Faulty CPC 464 troubleshooting

[Liartes]

Hi all,


A new-old CPC 464 as arrived and this one is quite ill : No video signal, no beep.
The keyboard and tape deck are working allright when plugged on another motherboard.

After cleaning the power plug I get a little voltage drop from 5V to 4.8V. The board draw 760Ma when powered up without the tape deck.
Nothing suspicious at first glance, except the transistor on the photo : I get a regular voltage pulse from 0V to 3,5V on the emitter pin. I compared to a working CPC 464 board, which have a steady 1.6V for this same measure.


Beside this transistor, no IC get hot except the gate-array, everything seems powered but a bit low on voltage.


I'm planning on replacing this transistor C1815, as it is in the video output part, I hope this will suffice to get a video signal but it won't fix the "no beep" part. What should I check next ? Does someone here can provide faulty IC replacements ?

[Bryce]

Don't bother swapping the transistor just yet. It's unlikely to be broken. Do you have another Gate Array that you can try? This is what has most likely failed.

Bryce.

[Liartes]

Thank you for the hint Bryce, can I swap a rev2 motherboard GateArray with a GateArray from a rev1 motherboard ? To be more specific, if they are both located on the bottom slot of their motherboard, are they swappable ?

[Bryce]

If they have the same part No. you can swap them. ie: 40007 with 40007, or 40010 with 40010.

Bryce.

[Liartes]

Surprise under the heatsink: The GateArray is a 40008 socketed on the 40007 slot, as documented here https://cpcrulez.fr/codingBOOK_weka_02322.htm this version is not pin compatible with the 40007. does a 40008 would work in the 40010 slot then ?

[gerald]

Surprise under the heatsink: The GateArray is a 40008 socketed on the 40007 slot, as documented here https://cpcrulez.fr/codingBOOK_weka_02322.htm this version is not pin compatible with the 40007. does a 40008 would work in the 40010 slot then ?

Except that the WEKA doc is wrong : 40007 and 40008 are footprint compatible.
I've never seen a 664 with anything else than a 40010 while they say that the 40008 was introduced with the 664.
See also that 464 with a 40008 and note on which socket it is mounted : https://www.cpcwiki.eu/index.php/File:CPC464_PCB_Top_(Z70200_MC0002D)_GA40008.jpg

[Liartes]

Yes this is this exact version of the motherboard.


Edit : I forgot to power my scart cable, I need to test swap again. Tall key cpc working fine 


Edit 2: GA 40008 is working fine on another motherboard. I am back to the suspicious transistor, waiting for supply.

[gerald]

Edit 2: GA 40008 is working fine on another motherboard. I am back to the suspicious transistor, waiting for supply.

That transistor is responsible for the monochrome signal generation.
Until you get a beep out of the speaker, just do as Bryce said, forget it 

If you don't have any video, usual suspect are Z80 / ROM / CRTC / GA
You already tested the GA.

[Liartes]

Indeed, after reading years of archives on this forum and others, I will do my best to follow Bryce advices


I'll try swapping Z80 next then.

[Liartes]

Z80 swap done, this IC is fried. I will keep you updated when I get a replacement part. Thanks again for your help in the troubleshooting process.


For now this cpc 464 will take a good bath !

[Liartes]

I teared down the cpc 464 for a good clean and while removing the tape deck, these pieces fall. Do you have any clue where they come from ?
Still waiting for the Z80 IC but regarding the general state after a good clean, I have very little hope to see it working again.

[gerald]

I teared down the cpc 464 for a good clean and while removing the tape deck, these pieces fall. Do you have any clue where they come from ?
Still waiting for the Z80 IC but regarding the general state after a good clean, I have very little hope to see it working again.

That's part of the tape deck pause key latch mechanism.
Part 21 on page 5 of the 464 service manual. There is also a small spring.

[Liartes]

There is also a small spring.

Gerald you saved this spring just in time before it get lost forever !

