• Please review our updated Terms and Rules here

Board scans wanted for IBM 5160 cloning project

I wonder what the reliability of the final product was in the field?
It's very hard to tell, many of these computers (and many, many more) were destroyed in the 90 by gold diggers. There were a lot of gold and platinum in soviet electronics, so they just ripped the components and dumped the rest. I have another Delta-128, which I got to debug the first one. It's spider was alive, but one of the RAM chips was dead. Dead RAM is quite common, sort of like MT RAM in Commodore 64.

As for wafer yield - I have a little story that my professor in the university told us at one of the lectures. I don't know if it's completely true, but from what I saw later and heard elsewhere it's probably 80% true.
The story goes back to when he - that professor - was a student doing his professional internship at one of the microelectronics plants in the 1970s. He was assigned as an assistant operator in the diffusion department, where silicon wafers underwent the diffusion process. On his first day, dressed appropriately in protective gear, he entered the room. Near the diffusion ovens, there sat the operator - an elderly grandmother - eating dumplings at the table. Seriously. In a clean room. Eating dumplings. Seeing his surprised and slightly bewildered expression (he remembered lectures about air cleaning and containment in production), she reassured him that everything was fine and that after she finished her meal, she would brief him on the task at hand. In the meantime, he could sit on a chair and wait. He sat down, looking around. Not far away was an oven processing the wafers. The process involved a large quartz tube heated to over 1000 degrees, into which a quartz boat with wafers was inserted, kept inside for a set time, and then taken out by the operator. As the boat emerged automatically, the operator - the grandmother - continued sitting, still having her lunch! He started to worry because the boat was heavy and needed to be handled, but seeing her reaction, he remained seated. Eventually, as the boat exited the oven, gravity took over and tipped it over with the wafers, causing them to fall and shatter on the floor. With a horrified expression, he turned to the grandmother, who waved her hand and said, "Oh, forget it, it's not your fault! Here, have some dumplings instead!" After laughter from the audience, he explained that we shouldn't be surprised that during the heyday of Soviet electronics, the yield of functional products did not exceed 15%.
 
Well, after all possible checkups (that I can do without the actual board in my hands), the boards are ordered. Four different boards are coming - a motherboard replica, a CGA card replica, an Adlib replica, and an XTIDE, all with white solder mask. :) It will probably take about a month to arrive, so while I wait, I can start amassing ICs. Together with the ICs for the CGA card replica, that is a total of 159 ICs for a single setup (XT motherboard with CGA card), with 55 different types. Oh boy, this is going to be the most complex and expensive project for me. And most fun.
 
The boards are in the mail, slowly moving from china to the 'glorious motherland.' That gives me some time to assemble all the parts I need for the motherboard. That includes three basic categories: 'chipset' (chip set, so this is where this word comes from, heh), logic ICs, and 'misc.'

The first group is easy. Lots of 41256, 27C256, 8088, and 8237-8253-8255-8259-8284-8288. The first three are easy, and the 82xx could be sourced from aliexpress and our local online flea market. There were soviet clone ICs of the 82xx line, so I can also use them if needed. I modified my test clone motherboard by installing sockets for all the major chips, so now I can test all of the above right inside a working system. Notice the Soviet КР580ВВ55А instead of D8255A-5. It was tricky to install since it has pins with a 2.5mm step instead of 2.54mm, but nonetheless it works like a charm. Now I have enough ICs for at least one whole new motherboard, so it's OK. Check!

The second is logic, and that is rather bothersome. I think I got everything I need, but I still need to check all of it in an IC tester and make sure that it's all present.

The third is miscellaneous. That includes passives (a surprisingly small amount of elements), crystal, and the most problematic component - delay line. TD1 in schematics, a 100ns 5-output delay line in a DIP-14 package. It's responsible for the generation of CAS and Address select lines. Without it, the machine won't start at all (as expected). I looked EVERYWHERE (where I can buy from of course) - nothing. In desperation, I even bought a couple of dead 286 motherboards with such components so I could salvage them from there, but that is not a way out, plus I HATE to destroy old hardware. But after some additional searching, I found DS1100-100 chip, which is actually a modern version of the one I'm looking for. But it's SOIC-8 package, and I need DIP. After studying the datasheet and making sure that it's compatible, I sketched a tiny adapter board from SOIC to DIP in KiCAD. Today the boards arrived (a friend helped to add my gerbers into his order, so they arrived much earlier than my motherboards), assembled the first prototype and installed it instead of the original. The result is in the photo; everything works :) So the problem of the delay line can now be considered solved; aliexpress is full of DS1100-100 for less than $2, and since the board outline resembles DIP-14, the boards are tiny and could be ordered very cheaply. Check!

