2101s. (I was a little confused when I looked at the board photo myself...)
 
Doh! I missed the 2101s in the video RAM. Thank you! Okay so that explains the video ram. So they're counting the system RAM and video RAM together as RAM?

This has relevance because I was trying to understand what the code in ROM for the prototype actually did, and thus what the prototype was capable of. I had asked Lee and Bob about it but there's no listing remaining and neither could remember exactly what functionality it enabled. One of the 5204 EPROMs dumped by Al at CHM was dead. If I understand the EPROM info correctly, with the three ROMs there would be about 1.5K ROM onboard, not enough to do anything too fancy with a fair amount taken up by text.
 
Eudimorphodon said:
... correction to the above: The video memory isn't 2101s, it's 2102s. (It's misprinted in the magazine article's parts list but the schematics are correct.) That's why you only see the four 2101s. (I was a little confused when I looked at the board photo myself...)
Click to expand...

Good catch! I wondered what was going on there. Darned misprints.
 
falter said:
Doh! I missed the 2101s in the video RAM. Thank you! Okay so that explains the video ram. So they're counting the system RAM and video RAM together as RAM?
Click to expand...

Yep. Off the top of my head I don't know what the hardware capabilities of the SOL's video card were compared to the VDM-1 it was loosely based on; I have vague memories that the VDM-1 had some funky/excessive hardware features like to-end-of-line and to-end-of-screen blanking that might make it slightly feasible to trade screen real estate for more CPU memory, but of course realistically you wouldn't want to do that.

1.5K of PROM doesn't sound like a *completely* unrealistic amount of space to implement a simple terminal emulator and screen editor program in but, again, pretty useless for a "computer".
 
Chuck(G) said:
Didn't I say something to that effect?

Maybe I should avoid speaking in ellipses.
Click to expand...
You did... I was just not cluing in that the machine had the separate video RAM. I should have known but I was so fixated on where the 1280 bytes was coming from. I've always thought of video ram and system ram as two different things, and usually when I hear RAM as a term, alone, I think of system RAM, not video. But now that I think of it I think Mindset for example may have advertised having 64K of 'RAM', when in fact only half of that was system ram.
 
Chuck(G) said:
Maybe I should avoid speaking in ellipses.
Click to expand...

The confusion is a little understandable given the Popular Electronics article uses confusing terminology (lumping video memory in with CPU memory), employs a pretty suboptimal format (with mistakes) for the parts listing, and the text description here, which is ironically on the same page where they say the memory size is 1280 or 1560 bytes... which should say 1536 if we're playing that game, doesn't mention the video memory at all, only the "up to 512 bytes" of CPU RAM. Frankly it's kind of a mess.

Skimming through the 1977 SOL systems manual it looks like a major difference between the magazine version and the shipping one is they replaced the 512 bytes' worth of 2101s with a second set of 8 2102s so they instead had 1K of RAM in addition to the 1K VRAM on the motherboard. (And they pitched the onboard ROM into a "personality module"?)
 
Yeah I just got way confused there. In addition to forgetting there had to be video RAM somewhere, I was assuming when they said it had 1280 bytes that they meant that was what was installed on the prototype. So even when Chuck pointed out that the video should have 1k of that, I'm like.. ok.. 8 2102s = 1k, but there's 4 2101s so how does that add up to the remaining 280 bytes? Yeah. Thank goodness for internet forums in 2022 where you can ask people who are smarter than you what's going on.

I'm going to revive my Sol Prototype Project thread, as the reason I'm asking these questions is I'm about to embark on building this thing, sort of an alternate history type deal. But yeah, I was really confused there. I'm glad you and Chuck were here to sort me out.
 
... Just editorializing... I'm not going to give Tandy too much credit because they did some kind of dumb things to the TRS-80 in order to save a couple bucks (pretty literally), like deleting the 8th bit of video memory, but comparing the designs broadly it's remarkable how close the TRS-80 at its introductory price of $599 for a 4K system plus monitor and cassette recorder comes to matching a SOL-20 bundle that was $2,129 assembled and tested just a few months earlier. They're almost the same computer(*) except Tandy A: hacked out all the weird window-shading and hardware scrolling bits of the VDM-1, which is a thing that frankly makes a lot of sense if you've gone all in on memory mapped video, B: Got DRAM support mostly for free by using the Z80, C: didn't care about being S100 or CP/M compatible, and D: just generally embraced being cheap. Crazy just how fast things were moving back then. Even if you make the case that a SOL was a Cadillac compared to Tandy's Pinto, well... given how much easier for most people it is to shake Pinto money out of their couch cushions it's easy to see why a lot of these pioneering firms took it hard on the chin at the end of the 70's.

(*Asterisk to "Almost the same computer": the TRS-80 is arguably an even closer rip-off of Polymorphic Systems' machines like the 8813, in that both had ROM in the bottom of the memory map and the video system's semigraphics were exactly like the Poly's VTI video card. (Which was also copied very closely by SSI's VB1B S100 card? Pretty sure the VTI came first but not positive, if I have that backwards then you can say the TRS-80 was ripping off the VB1B.) But they were all still very much in the same capability ballpark.)
 
