News:

Printed Amstrad Addict magazine announced, check it out here!

Main Menu

CPC 6128 - white screen, black border

Started by mmldrm, 22:02, 31 March 24

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

McArti0

n is the same as a line above the text.
_____
CPU = nCPU -  Pin14 GA
CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

McArti0

Quote from: mmldrm on 20:34, 06 April 24Select Expand
pin7 (2.18v) - pin17 (3.63v)
pin8 (1.67v) - pin12 (3.73v)
pin9 (0.66v) - pin19 (0.75v)
Old measure pin17 PAL CAS0 work.
CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

Bryce

Quote from: mmldrm on 20:34, 07 January 25
Quote from: McArti0 on 18:17, 07 January 25Either PAL/HAL does not work or nCPU does not reach.
Where does nCPU originate? I can't find any references to it. Only on the PAL pin-outs.

I thought I would try to trace the origin (still learning, here) and try to measure the value before it reaches the PAL. (Also learning to call it PAL and not HAL.)

Edit: I see it on the GA, in the circuit diagram. But I can't tell how that relates to the PAL - difficult to tell from the diagram.

All "n" signals originate from the GA. That's  what the "n" means.

Bryce.

mmldrm

Thanks. I'm slowly learning the notation as we go. (I should look up a reference to minimise the simplest questions.)

nCPU is a clean signal, both on GA pin 11 and PAL pin 14.

So it's looking like (at a minimum) the PAL is toast?

mmldrm

Quote from: McArti0 on 23:09, 07 January 25
Quote from: mmldrm on 20:34, 06 April 24Select Expand
pin7 (2.18v) - pin17 (3.63v)
pin8 (1.67v) - pin12 (3.73v)
pin9 (0.66v) - pin19 (0.75v)
Old measure pin17 PAL CAS0 work.
You are right. I just repeated the above measurements with my multimeter and they are very different. :(

eto

Quote from: mmldrm on 09:12, 08 January 25Thanks. I'm slowly learning the notation as we go. (I should look up a reference to minimise the simplest questions.)
That unfortunately doesn't always help as it's not used everywhere properly.

E.g. the CPU signal is shown as nCPU on the GateArray but is then called /CPU on the PAL. And on the Wiki pages about the Gate Array signals nCAS is documented as /CAS. (https://www.cpcwiki.eu/index.php/Gate_Array_and_ASIC_Pin-Outs) In some other docs you will find the CAS0 being called nCAS0.

What usually helps me most is to look up the schematics, then follow the lanes. That also helps to better understand how everything works together. https://www.cpcwiki.eu/index.php/Schematics

McArti0

Quote from: mmldrm on 11:08, 08 January 25You are right. I just repeated the above measurements with my multimeter and they are very different. :(
But still bad.

ps. how many now?
CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

SerErris

Quote from: eto on 11:15, 08 January 25
Quote from: mmldrm on 09:12, 08 January 25Thanks. I'm slowly learning the notation as we go. (I should look up a reference to minimise the simplest questions.)
That unfortunately doesn't always help as it's not used everywhere properly.

E.g. the CPU signal is shown as nCPU on the GateArray but is then called /CPU on the PAL. And on the Wiki pages about the Gate Array signals nCAS is documented as /CAS. (https://www.cpcwiki.eu/index.php/Gate_Array_and_ASIC_Pin-Outs) In some other docs you will find the CAS0 being called nCAS0.

What usually helps me most is to look up the schematics, then follow the lanes. That also helps to better understand how everything works together. https://www.cpcwiki.eu/index.php/Schematics

That is actually all the same. nCPU ^CPU /CPU !CPU all means, that this is active low signal - nothing more nothing less. 

And there is a difference between high and low active and you can use the same signal for different purposes. For instance if you have on chip that has for instance chip enable as active low, and the other one active high, you can connect both to the same signal and the will activate at oposite times. 

This is actually something that is beeing used by the CPC. 

nCPU is at the same time the Soundchip clock. That is because it is just a 1mhz clock signal, which is what is required by the sound chip. The beauty of it is, that the sound chip is automatically synced to the 4Mhz clock (exactly 1/4 the speed) and to CRTC and other components. Not that it is really required in this scenario :-)

Proud owner of 2 Schneider CPC 464, 1 Schneider CPC 6128, GT65 and lots of books
Still learning all the details on how things work.

mmldrm

Would you like me to check the values on the PAL now?

mmldrm

Quote from: McArti0 on 17:09, 03 January 25You need to be sure that after inserting the IC or adapter all connections are secure and stable. Sometimes it is thought that it is, but it is not.
Check by oscylloscope signal RAS, CAS, CLK 4 MHz, READY.
On Probe setting 10x. And trigger in oscyllo single.
So, with the PAL adapter in place the readings are much cleaner.

