Amstrad CPC 464 - Screen with Garbage

[jarcher1980]

Anything below 4.75V could make the computer unstable.

Bryce.

Thanks for confirming. I'll keep checking the voltage rails and hopefully I can detect where it drops.

[jarcher1980]

Anything below 4.75V could make the computer unstable.

Bryce.

Thanks for confirming. I'll keep checking the voltage rails and hopefully I can detect where it drops.

Still investigating... when checking the power rails from CP001 I decided to check the nearby R145. I suspect this isn't connected to the issue, but the Ohms reading was 429. The schematic suggests it should be 1k. What do you think? Thanks.

[Bryce]

If you were measuring it in-circuit, then it was in parallel with several other devices, so the value could be anything!

Bryce.

[jarcher1980]

If you were measuring it in-circuit, then it was in parallel with several other devices, so the value could be anything!

Bryce.

So, perhaps not an issue.

Thanks for helping on this Bryce. Much appreciated.

Please bear with me as I present another potential 'non-issue' as the fault may be elsewhere...

I've been checking voltages at different locations. I'm using a simple wire bridge (with two female connectors at the end so there's no contact with other pins) to connect pin 1 and 3 of CP001 to switch on the CPC. It seems to work fine—with the same garbage centre or grey centre as before showing on screen.

If I remove this bridge and measure the voltage at pin 3 (labelled 'A' in my image below) I get 5.1V—which is exactly what the power supply is giving out. If I use the bridge and measure the same pin 3 I get 4.9v. If I measure pin 1—the other side of the bridge and labelled 'B' in my photo—I get 4.81v.

4.81v is what I'm getting at the voltage pins on the CPU, RAM etc.

Could CP001 be the cause of the voltage drop? Or is the fault elsewhere in the circuit and I'm just detecting this here when I join the whole circuit?

https://photos.google.com/photo/AF1QipNkYo5-XwqcN4PLECovWXYadnTNVK4jPDkNl3M5

[Bryce]

No, the drop from 5.1V to 4.8/4.9V is what would be expected because a load (the CPC circuit) has been added.
Can't check the photo as it brings me to a login page (and I don't have an account).

Bryce.

[jarcher1980]

No, the drop from 5.1V to 4.8/4.9V is what would be expected because a load (the CPC circuit) has been added.
Can't check the photo as it brings me to a login page (and I don't have an account).

Bryce.

Thanks for confirming. Perhaps voltage around the PCB isn't an issue then. It must be something else. If it's not the RAM, then perhaps it's one of the other ICs. 

[Bryce]

It could still be many things. Where are you located?

Bryce.

[jarcher1980]

It could still be many things. Where are you located?

Bryce.

Yorkshire, UK. 

[Bryce]

Ok, then sending it to me isn't an option, but I can recommend someone in the UK that can fix it if you don't have any success.

Bryce.

[jarcher1980]

Ok, then sending it to me isn't an option, but I can recommend someone in the UK that can fix it if you don't have any success.

Bryce.

Good to know Bryce. Many thanks.

[jarcher1980]

Ok, then sending it to me isn't an option, but I can recommend someone in the UK that can fix it if you don't have any success.

Bryce.

Hello. Still attempting to fix! :-)

I've been testing continuity around the memory modules—mainly between each module's D0 and D1 (pins 14 and 2 respectively) and the D0-D7 of the Gate Array, along with lines to IC115 and IC114. (Marked on a schematic here.) Everything checks out fine as far as I can tell.

I noted that GadgetUK had a similar fault with a garbage screen in this Youtube
video. He identified IC112 as the issue. I decided to get a logic probe and thought it worthwhile checking the same chip. You can see my results at this link. Each of the 4 OR gates was showing a HI output. I think this is correct, but it would be interesting to hear your thoughts.

If IC112 isn't the issue, are there any other ICs you think it would be worth checking next?

Thanks again!

[Nworc]

Hi, just a few thoughts from my side:

A logic probe is an interesting instrument to test a hand crafted circuit, e.g. circuits that operate at low speed. I'm not so persuaded that this will be of much help in your hands in order to hunt down the error in a circuitry that runs at 4MHz.

I would invest the money in a few cheap chips that you find on this board, and try the piggy-back trick shown here instead (one chip after the other, not all at the same time).

Another thing is to closely observe the picture the CPC is still producing, as it tells you something. It is not a good idea to refer to a picture you found on the internet saying it looks like this. If I were you, I would produce a good picture showing how it looks like in your situation. Everyting else is half hearted.