The same approach I want to use for resistor networks. There are five of them on the board, two different types and three different values. I made similar adapter boards so one can just solder 0805 SMD resistors and get a resistor network of a desired value in a DIP package. And it's possible to use original components if required. A win-win situation! They are so far untested, but I don't think there will be problems with them; they are dead simple. Check!

Now I have to finish XTIDE and Adlib board and that's basically it, I am fully ready for assembling and testing.

By the way, I am experimenting with running a Discord server (never done it; could be fun), so if anyone is interested - here is the link: https://discord.gg/3KUundc9
I think that I have some retro hardware that could be of interest to people, plus a little bit of PCB design knowledge to help others.
 

Attachments

  • 1.jpg
    1.jpg
    359.7 KB · Views: 20
  • 2.jpg
    2.jpg
    174.8 KB · Views: 17
  • 3.jpg
    3.jpg
    253 KB · Views: 14
  • 4.jpg
    4.jpg
    114.3 KB · Views: 14
  • 5.jpg
    5.jpg
    35.6 KB · Views: 15
  • 6.jpg
    6.jpg
    155.1 KB · Views: 15
  • 7.jpg
    7.jpg
    175.3 KB · Views: 15
  • 9.JPG
    9.JPG
    63.9 KB · Views: 18
  • 8.jpg
    8.jpg
    232.3 KB · Views: 18
... But after some additional searching, I found DS1100-100 chip, which is actually a modern version of the one I'm looking for. But it's SOIC-8 package, and I need DIP. After studying the datasheet and making sure that it's compatible, I sketched a tiny adapter board from SOIC to DIP in KiCAD. Today the boards arrived (a friend helped to add my gerbers into his order, so they arrived much earlier than my motherboards), assembled the first prototype and installed it instead of the original. The result is in the photo; everything works :) So the problem of the delay line can now be considered solved; aliexpress is full of DS1100-100 for less than $2, and since the board outline resembles DIP-14, the boards are tiny and could be ordered very cheaply. Check!

The same approach I want to use for resistor networks. There are five of them on the board, two different types and three different values. I made similar adapter boards so one can just solder 0805 SMD resistors and get a resistor network of a desired value in a DIP package. And it's possible to use original components if required. A win-win situation! They are so far untested, but I don't think there will be problems with them; they are dead simple. Check!
Progress :->. Would you post your two set of Gerbers here for community use?
 
I plan to make all of them opensource on github after testing, both PCB design project in KiCAD and gerbers, with all the additional boards. The boards aren't cheap, so I don't want other people to suffer financially from my mistakes. But of course if anyone is ready and want to take the risk - here they are. Inside are two folders - tested (Delay line PCB) and untested (Resistor network and the motherboard itself).
 

Attachments

I decided to go with the CGA card first. It's a proven design and is needed for the diag ROM to work with the motherboard. I don't think it will run on the first try, so it's better to prepare myself.
It took me 2 evenings to assemble the card, but alas, I missed two ICs when preparing the components. I think they will arrive by the weekend, hopefully.
It's kind of funny - american 74LS and soviet K531 series (74S copy) working together to bring CGA back to 2024.

photo_2024-07-06_01-50-41.jpg
(Sorry for the mess ^_^ )
photo_2024-07-06_01-50-42.jpg
photo_2024-07-06_01-50-42 (3).jpg

Meanwhile, I finished the Adlib replica card, the first card from the "The Great White Five" (motherboard, CGA, Adlib, Floppy controller, and XTIDE). Assembling it was like a walk in the park compared to CGA. It works like a charm!
 