[Liartes]

Z80 replaced : The cpc 464 is back to life 


I replaced the broken tape counter reset button with a 3D printed part, it is working allright.
I am impressed there is nothing else fried as I suspect this computer was water damaged. Keyboard springs were very rusty, some elements in the case were rusty too.



Thanks again Gerald and Bryce for your support !

[dragon]

Except that the WEKA doc is wrong : 40007 and 40008 are footprint compatible.
I've never seen a 664 with anything else than a 40010 while they say that the 40008 was introduced with the 664.
See also that 464 with a 40008 and note on which socket it is mounted : https://www.cpcwiki.eu/index.php/File:CPC464_PCB_Top_(Z70200_MC0002D)_GA40008.jpg

I think, they refer in the sense of" we introduce it in the time when 664 was released". In the 464 board manufactured at time when 664 was released.


Litte offtopic at finish you have documented the changes between 40007/8/10a/10b?

[gerald]

I think, they refer in the sense of" we introduce it in the time when 664 was released". In the 464 board manufactured at time when 664 was released.

Well, they clearly introduce the 40008 as an improved version of the 40007 made for the 664. "La version 40008, utilisée sur le CPC-664 est la version améliorée du 40007 (pour le CPC-464)"

Litte offtopic at finish you have documented the changes between 40007/8/10a/10b?

I did not do the reverse engineering of the 40007/40008, but the die pictures are strictly identical from a mask point of view. So either the decaped 40008 was a re-marked 40007, or the 40008 is only an process evolution of the 40007.
40010 version are, from a quick check, only a re-implementation of the same logic on a smaller (and therefore cheaper) gate array.

[dragon]

Well, they clearly introduce the 40008 as an improved version of the 40007 made for the 664. "La version 40008, utilisée sur le CPC-664 est la version améliorée du 40007 (pour le CPC-464)"

I did not do the reverse engineering of the 40007/40008, but the die pictures are strictly identical from a mask point of view. So either the decaped 40008 was a re-marked 40007, or the 40008 is only an process evolution of the 40007.
40010 version are, from a quick check, only a re-implementation of the same logic on a smaller (and therefore cheaper) gate array.

But the 664 pcb are designed only with 40010 in mind. They not have  two gate array slot. As the first 6128 made.


I read in another site that the dual gate array pcbs was introduced a cause of the 40010. But i don't know if the last batch of cpc 464  made in factory have it or not. Indicating if the 40010 it's most early that the 664...




Edit i found the info. The first record of the mc0001A(first with double gate array slot) was k32-4X October 1984. And k31-4y November 84


And one 464 with a k31-51 mc002b have 40010 installed.


Assembly of 664 begin in k31-54 and k32-53. Abril 85 and March 85.


So if  firsts cpc 6128 are double slot gate array and begin assembly after stopped 664 production in June.

Its little rare 664 is only single 40010 gate array and don"t support 40007/8 yeah.


And what's mean the 40007 subfix -X - 7 - 4 etc. 40007 subrevisions?.
...




No, remarked not. At simply view they change things. They produce less heat.  And at least the draw are different in some zones but same [size=78%] in others. [/size]

[gerald]

No, remarked not. At simply view they change things. They produce less heat.  And at least the draw are different in some zones but same in others.

Attached is a picture of the 40007/40008 die photo differences.
These are mainly :
- scratches on the die
- dirt remaining from decapping (mainly on the 40007)
- slight offset due to stitching errors.

The metal layer is clearly identical on both die, and this is the only thing that customer can customise in a GA.
The only explanation for a lower consumtion is a change in the manufacturing process.

[dragon]

Attached is a picture of the 40007/40008 die photo differences. These are mainly : - scratches on the die - dirt remaining from decapping (mainly on the 40007) - slight offset due to stitching errors. The metal layer is clearly identical on both die, and this is the only thing that customer can customise in a GA. The only explanation for a lower consumtion is a change in the manufacturing process.

O.k now I have a theroy.


I think the 40007 works but it not work as it should work. But the problem is not the metal layer, was ferranti technology. And they need compensate outside in the motherboard the deficiece.