Referring to the pictures you refer to: obviously the clock is running, CRTC and Gate Array are also functional, otherwise you wouldn't see the pattern on the screen. The CPU might be okay, but apparently it crashed. There seems to be a problem with reading out the RAM, which caused the CPU to crash but which also produces the noise you can see on the screen. Try to do something which reduces this noise. As soon as it's gone, I would expect the machine to revive.

Also, people that connect an LCD monitor to the CPC obviously use a power supply not designed for it. Can be that it's a good one, but can also be the opposite, can be the power is somehow distorted and you can't see that with a simple volt meter. I would try a different power supply to double check. Alternatively, test with 4 NiMh batteries in series (at 1.2-1.3V each), this gives a very clean linear power (double check the polarity!).

[jarcher1980]

Hi, just a few thoughts from my side:

A logic probe is an interesting instrument to test a hand crafted circuit, e.g. circuits that operate at low speed. I'm not so persuaded that this will be of much help in your hands in order to hunt down the error in a circuitry that runs at 4MHz.

I would invest the money in a few cheap chips that you find on this board, and try the piggy-back trick shown here instead (one chip after the other, not all at the same time).

Another thing is to closely observe the picture the CPC is still producing, as it tells you something. It is not a good idea to refer to a picture you found on the internet saying it looks like this. If I were you, I would produce a good picture showing how it looks like in your situation. Everything else is half hearted.

Referring to the pictures you refer to: obviously the clock is running, CRTC and Gate Array are also functional, otherwise you wouldn't see the pattern on the screen. The CPU might be okay, but apparently it crashed. There seems to be a problem with reading out the RAM, which caused the CPU to crash but which also produces the noise you can see on the screen. Try to do something which reduces this noise. As soon as it's gone, I would expect the machine to revive.

Also, people that connect an LCD monitor to the CPC obviously use a power supply not designed for it. Can be that it's a good one, but can also be the opposite, can be the power is somehow distorted and you can't see that with a simple volt meter. I would try a different power supply to double check. Alternatively, test with 4 NiMh batteries in series (at 1.2-1.3V each), this gives a very clean linear power (double check the polarity!).

Hi Nworc. Thanks for taking the time to reply. :-)

I did post some images of the screens that I'm getting in earlier messages in this thread, but I've just taken some new ones. You can see them at this link. If you can point to any likely tracks/ICs that may be causing this issue, that would be great.

I did swap the CPU with one of my spare Spectrum ones and I got the same screens so I can probably rule of the CPU. I don't have any spare gate arrays so if it's that I guess it's game over. All the memory ICs have been checked with a tester a few times and they seem fine.

For some reason, under load, the chips are getting around 4.6 - 4.8v—which I understand can cause issues below 4.75v. I guess there must be a short somewhere. Tracks all look fine so my guess is it's an IC.

I've tried two different modern replacement 464 power supplies as well as an Amstrad CTM 644 monitor.

During today's test, aside from the very hot gate array I did note that IC108 was getting quite warm. All other ICs seem cool.

[Nworc]

IC108, the CRTC is innocent, it's apparently doing it's job.

Well, the pictures are not very sharp, but good enough to see that the vertical stripes seem to be on byte level, meaning it's always the same bit / set of bits which is returning an unexpected value.

Now following the idea, that it's always the same bits which are affected. The point I see is, if the Gate Array is accessing the RAM to read the gfx data, the only chips which are connected to Data Out of the RAM chips are the Gate Array and the Data Latch/Buffer IC114 and IC115. Apparently the Gate Array is receiving wrong data since this is what we see on the screen, so either it's having an internal problem, or one of the other mentioned chips (the data latch/buffer) puts a load on the Data Out lines interfering with the RAM.

In case it's not that, I'm also not convinced that the voltage is at the correct level at all RAM chips at all the time, for whatever reason.

I think that's all I can say.

[jarcher1980]

IC108, the CRTC is innocent, it's apparently doing it's job.

Well, the pictures are not very sharp, but good enough to see that the vertical stripes seem to be on byte level, meaning it's always the same bit / set of bits which is returning an unexpected value.

Now following the idea, that it's always the same bits which are affected. The point I see is, if the Gate Array is accessing the RAM to read the gfx data, the only chips which are connected to Data Out of the RAM chips are the Gate Array and the Data Latch/Buffer IC114 and IC115. Apparently the Gate Array is receiving wrong data since this is what we see on the screen, so either it's having an internal problem, or one of the other mentioned chips (the data latch/buffer) puts a load on the Data Out lines interfering with the RAM.

In case it's not that, I'm also not convinced that the voltage is at the correct level at all RAM chips at all the time, for whatever reason.

