Dodgy RAM pack

[angus]

I recently obtained a 16K RAM pack for my ZX81, looks almost brand new. If relevant it is the USA version.

Anyway the Zeddy 'boots', taking a second or so to scan the memory before the K cursor appears....then after about 10 seconds the machine crashes. This happens reliably with the RAM pack but not without it...it'll run all day on 1K, as indeed it has on a few recent occasions.

Any ideas?

And yes, I realise the best idea is probably to save up for a ZXpand!

[sirmorris]

Is it the same period of time every time?

[angus]

I have not timed it accurately, but I believe it is the same time. It certainly does not vary much.

I managed to type the first two lines of the POKE program which returns the total RAM before it crashed. This I did twice. On other occaisons I have just let the machine sit and it crashes, or have tried loading a 16K game.

It ran for 4 hours in 1K today....playing Lazy Frog

[sirmorris]

This is very strange - once the zeddy is running then there's nothing that changes the path of execution according to time. The logic is cleared of any implication by this fact, imho. The only thing that I can think of with a time-based component is a build-up of capacitance. This should point you to the caps on the board - electrolytic caps can 'dry' out but the ceramic disks generally behave themselves. The fact it's a TS rampack should have no bearing - the only difference is the internal aluminium paint shielding and the intrusive FCC warning on the back The design is the same. That said there are 2 types of rampack - I've only ever owned the ones with all discrete logic - there are ones with their own little ULA. Which have you got? If the ula overheated then that might interfere... Or a ram chip, actually.

Freezer spray is often recommended in these situations - bad chips heat up quicker than good ones.

[angus]

I'm keeping it at work so I shall open it up again on Monday but I didn't notice any ULA. Everything looked fine, as you said I noticed the silver paint on the inside.

What I shall do on Monday is time the problem accurately, and have a look at all the capacitors...and look for a ULA.

[sirmorris]

If you can do so without too much wobble you could feel around all the chips with your finger-tips and see if any one of them is heating faster than the rest. Old chips ran hot but they'd take time getting up to temperature

[Prime]

I think it's quite easy to tell which version you have, if it has only one board inside it is the later version with the ULA, if it has 2 boards linked by some ribbon like cable then it's the descrete logic version.

Cheers.

Phill.

[angus]

Its definitely two boards linked, I just assumed they were all like that as the one I had 30 years ago was the same (but without the paranoid FCC warnigns)

[angus]

OK so I had a look today, the RAM pack contains two circuit boards linked by a ribbon cable and says "ISSUE 2" on the first board (the one with the edge connector)

On Friday it would run for an estimated 10 seconds. This time I ran it for 4 minutes on three separate occasions without any crashes. BUT...I typed in the little program with the POKES and PRINT USR that is supposed to return the amount of RAM. This works in 1K but with the RAM pack it causes a crash.

The only other thing I noticed was that the chips on the 2nd board (left side in the photo) ran warmer than the others....but I wouldn't say they were hot.

[gozzo]

First thing is to check the three supply lines to the 4116(or equivalent) RAM chips - I've been told they're very touchy about voltages and whatever line powers up before the others! If any chip gets hot enough to make you swear when you touch it, it is definitely duff! I think the most likely problem is either a dried up electrolytic (I would change them anyway just to be sure) or a faulty zener causing an incorrect supply line, but anything is actually possible. Also wipe the ZX's edge connector with contact cleaner or WD40 to try to minimise rampack wobble crash.