I think quite literally they borrowed from the SOL. This is what Lee said about it:

"...I should point out that by the fall of 1977, the SOL design was imitated fairly well by the Radio Shack TRS-80 Model I, although it didn't use an S-100 bus. It used dynamic memory for storage and, I think, static memory for the display, but it was the same memory space, which was important, because they were still doing a basic memory access to get something on the screen. John French, who designed that - if I'm remembering his name correctly - credits the SOL as his inspiration, and I'm happy enough with that."
 
falter said:
I'm going to revive my Sol Prototype Project thread, as the reason I'm asking these questions is I'm about to embark on building this thing, sort of an alternate history type deal.
Click to expand...

Are you actually planning to build one based on the magazine PCB layout? I... wish you luck. It very much sounds like that version wasn't ready for prime time.
falter said:
I think quite literally they borrowed from the SOL. This is what Lee said about it:

"...I should point out that by the fall of 1977, the SOL design was imitated fairly well by the Radio Shack TRS-80 Model I, although it didn't use an S-100 bus. It used dynamic memory for storage and, I think, static memory for the display, but it was the same memory space, which was important, because they were still doing a basic memory access to get something on the screen. John French, who designed that - if I'm remembering his name correctly - credits the SOL as his inspiration, and I'm happy enough with that."
Click to expand...

The memory space thing isn't true; the TRS-80 has its video at 3C00h, (15K mark) while the SOL has its at CC00h. (51K. If the manual is to be believed all the ROM and RAM built into a SOL starts in the C-page, and expansion memory is intended to start at 0000h? Something that's not entirely clear to me is if the SOL has some sort of page-flipping mechanism to temporarily map ROM to 0000h for cold startup or if there's a separate circuit that inserts a jump to C000h. The 8080/Z80 starting execution at 0000h is why the TRS-80 and Polymorphic Systems machines have their ROMs there, at the cost of CP/M compatibility.)

Steve Leininger is generally credited as the guy in charge of designing the actual production prototypes for the TRS-80. (I think "John French" is a mixup between "Don French" and "John Roach", two people who were instrumental in getting the TRS-80 to market but not the actual designers.) Don French may well have been inspired by the SOL in persuading Tandy to make a computer, the timeline matches up, but, well, after you see a screenshot like this of a Polymorphic VTI showing off its semigraphics capabilities:

blockgr1.jpg


It's kind of hard to believe that the video system that made it into the final product didn't lift some inspiration from these other sources.

(I did just learn something amusing from looking at the VTI and VB1B manuals, though... all three systems, IE, the two S100 cards and the TRS-80, used different bit ordering for their semigraphics modes. So even though they're practically identical hardware-wise you'd have to rewrite graphics software targeting them if you moved it between systems, otherwise the wrong dots will be set/reset. Sheesh.)
 
Very interesting. I suppose it's nice to think some DNA from the Sol, which I guess could be considered a commercial failure in the end, made it to a much more successful machine.

And yes, I am planning to build a replica of the Sol prototype. I've been warned this one likely won't work out, but I'm prepared to accept a static display. Plus I'm going to document the whole thing for my YouTube channel so who knows, maybe folks in the audience will have ideas.

Something that comes up for me again and again with these old projects is the what ifs or alternate history, which I find really interesting. What if microcomputer development didn't take off for a few more years and Don Lancaster followed through on plans to add a serial computer component to the TVT? What if the Tom Swift terminal had actually been built as intended? Or in this case, what if this Sol project had been just touched up to work, and hobbyists had developed it into a proper computer?

I just like to do stuff that is under the radar or untried. Building an Apple 1 replica would be easier but it's being done to death. I'm looking to stupidly go where others aren't. Even though the original plans were available I don't think anyone actually attempted to build this thing. So why not? Road less traveled and all that.
 
Eudimorphodon said:
Something that's not entirely clear to me is if the SOL has some sort of page-flipping mechanism to temporarily map ROM to 0000h for cold startup or if there's a separate circuit that inserts a jump to C000h.
Click to expand...
Look for PHANTOM on the schematic. It's a chain of four D flip-flops that invert A15 and A14 (0xxx -> Cxxx) for the first three memory reads after reset.

I'm always interested in how this can be done on a Z80/8080, I was looking through old BYTE issues the other day and wrote down references to two or three other methods.
 
falter said:
I just like to do stuff that is under the radar or untried. Building an Apple 1 replica would be easier but it's being done to death. I'm looking to stupidly go where others aren't. Even though the original plans were available I don't think anyone actually attempted to build this thing. So why not? Road less traveled and all that.
Click to expand...

To be clear, I'm by all means not trying to discourage you from tackling the project. Just... expressing the requisite "you know what you're in for, right"? ;)

Something that's kind of gnawing on me is when I was doing some research recently for a project I've got in the works I would swear I tripped across a mention in... it's killing me but I can't remember where, either a "project" or a "product" to convert a VDM-1 video card into a self-contained terminal. It was literally just a line or two, now I wish I could find the reference and dig into it to see if it was:

A: actually about the prototype SOL,
B: Someone else getting the idea of combining a VDM-1 with a really minimal CPU setup to make a software driven dedicated terminal, or
C: Someone building a discrete logic state machine to essentially turn a VDM-1 into an ad-hoc SWTPC CT-64...
 
You must log in or register to reply here.
Back
Top