Then amstrad move to sgs and begin designing the 40010. But in the mean time ferranti(because that hace recieved the amstrad original amstrad paid). Try fix it. And that generate the - 7 - X. And the final version was the 40008.


Then amstrad   continue working with ferranti and the 40008 definitive edition transform in the definítive 40007(without any - x)


But they broke contact with sgs when they begin the design of the cost down...


And made the special 6128 board that support ferranti.


Why I tell this. Because the numbers in the 40007 are incremental. And the manufacture date of the ulas can be read year/week.


40007-4 year 84 week 24
40007-4 year 84 week 28
40007-X year 84 week 36
40007-4 year 84 week 44( I think the result of the x version don't convence and they return to the previous version).
40007-4 year 85 week 2
40007-4 year 85 week 08
40008    year 85 week 08!
40007-4 year 85 week 12
40007-4 year 85 week 16
40008    year 85 week 48
40007-4 year 86 week 17
40008    year 86 week  08
40008    year 86 week 29 not 100: sure picture not good.
40007   year 87 week 22(without sufix)

https://retrohax.net/amstrad-cpc-464-the-turmoil-part-two/

40007   year 88  week 04 (year of the cost down). Without sufix.




another option If the die= because we have decapped the fixed versions?
.


I'n one  service manual they call 40007 mk-4..

[gerald]

If only we had the date code of the decapped IC !Everything else is speculation.And for using the 40008 design in a 40007 marked chip, that would be insane.

[gerald]

Interestingly, the 40007 is referred as a 20RA023 (2000 gates), while the die picture of both 40007/40008 are 16RA032 (1600 gates).

It mean, at least, that amstrad moved to a smaller array. But was it during the 40007 development, during 40007 shelf life (like 40010 AA36/AA37) or as 40008 ?
@robcfg : did anyone took a picture of the package before decapping ?

[robcfg]

I don't think we did any picture of the ICs before sending them, and if we did (there's always the possibility) it will take some time to look for pictures 4-5 years old...

[dragon]

Interestingly, the 40007 is referred as a 20RA023 (2000 gates), while the die picture of both 40007/40008 are 16RA032 (1600 gates).

It mean, at least, that amstrad moved to a smaller array. But was it during the 40007 development, during 40007 shelf life (like 40010 AA36/AA37) or as 40008 ?
@robcfg : did anyone took a picture of the package before decapping ?

Yeah I think there something that change between - sufix version and 08/07 witouth - sufix.

I only read 20ra043(service manuals and 16ra023 decapped pictures).

Take a look at these schneider service manual.


https://acpc.me/ACME/DOCS_TECHNIQUES/SERVICE_MANUALS/CPC464_neue_Platinenversion_SCHNEIDER_Service_Manual%5BENG%5D%5BGER%5D%28acme%29.pdf

They call 40007-4 "mk4" 20ra043 that board was scanned and mount 40007-4.


And there is another insane note in that service manual . They tell in other board schematic part 40008 change to part 40008/A.  With date 16.5
85 Wtf!!!

And 40007 part change to 40010.


MC0002B

I remember another service manual that.
Mention 40007-x I m in search of it.



Found: mc009a there is a note tell 40007-X or 400010 should be installed.


http://www.cpcwiki.eu/manuals/service.manual.cpc6128.ctm644.gt65.zip

the pcb 00057 is in the adment manual it tells 20ra043 40007. 




But... I have compared the gate array of the adment manual with the schneider mk4  manual.


I'n the schneider just up the pin 1 move the circuit to the right part.

And the 002b lost pin 35.


If when service manual say "type" it mean any ula type of 5000 series ula. But the first was 20ra043 and they use it as a example in the manual services.




In the other hand I have found the origin of the history of 40008 in 664. The origin was amstrad action number 39 December 1988


They write this:


"The video gate array is a more exclu-
sive chip, designed by Amstrad just for the
CPC's. Three versions of this were made:
the 40007 was included in most. 464s. and
il had to have a metal plate clipped on top
to dissipitate heal. The 40000 for the 664
v/as similar, but just warmed up. The
40010 6128 for the use of. barely changes
temperature. "