News:

Printed Amstrad Addict magazine announced, check it out here!

Main Menu

Quick and dirty RAM test for CPC

Started by gerald, 17:07, 12 August 15

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

gerald

Quote from: jonnixx on 20:25, 18 September 18
Hi,


I have two problems.
The CPC 6128 with standard RAM.
Original problem:
If the machine is cold (room temperature) then i can load the R-type Reloaded from floppy (3") and HxC and run the program.
... and then not cold the machine (after 3-4 minutes) press reset and can not load the game. (Freeze the program during loading)


Second problem:
The dirty ram test is failed. (attached picture)


I have replaced:
- 8 RAM modules (new rams: TMS4164-12NL)
- 74LS244 and 74LS373
Could you please help me?


Jonni



Did you run the RAM test with a cold or warm CPC.
Also, there are other ICs involved in the RAM access. Mainly the gate array and the address multiplexors (4 74LS153 : IC104 IC105 IC109 and IC113)
The gate array being on socket it may suffer a bad connection when getting warm. Can you try re-seating it ?

jonnixx

Did you run the RAM test with a cold or warm CPC.Yes. Both.
i am waiting for 74LS153... next week, maybe...
Can you try re-seating it ?Yes. Tomorrow.
Is it exist any test program for GA?
Many thanks.



Audronic

Please send a Photo of the RAM test when the machine is COLD
Thanks    Ray
Procrastinators Unite,
If it Ain't Broke PLEASE Don't Fix it.
I keep telling you I am Not Pedantic.
As I Live " Down Under " I Take my Gravity Tablets and Wear my Magnetic Boots to Keep me from Falling off.

jonnixx

Hi Ray,
The ram test is same (failed) in both.Thanks.


jonnixx

Hi,
- GA re-seating is done- replaced 4 x 74LS153- replaced 2 x 74LS32- replaced 765 FDC
Memory test still failed as before.
any idea?  (GA?)

Regards,Jonni

gerald

Quote from: jonnixx on 16:19, 20 September 18
Hi,
- GA re-seating is done- replaced 4 x 74LS153- replaced 2 x 74LS32- replaced 765 FDC
Memory test still failed as before.
any idea?  (GA?)

Regards,Jonni
Without looking at the RAM without a logic analyser, your last bet is indeed the GA itself.

jonnixx

#81
Without looking at the RAM without a logic analyser, your last bet is indeed the GA itself.
I have oscilloscope (Tektronix) and multimeter and any other but i do not have logic analyzer.I am looking for a 8 channels logic analyzer ... now.
I will come back if i have a new logic analyzer.
Regards,
Jonni


LambdaMikel

#82
A logic analyzer and oscilloscope is unnecessary for 99% of the repairs.
I suggest you get the CPC schematics and start doing some continuity checks first.
Next, get a 10 $ logic probe and check if the chips are getting their select signals, if
their outputs are active, etc. That way you can track down non-functioning chips and replace them. 

I have done a couple of quite challenging repairs, and none of them required an oscilloscope or logic analyzer. Unless you are very good at digital electronics, you will have a hard time understanding what's going on anyway.

The suggested method allowed me, for example, to perform this repair, so I know what I am talking about (that thing had about 20 defects - including broken traces, defect RAM, defect glue support chips, ...): 

https://stardot.org.uk/forums/viewtopic.php?f=3&t=15501&start=30&hilit=repair+problem

In addition, you will need people that can guide you through the checking process. Probably the most important thing.

Good luck! 

gerald

For those interested, I've added a zip file with the source code of the tool to the 1st post.You will need SDCC to compile it.

gotcha

I just noticed this post  :o This is very interesting, thanks @gerald !

Yesterday, I used one of my 6128 as a test bench for testing the RAM of a old PC (they use the same RAM) https://forum.system-cfg.com/viewtopic.php?f=18&t=9388&start=30#p144477
I used a memchecker program (in German) that I had found some time ago. This solution worked because I believe I've put the potentially non working memory on the second 64K bank only. 
Can anybody confirm that IC119->IC126 is the second bank ?

But in case of a CPC with a failing first memory bank, the solution I was using until now (also given by @gerald ) was to switch banks and/or disable a bank by removing the PAL and re-wiring the PAL socket.
This software solution looks ideal. I'm going to give it a try with the X-MEM  :D

gotcha

#85
Quote from: LambdaMikel on 17:24, 21 September 18
A logic analyzer and oscilloscope is unnecessary for 99% of the repairs.
I would say, yes and no.

Most of the issues on CPCs are indeed simply on-off switch to clean, keyboard to clean etc...

But when it comes to a problem on the motherboard, the scope is IMO more than useful, just to check clocks and some other signals like the video output, activity on busses etc....
For some problems, the logic analyzer is also needed. As an example, I've got a failing CPC and I didn't want to change blindly chips:

       
  • With the scope, I detected that the synchro signal was not generated by the CRTC. Without a scope, you can't see this
  • But the scope is not enough. This problem can come from many parts : failing CRTC chip ? Not well programmed CRTC due to a failing memory ? bus ? Z80 ?
    I then had to use a logic analyzer to check that the CRTC initialization sequence (at boot) was correctly coming on the data bus on the input of the CRTC and determine that the CRTC was faulty.
Without logic analyzer, i could not have located the actual problem accurately.