The work is progressing slowly but steadily. The CGA video card was dead after assembly, but after a day of troubleshooting with the oscilloscope, I managed to get it to an almost working condition. I have a nice black and white image via composite (wrong symbols were due an aliexpress LS244 chip, after replacing it with an original LS244 the symbols were fixed), but the TTL output is somewhat distorted and shaky. Area5150 works for 90% of the scenes, but I haven't tested it via TTL, only composite. As expected, chips from aliexpress are of poor quality. They work in an IC tester but fail in the card itself in various unexpected ways. Now I think I will have to restock all the required 74LS chips for the video card and the motherboard. Damn! :( But even in this state it will be more than enough for running the diagnostic rom in my motherboard, as it requires a CGA card to work.

Regarding the motherboard itself, I'm waiting for the delivery of sockets (it's already more than a week late!). In the meantime, I assembled it as far as I could, assembled the resistor packs and the delay line, and programmed the GAL for the 640k mod.

cga1.JPG
cga2.JPG
cga3.JPG
2024-07-17 01.31.29-min.jpg
 
Did you clone the board using same track as 5160? from your public file I found it's not the same. the left is original board picture and the right is gerber file.
BTW, did you found the 5160 using some inner layer to layout signal track? seems not only power and ground.

Untitled.png
 
Did you clone the board using same track as 5160? from your public file I found it's not the same. the left is original board picture and the right is gerber file.
BTW, did you found the 5160 using some inner layer to layout signal track? seems not only power and ground.

View attachment 1283267
The traces are not like the original, I know. There was no way for me to copy original layout without having a sacrificial board and/or x-ray machine for inner traces study. I had to re-route it completely, using a combination of auto-routing and manual routing. I know that auto-routing is bad for large scale boards, but I had no choice. Low frequencies, high voltage levels - maybe signal integrity won't be such a problem for this project. (I hope) After all I made a ZX Spectrum board using this approach and it worked really well from the first go. And that board wasn't much simpler - 75 ICs, only two layers and about 2/3 of 5160 size.
Oh, and I managed to route all the signal nets using just top and bottom layerы, so inner layers are for power and ground only. :)
 
Soldering is done, and as I expected, I don't have some of the chips on hand, so I had to order a restock. For everything else, it's on the board. A fierce, frenzied mess of original chips, soviet ones, and probably fake ones from ali. There are even chips pulled from a rotted 286, on which the markings have come off over time, but after cleaning the pins and tinning, they work quite well. A ceramic CPU from 1984, the same age as I am. :) Everything tested OK in chip tester and/or another board. The probability that this monster will start on the first try is almost zero; I anticipate a long debugging procedure.
photo_2024-07-20_17-40-18.jpg
photo_2024-07-20_17-40-19.jpg
photo_2024-07-21_00-06-52 (2).jpg
photo_2024-07-21_00-06-54 (2).jpg
photo_2024-07-21_00-06-55.jpg
photo_2024-07-21_00-06-56 (2).jpg
 
Thanks for reply, You resistor "chip" is so cute:)
I saw you got most of the ICs ready now, hope you power on it soon.
The reason I'm searching for the board layout is , I want to repair a damaged board, the PCB was broken, sadly.
I want to find the layout and recover the wires, mabe crazy, still trying.
if i got some info about original board layout, then this work will be pretty easy.
Here is my board:
1.jpg

3.jpg

2.jpg
 
Hello! I have what I believe to be a very late model 5160 board in my machine, manufacture date sometime in mid-late 1986. Zero bodge wires on the back. I see you've already produced boards, would scans of my board be of any use to you at this point?
 
Hello! I have what I believe to be a very late model 5160 board in my machine, manufacture date sometime in mid-late 1986. Zero bodge wires on the back. I see you've already produced boards, would scans of my board be of any use to you at this point?
Probably not much help at this stage, but still thanks for the reply.

Anyway, today is THE DAY. I got the final IC, installed it, prayed to computer gods and powered up the motherboard. I estimated the probability to be 95% for a black screen with zero life signs and 5% for partial success. But boy, was I surprised!
1722206731455.png
It's 2 a.m. here at the moment, so further diagnostics will be for tomorrow. But I am going to bed REALLY HAPPY tonight. :)
 
Back
Top