16K RAM Pack Schematic / Chip Layout

[TMAOne]

Hello, new poster in Canada here. I bought the ZX81 kit when it first came to Canada in May 1982 and cost $150 plus taxes and separate power supply. My soldering skills were poor back then, but got the job done. 30 years on, that one has been Frankenstein'd in various ways, but still beats me at chess. Dispassionate ungrateful little troll,...

But anyway, I have the following issue. I have a 16K RAM pack that has a bad bit. After much programming and scanning memory to identify the bad part, (errant programs frustratingly commit suicide by writing over themselves or active memory), I have identified the bad bit as the 2**1 (2) bit of address 26180, stuck at 0. In other words, POKEing 255 to that byte and then PEEKing it back returns 253.

I have searched the internet in vain for a schematic of the RAM pack or any other information that will tell me which 4116 chip of the eight corresponds to 2 to the 1st. I actually have some spare chips I ran across somewhere, (also other even "deader" RAM packs that could donate a chip), and would undertake meatball surgery on the dual-sided board to extract the bad tooth and replace it, but I do not fancy doing that more than once in a trial-and-error exercise.

Does anyone have a breakdown of the RAM pack layout?

Ian

[makeit2day]

Here a link to view the schematic.
http://www.retroisle.com/sinclair/zx81/ ... matics.php

[RetroTechie]

Welcome here, Ian!

I have searched the internet in vain for a schematic of the RAM pack (..)

You call that searching? Googling for "zx81 ram pack schematic" (just the words, without the quotes), 1st result: http://www.myprius.co.za/ZX81.htm
Note that there are different versions of the RAM pack, so schematic on that page may not be exactly what you have. IIRC some RAM packs use 74LSxx chips for the DRAM control logic as in schematic, some use a custom IC for that logic. I'm afraid I don't know whether RAM IC numbering (or location on the board) is the same among those versions.

2^1 corresponds to D1 line, so you could just trace that back to the 4116 chip (pins 2 and 14, connected) starting from the RAM pack's edge connector. Schematic above points to IC9, but I wouldn't rely on that, just trace the connection & verify with an Ohm meter.

FWIW: You may consider moving 1 topic down. If you can unsolder IC's cleanly, internal RAM expansion is easy to do. Does away with the infamous RAM pack wobble, supports true Hi-Res, and keeps the edge connector free. Also the RAM pack contains some voltage conversion circuitry that gives problems in many cases, and those 4116 RAMs are nasty critters in general (and 30 years old).

[1024MAK]

Have a read through of this thread.

This has details (inc. pictures and a partial diagram) that may help if your RAM pack is the same version.

If your RAM pack is not the same, post a hyper-link to a picture of the PCB.

I would say upload a picture to this forum, but we are currently getting a "Sorry, the board attachment quota has been reached" error message

Mark

[TMAOne]

Thank-you for the welcome and the great information. It was some time ago I did my searching for a schematic. My bad. I found this forum and got a little excited, and posted without doing a renewed search to see if anything had changed.

I'm glad I found you nice knowledgable Zeddy folks! The diagram was great, and the suggestion on how to correlate D0-D7 to the actual chips was just what I needed.

I wanted to include a photo of the RAM pack in question, and the chip layout which the schematic doesn't help with, but got the upload error I was warned about.

As you would have been able to see: (1) I have the 74LS flavor of logic chips, (and didn't know there was any other kind), (2) I was mistaken about this particular pack using 4116 chips--it uses the "direct replacement" AMD 9016EPC variety, and (3), I'm glad I didn't use trial and error to find the 2^1 chip!

In case it will help someone else some day who wants to do the same thing, the layout is as follows: with the memory board positioned away from you (the short ribbon bridge at the bottom), there are two rows of chips. Two in the top row, six in the bottom. Left to right the chips are, top row: D2, D1, and bottom row: D5, D3, D4, D6, D0, and D7. Not exactly intuitive--probably a quirk in how the traces needed to be layed out for circuit design. (I wonder if it was designed on a ZX80.)

At any rate, here's what happened: The unsoldering of the chip went better than I expected. I use a desoldering iron with a built-in vacuum bulb, but was concerned about connections on both sides of the board. It went fine, actually.

I tried to put a socket in, but then I couldn't solder the component side if I did that, so that was out.

I soldered in a 4116. No go. Tried another chip. No go. Perhaps the "direct replacement" 9016's have a timing issue or somehow resent a 4116 in their midst, but for whatever reason they didn't play well together. I got white-outs, occasional video puke, and sometimes the customary faint vertical stripes, but never a K prompt. Darn.

So to test the validity of my soldering I put the original chip back. It works as before. One bit of one byte always 0.


I did look at the other thread, and it has good information too. I have a pack that produces similar garbage on the screen to the pictures there, and one that works memory-wise, but produces distortion on the screen. I've tried replacing electrolytic caps, but it hasn't cured either of those packs. I will check the voltages another day. It is encouraging to finally find others working on similar issues.

One of my 2 ZX81's does have an internal 32K modification, which works very well in spite of my clumsy implementation. I just want to keep the other machine as a 1K / 16K "original". When you get to be an old computer geek you're entitled to be eccentric and romantic,... I think it's in the bylaws somewhere.

