Killed my CPC 464.. help with diagnosis please

[||C|-|E||]

Very useful tutorial! Yes, I didn't have these very fine cutters, so I couldn't get too close to the pins somehow. Need to get better equipment for that kind of job.

Has anyone ever tried compressed air to blow out the solder from the holes? This is really the hardest part of the job I think.

Bryce, what kind of iron are you using for the wick. It never seems to be doing a lot for me. Maybe the tip of my iron is to fine, or it doesn't have enough watts?

Huuuum... that solder should definitely go just with braid and enough flux. I tend to use this one:

https://www.ebay.co.uk/itm/15m-Reel-of-2mm-Chem-Wik-Solder-Remover-Wick-Mop-Desoldering-Braid/272289238812?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649

with flux similar to this:

http://www.ebay.co.uk/itm/New-flux-soldering-paste-in-the-20g-tin-for-electronics-SMD-plumbing-DIY-etc/253018379203?epid=2255531445&hash=item3ae91213c3:g:Bk8AAOSwIVhaKCtC

What I do is to impregnate the braid with the flux, then I put it on top of the hole with the solder and then I heat it until all the solder goes into the braid by capillarity. In very rare cases, it may be necessary to add a little bit more of fresh solder to help the process, but this is not common. Then, as Bryce said, if you cut the pins close to the IC, it is very easy to take them off with a couple of pliers .

However, I would say that the way I do it is pretty old fashion. Nowadays you have this odorless liquid flux that works very well and it is not messy, like the one I use. Actually, I always need to clean all my boards afterwards with isopropanol. I just keep working with it because I have tons and I feel really at home with the stuff .

Regarding the iron, I would definitely go for a soldering station. You can buy reasonably cheap ones that do the job very well and will allow you to avoid overheating the board. I bought an Atten from RS components and I am really happy with it. It is nothing special, but more than enough for my needs, and it was less that 50 pounds 

P.D: for the traces, Audronic advice is very good, you can just use a very thin cable and it will do the trick. The picture shows an AMX mouse that had a bad track and I just bridged the two positions with that little yellow wire. You could try something like that.

[pabloingreece]

Hi LambdaMikel, I had to do exactly the same process as you in a non working cpc, the same "grey screen of death" and finally I repaired it and came to life after changing the ram and, as strange as it sounds, a resistor which was in the way of the INT signal from the 40010 IC... read here: http://www.cpcwiki.eu/forum/amstrad-cpc-hardware/grey-screen-of-death-(revisited)/ and here (in greek, but you can google translate) http://underground.iamretro.gr/index.php?threads/%CE%95%CF%80%CE%B9%CF%83%CE%BA%CE%B5%CF%85%CE%AE-amstrad-6128-%CE%BC%CE%B5-%CE%BA%CE%B1%CE%BC%CE%BC%CE%AD%CE%BD%CE%B5%CF%82-%CE%BC%CE%BD%CE%AE%CE%BC%CE%B5%CF%82.8555/

Of course you should put dip sockets, because it will be easier in the future to replace damaged ram chips, I did it in that way, and it's also better for testing the ram.

I have an iron with 25 watts, and it was hard to desolder, it took me two days to achieve it, and in the process I broke some tracks which I had to patch with small cables. A trick is to hot the board with some heat air pistol or a dryer for the hair, and the apply the iron, it will be more easy.


So, review your tracks with a multimeter in order to check continuity. I recommend you to find an amstrad service manual in order to detect where the problems with tracks could be.


Good luck!

[Bryce]

Very useful tutorial! Yes, I didn't have these very fine cutters, so I couldn't get too close to the pins somehow. Need to get better equipment for that kind of job.

Has anyone ever tried compressed air to blow out the solder from the holes? This is really the hardest part of the job I think.

Bryce, what kind of iron are you using for the wick. It never seems to be doing a lot for me. Maybe the tip of my iron is to fine, or it doesn't have enough watts?

I use an Ersa iCon Nano with a chisel tip for most work. Usually set around 360°C. Definitely only use branded braid, the cheap shit usually doesn't contain enough flux (if any). Also, buy yourself a flux pen to add more where required.

The old solder will be very dry and, as the PCB was wave soldered, it will contain zero flux. So although it may sound counter-productive, add some new solder to each hole before you try removing the solder with braid.

Bryce.

[khaz]

I've been using a cheap desoldering iron:



https://www.ebay.co.uk/itm/30W-220V-50Hz-Electric-Vacuum-Solder-Sucker-Desoldering-Pump-Iron-Gun-E-Class-/172584401671

For $30, it makes removing through hole chips ridiculously easy. You just have to be careful when releasing the red button, as it spits out everything it sucked (so point it away from your board!)

[Bryce]

But you've no idea of how hot it is, so it could be destroying parts or even the PCB!

Bryce.

[khaz]

I have no idea how hot my regular iron is either ¯\_(ツ)_/¯

You raise a valid concern, but these things tend to be not hot enough instead of way too hot.

I've damaged a pcb before, trying to remove a chip with a simple soldering iron. The thing was stubborn and I was new to soldering, I ended up breaking some of the pads when yanking it. With this desoldering tool, I haven't had a single problem. Put it over the stub, press the button and do a circular movement to detach the pin, and done. Sometimes I use a pair of tweezers and shake the pin back and forth to snap it from the hole if its still attached.

[Bryce]

For freeing the pins from the solder in the hole you can get these: https://www.ebay.de/itm/8PCS-Lots-Hollow-needles-desoldering-tool-electronic-components-Stainless-steel/172280051221
You just push the hollow point over the end of the pin and into the hole. Much safer than shaking the pin with a tweezers.

Bryce.

[khaz]

Oh, very nice! I like that.

