Spectrum Computing

Hi all,

After finally acquiring another ULA - albeit from a Spectrum +2 - I replaced the one in my 128K, turned on the machine and was greeted with a screen alternating between these two...





At first I thought it was a problem with the ULA, so swapped them back, but the problem remained. It therefore seems my 128K has developed this fault during the last few weeks of not being used?!

Any suggestions?

Broken contended RAM chip.

Not broken, because the CPU can read and write it just fine.... Just out of tolerance.

Write a basic program to write bits to the screen to work out which one it is, and replace it.

It's an issue that only seems to affect Greys.

David

Thanks David.
This is a Spectrum 128k, not a +2.
Do you have a link to a sample program?

I've had 3 RAM ICs fail in quick succession in my +2 since resurrecting it recently. The +5V rail is fine, so I suspect it's just old age catching up with them. It's probably overkill, but all of mine are socketed now just in case.

You may find it easier to order a "Switchable Replacement ROM Module for Spectrum 128s" from Retroleum.

https://www.retroleum.co.uk/128-chips

One lives permanently in my +2 now configured as a 'Toastrack' for slot "A" and Brendan Alford's ZX Diagnostics for slot "B".

Phil also stocks the Unrainer mod now since I requested one a few weeks ago so, if you're handy with a soldering iron, you may want to drop one of those in too while you're in there.

I realized I could use the RAM tester on my diagnostic hard. It reported no errors.

Could it be a false negative?

KeplerOfMars wrote: ↑Sat Jul 29, 2023 4:02 pmCould it be a false negative?

Possibly. One of my 3 failed RAM ICs passed the memory test in Brendan Alford's Diagnostics and the screen corruption looked very much like yours.

I’m not entirely convinced it’s the RAM, but from the pictures you posted it looks like it’s either IC 22 or IC 13. If you don’t have spares, try swapping them with any other RAM IC and see if the vertical lines change position.

Things that make you "hmmmm"...

It looks like bit D7 in the contended RAM is stuck high, hence the stripes. But if that's true then how come the attributes aren't similarly affected? That would the whole display flash.

I must say, the QL Minerva ROM has never given me a false negative.

Is there a way to identify the bad IC using a logic probe?
I don't really want to desolder all 14 ICs and test them in my RAM checker...

OK...I tried the old piggyback technique and found that placing a 4164 on the back of IC19 & IC20 produced the following results in my diagnostic cartridge:




So I decided to quickly remove, test and socket those two ICs even though the diagnostics refer to other banks? But at this stage I don't trust the results...

Suffice to say, in my RAM tester, those ICs were fine and the problem remains. It looks as though I will have to remove them one by one and test them.

KeplerOfMars wrote: ↑Fri Jul 28, 2023 8:33 pm

Why would bad 4164s cause the whole display to have inverted colours, though?

Ah, bad 4116s would cause the stripe, and maybe force set the flash bit. Is the display alternating at the usual flash rate? What does PRINT PEEK 22528 give you?

Wait a minute!

KeplerOfMars wrote: ↑Fri Jul 28, 2023 8:33 pm turned on the machine and was greeted with a screen alternating between these two...

I think he means that it is flashing.

Correct...I have now removed, tested and socketed 5 ICs...never have I wanted to find a bad RAM IC as much...

Okay, so that being the case, looking at the 128k schematic, I think it's probably IC13 that's snuffed it. That looks to be bit D7 of contended RAM.

But IC13 is one the 5 I have checked, and it passed with flying colours...

A bad soldering or cut track would also keep D7 high.

Seven.FFF wrote: ↑Mon Jul 31, 2023 3:27 pm Why would bad 4164s cause the whole display to have inverted colours, though?

Ah, bad 4116s would cause the stripe, and maybe force set the flash bit. Is the display alternating at the usual flash rate? What does PRINT PEEK 22528 give you?

PRINT PEEK 22528 gives me 56.

I have now checked five RAM ICs and just tested the machine.

The good news is, it is no worse than before. The bad news is mentioned in the previous sentence.

Should I continue and check the rest?

Is there a little solder splash or wire offcut bridging pins 38 and 39 on IC1 (the ULA)?

No.
So...I have now removed all the ICs - and managed to break two in the process - tested them all, soldered new sockets...replaced the broken two ICs, the same...

This morning I removed all the RAM ICs and put them back in different positions, same result. Correct me if I am wrong, but does that not prove that it is not a RAM issue, unless of course a faulty RAM IC had been put in the position of a previous bad IC?

I'm not an electronics expert but could it be the capacitor for the ram chip for bit7 that's faulty?

I suppose I could test each capacitor without having to remove them...