464 grey screen / black border, always the RAM ?

[reno]

Hello,

I have a 464 with a MC0002B board (40007 GA) that's booting to a grey screen / black border.
I know that this is usually the RAM, so I've ordered and just received new 4164 chips + sockets.

I opened the 464 :
- none of the RAM chips are getting particularly hot. Impossible to pinpoint an obviously dodgy one.
- Piggybacking a new chip or 8 new chips doesn't help either : depending on which chip and how well I made the connection, it gives me either no change i.e. grey screen (most of the time), sometimes a black screen, or garbled grey/black output, and sometimes even a relay sound coming from the tape assembly.

Given that, I'm a bit reluctant to go ahead and desolder / replace all 8 chips. 
My soldering skills are average, I have a soldering station, some wick and a pump and that's it.
If there's a good chance that replacing the chips still won't help, I'm not sure I'll have time to look further and I might just sell the 464 for parts now.

The board looked very clean (possibly cleaned up by the previous owner), but while taking close up pictures I noticed some rusty (and possibly dry ?) solder joints on most of the RAM chip pins, and some of the 74LS chips too. See pics attached.

I used some contact cleaner + a toothbrush but it didn't help.

Questions :
- What would be the next steps ? Does it make sense to try and fix the solder joints first on the current chips, before installing new ones ? If so, how ? Can I do it from the bottom ? Or should I just cut the current chips, desolder and clean everything up then install sockets ?
- I have an M4 board : is there a ROM I could try that would help me pinpoint the failed chip(s) to at least reduce the amount of work ? My M4 is set up for my 6128 currently, and I can't remember what is in what slot, but using it on this 464 makes no difference to the grey screen. The reset button works, if that helps.

Thank you !

[Bryce]

That's not rust, it's exposed copper. You should run a RAM test to find out which chip has failed.

Bryce.

[reno]

That's not rust, it's exposed copper.

Thank you. Do you mean this looks normal to you and doesn't necessarily need fixing ?!

You should run a RAM test to find out which chip has failed.

Ok I've managed to run Noel's Amstrad Diagnostics from the M4. It shows chips 2, 4, 6 are dead.
I notice they are all even numbered : could that indicate the failure of another logic component ?

Can I assume the chips are dead and start replacing them, or should I continuity test something / inspect their solder joints and maybe try to fix that first ?

I'm puzzled that piggybacking didn't make a difference yesterday, so I don't want to do replace chips for nothing. But now that I have the RAM test running, I'll try again.

[reno]

Ok, piggybacking a new chip in fact works for 2 of the 3 failed chips. Only chip 2 (IC118) is still stubborn.

[eto]

So, I'm no expert in electronics, but I would expect, that piggybacking will only help, if the RAM chip is fully dead and does not interfere with the data lines any longer. 

Before checking any other IC I would replace it and see if it solves it. 

[TotO]

Piggyback is perfect to lose the extra soldered parts.
Cut the defective IC pins, remove them one by one, add a socket.

[Bryce]

Piggybacking only works for very specific failures. If the broken IC is holding any pin high or low or the IC is pulling too much current, then piggybacking won't work. I never bother even trying because the RAM test will tell me what to swap and I swap them.

The fact that the failed chips are 2,4 and 6 is a coincidence, it doesn't point to any other failure.

Bryce.

[reno]

Well, thanks to Bryce's IC replacement tutorial on the site (and a lot of flux and patience) I managed to replace the 3 failed chips. Success !

But short lived success ... after about 30 mins of testing, I got a grey screen again, this time slot 3 (IC117).
This one had never failed before. I wonder if I should bite the bullet and replace them all. If I can get away without it I'd love that, but if they all start failing one by one ...

[eto]

this time slot 3 (IC117)

Are you sure you have the right power supply? 

[reno]

this time slot 3 (IC117)

Are you sure you have the right power supply?

I thought about that too because I sometimes use a USB-DC 2.1 adapter powered from a USB hub, and that could be a little tight especially with the M4 plugged into the 464. But I tried with a proper 5V 2.5A PSU and had the same issues.

I've now replaced IC117 as well and ran over 100 iterations of the soak test in the Amstrad Diagnostics 1.3 ROM, so hopefully all should be good now.

I've read somewhere that it's best to avoid using different RAM chips though, but I'm not sure why. Do you think I should replace the other 4 while I'm at it ? At least they would all be socketed ...

Also, would you replace the thermal paste on the Gate Array heatsink ?