[LambdaMikel]

For freeing the pins from the solder in the hole you can get these: https://www.ebay.de/itm/8PCS-Lots-Hollow-needles-desoldering-tool-electronic-components-Stainless-steel/172280051221

Exactly what I was looking for! Will order a pair these... much better than the wire I used (and the leads of some components, such as resistors).

I finally have sockets in the board. I used a can of compressed air to blow out the solder from the holes. I guess I really need a better iron or better desoldering wick. And then, who on earth designed this board??? The spacing / pitch of the RAM ICs doesn't fit the IC sockets! I had to bend the IC socket pins  - the width didn't match! For that reason, I also couldn't use the standard spring-loaded IC sockets, but had to use the round ones. The spring-loaded ones would just pop out of the socket while I was trying to bend the pins to make it match and push them in.... what a nightmare.

[LambdaMikel]

Hi LambdaMikel, I had to do exactly the same process as you in a non working cpc, the same "grey screen of death" and finally I repaired it and came to life after changing the ram and, as strange as it sounds, a resistor which was in the way of the INT signal from the 40010 IC... read here:

Thank you Pablo, that is good information, very useful!


I did the memtest.bin again with all RAM ICs removed by now, and I would have expected an all red screen. It still shows some weird patterns after the border changes, so maybe some other components are damaged as well. I will know soon - my RAM should arrive in 2 days.


Btw, does anybody know what the screen of memtest.bin should look like with all RAM chips removed, and all other chips being fine?


Another concern I have - I accedentially heated some ceramic capacitors quite a bit during the unsoldering process... are they going bad easily from that?

[LambdaMikel]

I've been using a cheap desoldering iron:

Wow! Quite a weapon! Hopefully I will never have to go through such an extensive desoldering job again, but if, I will consider using this tool.

[pabloingreece]

Thank you Pablo, that is good information, very useful!


I did the memtest.bin again with all RAM ICs removed by now, and I would have expected an all red screen. It still shows some weird patterns after the border changes, so maybe some other components are damaged as well. I will know soon - my RAM should arrive in 2 days.


Btw, does anybody know what the screen of memtest.bin should look like with all RAM chips removed, and all other chips being fine?


Another concern I have - I accedentially heated some ceramic capacitors quite a bit during the unsoldering process... are they going bad easily from that?

I think the test is only for testing with ram connected, if you don't have any ram, I don't think it tests any other component in the board.


In my case, I think the ram stopped the overvoltage to go further, because all the other components were fine, also the fried resistor could have acted as protection. So wait for the ram chips, check track continuity and check then if the ram test is ok.


If this does not resolve the problem, then will see.

[gerald]

I did the memtest.bin again with all RAM ICs removed by now, and I would have expected an all red screen. It still shows some weird patterns after the border changes, so maybe some other components are damaged as well. I will know soon - my RAM should arrive in 2 days.


Btw, does anybody know what the screen of memtest.bin should look like with all RAM chips removed, and all other chips being fine?

I've just tested that and what you see is expected. RED mean all bit a read at 1, which is expected for floating TTL bus. The moving pattern is a effect of the floating bus and other activity.
The pattern changes after the border change are linked to the Z80 activity.

If all RAM is good the screen should be black. The memory is cleared by the last test pass.

Another concern I have - I accedentially heated some ceramic capacitors quite a bit during the unsoldering process... are they going bad easily from that?

That should not be a problem.

[LambdaMikel]

I've just tested that and what you see is expected. RED mean all bit a read at 1, which is expected for floating TTL bus. The moving pattern is a effect of the floating bus and other activity. The pattern changes after the border change are linked to the Z80 activity.

That is good to know! This gives me hope that by replacing the RAM the CPC can be fixed. Thanks for this test, Gerald! 

[LambdaMikel]

My RAM arrived today. I also tried to fix the traces. It got better for sure, but still some errors. Is it possible to tell what the remaining problem is? I checked all PINs of the RAM ICs with the logic probe if they receive pulses while RAMTEST is running, they are all active by now, after having fixed the tracks. Maybe the patch cables are too long?

[LambdaMikel]

Oh, I forgot one defect track... now the screen grizzle is gone, but the border still shows a red bar.
What could that be, what does it mean?

[Audronic]

Mark the Ram chips.
Move 4 up to new position, replace the 4 that were moved from the old position.
and see if the Red bar (s) has changed position


Good luck Ray

[LambdaMikel]

Welcome back, beloved 464 
Sorry I did that to you... I did such a poor job desoldering. But good learning experience.
Happy I was able to resurrect it.

Thanks to all who helped, you guys rock!!! 

[Audronic]

Ok thats good.


? what was finally wrong ?


Thanks     Ray

[LambdaMikel]

? what was finally wrong ?

I replaced all the RAM chips, socketed them. I don't know which individual RAMs were bad, I suppose all, according to MEMTEST.

Then, there were more traces disconnected than I thought, caused by crude desoldering. I had made highres pictures of the traces before I put the sockets in; I inspected them, and that way I found the first bad tracks.

But then some disconnected traces weren't very visible. I finally looked at the PINOUT for the 4164 RAM chips and figured that A0 - A7 should be connected identically to all 8 chips, so I tested continuity with my multimeter between all pairs of corresponding data bus pins, and figured out that a couple more connections were broken -> more wires. Finally, everything was green in RAMTEST.

So it was really the combination of understanding the pinout of the 4164 and the continuity tests that enabled me to fix the problem.

[||C|-|E||]

Look at it, working again  Congratulations! 

[LambdaMikel]

Look at it, working again  Congratulations! 

Thanks 

[pabloingreece]

Great news! as I told you, I had te same problem, same broken tracks. I reviewed the schematics of the service manual and before removing I made some photos. So, I understand completely the awesome feeling of bringing back to life the old friend of us, the cpc.


Good job!