CPC 6128 likely faulty RAM?

[tomek_c]

Hello everyone,


I obtained a CPC 6128 in "untested" condition, and wanted to make it up and running.


When powered on, I saw all-white screen and nothing more happened.


Thanks to very helpful thread Need some help fixing CPC6128 not booting I followed the procedure of pulling out PAL, wiring the socket so that I switched to bank 1. Boom! It worked! Basic showed up.


I went ahead and switched to Bank 0 only - white screen.


Do I assume correctly that there is faulty RAM in Bank 0? How can I proceed with my average electronics skills (soldering and can tell the difference between scope and multimeter)?

[Bryce]

Hi Tomek,
     welcome to the Forum and congratulations on getting that far with the diagnostics. Yes, your lower 64K has a damaged chip. The broken chip may be getting hotter that the others and it may not. Either way, I would suggest you get yourself 8 new RAM ICs and 8 sockets for them. Remove the old ICs, install the sockets and put in the new ICs. Then all should work as expected when you put the PAL back in.

Bryce.

[endangermice]

Well done on getting so far with the diagnostics you're definitely on the home run now!


In addition to what Bryce says, if none of the chips are running hotter than the others, you may be able to piggyback a new one on top of each chip in turn to identify the faulty one (assuming only one is faulty which I suspect is the case). It doesn't always work but I've had a couple of successes in the past with this and it can save a lot of needless resocketing.

[tomek_c]

Thanks guys, I wanted to approach forums with homework done I ran computer for couple of minutes, and all chips become warm, but none I could call hot. Piggybacking means placing a good memory chip on top of suspect, power on, check, rinse and repeat? Maybe is there a way of checking certain pin with multimeter? I already tried Vcc, and on every chip it is 4,73 V.P.S.I am using scalvaged power supply (5V, 12V, 2A) which is switching PSU and I guess the original was transformer based. Can this make a difference?

[radu14m]

i try this method: placing a good memory chip on top of suspect, power on, check, rinse and repeat
but did not work for me, maybe i had bad memory chips too.

let me know if this method work for you...

[tomek_c]

Another quick thought:


Is possible to switch bank 0 and bank 1 logically and run a kind of self-test to diagnose the problem closer?

[KaosOverride]

Two little tips


Memory chip can be shorted or broken. If shorted then it gets hot, and in my cases, 5v power of the memory block drops to 2.5v +-. If broken, then the other 7 chips have 5v power and the damaged is not hot.


Also, check your board revision. If it's the rev3, with the preASIC which replaces the crtc and gate array, you are lucky. There are 4 free chip places that accept the 41464 memory chips from the CPCplus. You can cut the power and ground pins from the defective bank or remove them, and fit 2 chips 4146, or do the hard way and replace the 8 chips. But only for rev3 boards...


And yes, the piggy back tip will work for broken chips, not for hot and shorted ones 

[KaosOverride]

Another quick thought:


Is possible to switch bank 0 and bank 1 logically and run a kind of self-test to diagnose the problem closer?

You need to place the PAL, and leave the cas0 and cas1 pin signals out, and then crosswire them. No PAL, no expanded ram 

[Bryce]

Generally I don't recommend using switchmode PSUs on the CPC, but even if you do, it wouldn't cause the problems you are seeing, it's definitely a RAM IC that's dead. You could as you suggested swap the banks manually, then test for a "stuck bit". Each address is spread across all 8 chips - each chip stores 1 bit of each address). It might only be one single bit or it could be the whole chip, so you'll have to make a routine to go through the entire 64K and comparing the data read to what you wrote (just &00 and &FF should be enough).

Bryce.

[tomek_c]

I just got brand new 4164 chips (and sockets). I read somewhere that Bank 0 are chips IC127 to 134, i.e. right hand side looking from top, is that correct?


BTW is there any sort of basic reference of CPC hardware to answer such questions?

[gerald]

I just got brand new 4164 chips (and sockets). I read somewhere that Bank 0 are chips IC127 to 134, i.e. right hand side looking from top, is that correct?


BTW is there any sort of basic reference of CPC hardware to answer such questions?

The best answer is in the service manual 
But yes, Bank 0 is IC127 to 134, right hand.

[tomek_c]

With friend's help I was able to replace all chips in Bank 0.


Still computer is not operational, though.


Only improvement so far is that instead of white screen, there is blue background with some 8x8 squares and random dots.


Any clues what can we check further?


Thanks,

[gerald]

With friend's help I was able to replace all chips in Bank 0.


Still computer is not operational, though.


Only improvement so far is that instead of white screen, there is blue background with some 8x8 squares and random dots.

Blue background means that the RAM is at least partially working.
Are you RAM on socket ?
Can you check that all new RAM are properly connected, ie no short/open on address/data pin.
You can check that bank 0 device are properly connected to back 1 pin per pin with a continuity tester (except for cas)

This could also be a ROM problem, but it was working fine when using bank 1.

[Bryce]

That's a big improvement. I assume you have reseated the socketed IC's?

Bryce.

[tomek_c]

@Bryce, not sure if I understood: to reseat is to pull, possibly clean and replace ICs which are socketed?


We'll check the connections thanks for the tip.

[Bryce]

Yes, just remove them from their sockets, check the pins and sockets are clean and re-insert them.

Bryce.

[tomek_c]

Thank you guys all, case closed.


RAM replacement did help indeed. After longer investigation the latter problem (after RAMs were replaced) was... tiny break in PAL socket connection to PCB.
I must have broken it when pulling PAL out of socket.

[endangermice]

Very easily done!


Great news you got it working again. Most of the time, the problem is faulty RAM so unless the symptoms tell you otherwise, usually my first port of call.