I think that's all I can say.

Thanks for that.
 
I suspected the CPU->RAM->BUFFERS->GATE ARRAY circuitry so I checked for continuity between them. You can see my checks here. All those in red were checked for continuity and they're fine.

Do you think IC115 and IC114 may be faulty, or even IC112? Do you think it's worth purchasing some replacements? Thanks again.

[Nworc]

Do you think IC115 and IC114 may be faulty, or even IC112? Do you think it's worth purchasing some replacements? Thanks again.

I can't guarantee a return on investment if you spend a pound on each of these. It can always be something you don't expect. I would resolder all chips in question before I would order them, just to exclude that kind of an issue. BTW you can't detect weak solder joints using a multimeter. Okay, if the joint is so weak that the component is falling out, you can detect it

[eto]

how/when do the colours change?

[jarcher1980]

how/when do the colours change?

It seems random whenever I turn it on. Occasionally, I get a completely grey center instead of coloured garbage. It's weird. 

[jarcher1980]

Do you think IC115 and IC114 may be faulty, or even IC112? Do you think it's worth purchasing some replacements? Thanks again.

I can't guarantee a return on investment if you spend a pound on each of these. It can always be something you don't expect. I would resolder all chips in question before I would order them, just to exclude that kind of an issue. BTW you can't detect weak solder joints using a multimeter. Okay, if the joint is so weak that the component is falling out, you can detect it

Good point. I'll solder in some new sockets and retry the same chips first before ordering new ones. Thanks!

[Nworc]

I would not add sockets for chips which aren't socketed. I had the "bright" idea once to socket RAM chips in one of my computers, and problems started with that. In the end I threw out the sockets. By resoldering I just meant: heating up the solder joints and applying a very tiny bit of fresh solder (the one with flowing agent inside) on each joint.

[jarcher1980]

I would not add sockets for chips which aren't socketed. I had the "bright" idea once to socket RAM chips in one of my computers, and problems started with that. In the end I threw out the sockets. By resoldering I just meant: heating up the solder joints and applying a very tiny bit of fresh solder (the one with flowing agent inside) on each joint.

Are okay. Good call. Thanks for clarifying.

[eto]

I would not add sockets for chips which aren't socketed. I had the "bright" idea once to socket RAM chips in one of my computers, and problems started with that. In the end I threw out the sockets. 

That's the opposite of what I heard as recommendations. It's usually said to always do sockets e.g. to avoid desoldering again if the new chip is also faulty.

[Audronic]

I would not add sockets for chips which aren't socketed. I had the "bright" idea once to socket RAM chips in one of my computers, and problems started with that. In the end I threw out the sockets.

Wrong !

As a techno with 40 years of electronic Maintenance the above comments are more about the installation of the sockets than the function of sockets ?
Please put in sockets it helps with further repairs and diagnostics.
Keep Safe
Ray

[eto]

I saw another discussion on this forum where the CRTC was broken and threw wrong addresses at the bus. It still produced an image, but somehow garbled. Not sure if that could be an option here, as it's such a weird, logical pattern (as mentioned be gerald), but if the CRTC or something else throws complete garbage at the address bus, then the CPC could still work, we would just see the data at the resulting locations.

Maybe a bit simple, but  have you tried to press DEL and check if you hear a beep? (volume level high of course).

[jarcher1980]

I saw another discussion on this forum where the CRTC was broken and threw wrong addresses at the bus. It still produced an image, but somehow garbled. Not sure if that could be an option here, as it's such a weird, logical pattern (as mentioned be gerald), but if the CRTC or something else throws complete garbage at the address bus, then the CPC could still work, we would just see the data at the resulting locations.

Maybe a bit simple, but  have you tried to press DEL and check if you hear a beep? (volume level high of course).

Thanks for this. I'm at a bit of a loss as to what to test next to be honest—short of just buying ramdom ICs and seeing if they work. :-)

When I press the DEL key I don't hear anything.

I did notice today that when powering up two or three times, the forwards and backwards buttons on the tape deck function as they should. However, the play button sometimes works and sometimes doesn't. Can anyone think of a reason why that would be? Would fluctuating power, or insufficient power, produce this behaviour? 

I'm personally thinking there is a power problem with this CPC but I've ruled out the power supplies; I've tried three. When the machine is not switched on power at the power connector is 5.1 volts but when switched on it drops to between 4.7 and 4.9; it seems to vary. Dropping would seem to be normal under load, but it seems on average to hover around the 4.75 mark and I know anything below this can cause problems.