LambdaMikel

#86
Quote from: gotcha on 09:23, 02 November 18

But when it comes to a problem on the motherboard, the scope is IMO more than useful, just to check clocks and some other signals like the video output, activity on busses etc....
For some problems, the logic analyzer is also needed. As an example, I've got a failing CPC and I didn't want to change blindly chips:

       
  • With the scope, I detected that the synchro signal was not generated by the CRTC. Without a scope, you can't see this

My logic probe has a "pulse mode". With that you can check if the output / chip is alive and generating a non-steady signal. I used that mode to check RAM chips (chip select, RAS, CAS, inputs, output).

A logic probe is onle 30 €, whereas a scope that can be used for Mhz digital signals is a couple of hundred €,  even the cheap ones (there are little DSO pocket scopes for about 100 $, but they are toys and don't have the bandwith for digital signals, they are only useful for audio / analog really)

I am not saying that a scope and/or logical analyzer is not useful. But I would not advise people to get one if 99 % of the situations (including some of the ones that you have just described) can be analyzed perfectly fine with a logic probe.

And from a time management point of view, before you have figured out how to hook up and read and understand the databus and CRTC signals, you would have already replaced the CRTC with a fresh one. The good thing about the CPC is that it only has a couple of big chips... in principle, I am not a fan of blind repairs, but for the CPC it is possible. Buying a new CRTC is not only more cost efficient than buying an oscilloscope or logic analyzer, but also much more time efficient.

All this is of course only advice for people who are not electronic geeks and don't already have an oscilloscope and logic analyzer. So I would not suggest to people to buy such equipment in order only to repair a CPC (unless they want that equipment anyway of course).


Just my few cents.

Audronic

Hi All


I have just been working on SOME CPC464's and have made a table for the Bits used in GERALD's RamTest


It may be of some use  ?


Thanks.    Ray
Procrastinators Unite,
If it Ain't Broke PLEASE Don't Fix it.
I keep telling you I am Not Pedantic.
As I Live " Down Under " I Take my Gravity Tablets and Wear my Magnetic Boots to Keep me from Falling off.

GUNHED

Great! It would be even greater to know which bit corresponds to which RAM chip.


Just to be clear: It's the case that for every bit of a byte we have one physical RAM chip, right?
So if a bit is 'broken' in one byte it is probably 'broken' in all bytes, right?

http://futureos.de --> Get the revolutionary FutureOS (Update: 2023.11.30)
http://futureos.cpc-live.com/files/LambdaSpeak_RSX_by_TFM.zip --> Get the RSX-ROM for LambdaSpeak :-) (Updated: 2021.12.26)

Audronic

Procrastinators Unite,
If it Ain't Broke PLEASE Don't Fix it.
I keep telling you I am Not Pedantic.
As I Live " Down Under " I Take my Gravity Tablets and Wear my Magnetic Boots to Keep me from Falling off.

GUNHED

http://futureos.de --> Get the revolutionary FutureOS (Update: 2023.11.30)
http://futureos.cpc-live.com/files/LambdaSpeak_RSX_by_TFM.zip --> Get the RSX-ROM for LambdaSpeak :-) (Updated: 2021.12.26)

CanonMan

Quote from: gerald on 17:07, 12 August 15

The program first test the display area (C000-FFFF), should see something like :

The content of the screen will change during the test. So if nothing move, more than the ram may be wrong.
Something else is wrong if you do not see that screen at all.


Is there any reason why it only tests ram from &C000 to &FFFF, rather than the whole 64k?


That's what appears to be happening in the source code.

gerald

Quote from: CanonMan on 22:59, 09 May 19

Is there any reason why it only tests ram from &C000 to &FFFF, rather than the whole 64k?


That's what appears to be happening in the source code.
The way the DRAM is organised makes a 16K test enough to check all memory ROWs.
Screen addresses are then used to have a visual feedback.

CraigsBar

#93
.
IRC:  #Retro4All on Freenode

SpDizzy


Hi guys!!
Does anyone could help with a diagnosis over the RAM test on attached image?
CPC 6128 AZERTY with grey screen on startup.
Several IC's so hot on both banks, reemplaced with new ones, socketed.
On the upper right side of the screen (as shown on the image) there are some blinking pixels on concrete positions.
(Sorry, green screen monitor test)
Thanks so much in advance,

TotO

It looks the 6th RAM IC is bad. (GREEN/RED rasters char-lines border)
"You make one mistake in your life and the internet will never let you live it down" (Keith Goodyer)

SpDizzy

Thanks so much for response @TotO
So it looks like IC132 is the culprit then, right?

llopis

Quote from: TotO on 21:51, 05 October 19
It looks the 6th RAM IC is bad. (GREEN/RED rasters char-lines border)
Wait, are you sure? The bottom row is the LSB, and this looks like the third row to me. So I would say it's D2 that is faulty, so IC129. Or am I totally off?

TotO

I'm not at home. May be @gerald could confirm what RAM IC number match with the 6th char line.
"You make one mistake in your life and the internet will never let you live it down" (Keith Goodyer)

Audronic

@SpDizzy


What is the model number on the motor board ?
Can we have a photo of the Ram area on the motherboard Please to identify IC Numbers


The fault is at Data 2 ( Read from the bottom up ) 0-1-2

Thanks.  Ray
Procrastinators Unite,
If it Ain't Broke PLEASE Don't Fix it.
I keep telling you I am Not Pedantic.
As I Live " Down Under " I Take my Gravity Tablets and Wear my Magnetic Boots to Keep me from Falling off.

Powered by SMFPacks Menu Editor Mod