So,... I looked. AMD 9016EPC on eBay, $2.37 plus $13.95 (!) shipping. Sheesh. Have to think on that one,...

Thanks again for the great advice.

Ian

[RetroTechie]

Also have a look through this thread: ZX81 memory troubles

(2) I was mistaken about this particular pack using 4116 chips--it uses the "direct replacement" AMD 9016EPC variety

There were many manufacturers of these chips, and if fast enough they're usually interchangeable. I use 4116 as a generic identifier: 16K*1 bit, 16 pin DIP with same pinout, +12, +5, and -5V supply, manufactured in early '80s or so. All the same crap, really... The only other thing to watch is speed (access time). There's probably lots of room, but I would only fit chips that are at least as fast as what was there originally. From the datasheet that AM9016EPC appears a 4116 replacement like any other.

Did you use a known good one? When is the last time you saw the replacement IC successfully in action? What's happened to it since?

I tried to put a socket in, but then I couldn't solder the component side if I did that, so that was out.

You only need access from solder side to put in a socket. These 4116's beg for sockets as they're very prone to failure. So if vertical space allows, I'd use sockets. Also if you have some socketed (+ software to test the RAM), it makes a good tool to test other 4116's.

But before you do: have you checked supply voltages in operation? Especially the -5V, low current but likely to kill a 4116 when missing.

So,... I looked. AMD 9016EPC on eBay, $2.37 plus $13.95 (!) shipping. Sheesh. Have to think on that one,...

IMHO a 4116 doesn't offer any advantage as replacement for a dead 4116 (not even as 1:1 replacement for a single chip), there are a number of better alternatives to choose from. 4116 chips should be in museums or collector's hands by now, not serving to add 16K to a ZX81...

[RWAP]

Have a read through of this thread.

This has details (inc. pictures and a partial diagram) that may help if your RAM pack is the same version.

If your RAM pack is not the same, post a hyper-link to a picture of the PCB.

I would say upload a picture to this forum, but we are currently getting a "Sorry, the board attachment quota has been reached" error message

Mark

Ah well why didn't someone say - I have increased the amount of space for attachments now

[TMAOne]

No, the 4116 chips that I have are not "known good". In fact, I don't even remember what I pulled them out of, or when. They've been in an anti-static mat in the "IC drawer" of my small parts box for ever.

Sooo,... I did get around to checking the voltages of my 3 non-functioning or semi-functioning packs. Got a surprise.

The one that doesn't work at all has voltages of -5.16, +12.09, and +4.98 at pins 1, 8, and 9 repectively. All good.

The pack that runs programs okay but makes for bad video shows -4.95, +12.24, and +4.98. Also cool.

The pack under scrutiny with the "bad bit" shows -1.27 (!!), +9.40 (!), and +4.99. Holy cats. It's a voltage problem.

This absolutely begs the question, how do the other 131,071 bits function properly on such a starvation diet?

So I guess I'm heating up my (de-)soldering irons again to swap a transistor and a bunch of diodes from the dead pack with good voltage into the almost perfect pack that couldn't run an LED on a dark night.

Curiouser and curiouser. Here's the photo I wanted to show you.

Ian

[TMAOne]

Argh.

I realized I haven't changed the caps in this particular unit, since I didn't think a cap could be responsible for one bad bit. But given that the voltages were off, I started with the big 100uF and replaced it with one from my salvaged parts supply. Suddenly, no boot.

Okay, maybe that was a bad cap. Replaced it with a brand new one. No boot. Strange.

Put the original back in. No boot. Aw rats, I've blown something else.

This pack now has poetentially 3 issues: a boot-preventing problem, a voltage problem, and a stuck bit. It is now my worst pack!

Another has good voltages but won't boot. The last has good voltages and does boot but causes interference in the video. I'm not willing to fool with that one until I have another that works better.

My potential "donor", the one which wouldn't even boot before, is actually in better shape now than the one I was trying to fix.

I think I'm punching above my weight class. Although I can solder better now than when I put this computer together, my diagnostic skills are lacking, especially when it comes to digital. I'm not professionally trained in electronics, rather just a "script kiddie" when compared with a hardware "hacker".

I do have access to an oscilloscope, but don't have the skills necessary to search out the defective part(s).

I'll read that other thread again for ideas, wait for whatever you guys have to say, and think about this later.

Right now I'm going out to cut the lawn and forget about it.

Ian

[RetroTechie]

Yes I'd start with those electrolytic capacitors too, just replace all 5 of 'm. Not much point in replacing with a used one, or putting the old one back in. Replace with a new one, and leave that. And watch polarity! Best do 1 by 1 so you don't lose track of what goes where & how. Optionally you can use multilayer ceramic capacitors for smaller values like 1 uF.

If that doesn't fix voltages, the transistor might be next likely suspect. When powered before, did you feel any of the RAMs heating up more than the others? Voltages might drop if a RAM IC fails to the point that it starts working like a wire. Better not power it now... before you've replaced all electrolytic capacitors, that is. When that's done, power up quickly just to measure voltages again. If not correct, power down again ASAP! If correct, back to software tests (hopefully).