PAL adapter:
CAS1 pin 16 signal - same as CAS pin 16 on GA
CAS0 pin 17 HI

Lower RAM Bank 0:
RAS pin 4 same signal as RAS pin 34 on GA
CAS pin 15 HI

Upper RAM Bank 1:

RAS pin 4 signal- same as RAS pin 34 on GA
CAS pin 15 signal - same as CAS pin 16 on GA


So the CAS and RAS signals are getting to the upper RAM, and the 'PAL adapter' CAS0 is HI.

Screen is still white with black border. But I have not replaced the upper RAM.

Is it possible to use the adapter to force the lower RAM? Those chips were replaced and socketed.

Looks like I can do this by swapping the wires in pins 16 (CAS1) and 17 (CAS0) of the adapter. Seems obvious, but I will wait for the experts to confirm before going rogue!

eto

Quote from: mmldrm on 11:10, 11 January 25Is it possible to use the adapter to force the lower RAM? Those chips were replaced and socketed.
yes. Just make sure nCAS goes to CAS0 and CAS1 is high. 

But don't be surprised if nothing changes: if one of your upper RAM ICs has a short, nothing will change. The data pins are not isolated and if any of them is setting a bit high or low constantly then there is no change. Same if the latch has an issue that isolates the RAM from the bus: nothing will change. 

mmldrm

Quote from: eto on 11:28, 11 January 25yes. Just make sure nCAS goes to CAS0 and CAS1 is high.

But don't be surprised if nothing changes: if one of your upper RAM ICs has a short, nothing will change. The data pins are not isolated and if any of them is setting a bit high or low constantly then there is no change. Same if the latch has an issue that isolates the RAM from the bus: nothing will change.
Thanks for the quick answer. To no-one's surprise, nothing changed. At least I knew in advance to not get my hopes up!

Am I going to have to replace the upper RAM before I can do further tests? (Given that it was £40 to do the first bank, I may have to be brave and replace the upper RAM myself...)

eto

Quote from: mmldrm on 11:43, 11 January 25Am I going to have to replace the upper RAM before I can do further tests?
Probably yes.

I would do two things before:
1) Remove the socketed RAM from bank 0 and try again with the adapter (which connects CAS to CAS1) just to make sure none of the new RAM ICs is fried
2) check for any shorts introduced by the repair guy - you never know if it's done well or not

If that does not show a change, then I personally would probably try the following:

  • Put in the RAM again in bank 0, PAL adapter CAS to CAS0 so it runs on bank 0
  • On bank 1, cut the Dout pin of each RAM IC one after the other and in between check if the CPC then works

If it's really just the Dout then the CPC should start once you have cut the pins of the failing IC(s) and just need to replace that IC. You probably also can reconnect the disconnected pins of the other ICs (at least if you don't cut it too close to the IC package) which is much less work and less risk than replacing all ICs.

To be honest, I never did that myself as I am usually using the Amstrad Diagnostics. And it might also be that it is as likely that an address pin is short, which might have the same effect. Let's ask someone who really knows what he's talking about...  @Bryce  would that idea make sense or is that a stupid idea?


mmldrm

Unfortunately step 1) still resulted in the same white screen with black border.

Quote from: eto on 12:26, 11 January 252) check for any shorts introduced by the repair guy - you never know if it's done well or not
Could you describe that process in more detail? Reading up on it, I understand the principle that there should be high resistance between neighbouring traces/pads, but not sure where is best to measure across in this scenario (checking the new RAM/sockets).


eto

Quote from: mmldrm on 19:12, 16 January 25Unfortunately step 1) still resulted in the same white screen with black border.
Not really a surprise but it was important to test that, so you don't exchange the ICs of bank 1 - and then it's one of the ICs in bank 0.

Quote from: mmldrm on 19:12, 16 January 25Could you describe that process in more detail? Reading up on it, I understand the principle that there should be high resistance between neighbouring traces/pads,

1) download the schematics, and have a look at the part with the RAM

2) check that all pins that should be connected indeed are connected. See attached one example: Pin 10 of all RAM ICs should be connected. With your multimeter you can put one probe to any IC and then check if all other ICs have a connection on pin 10. Finally also check if there is a connection to pin 7 of IC104 (resistance should show 47 Ohms). If any connection fails, there must be fault somewhere. Just an example what happened to me: When removing a RAM IC I accidentally removed a solder pad - and once I inserted the IC socket and didn't recognize that one leg was bent - and then it had no connection.

3) check that there is no connection between pins that should not be connected. As you already checked the correct connections in step 2 it should be sufficient to test
a) on each RAM IC: check all pins - start with one probe on pin 1, then check Pin 2-16 with the other probe, move on to pin 2, check pin 3-16 ... and so on
b) on one bank: check all D-Out pins. start on Bank 0, RAM IC 1, put probe on Pin 14 and check Pin 14 on all RAM IC 2-8 of bank 0 with the other probe, move on to IC 2 and check PIN 14 on RAM IC 3-8
If there is any connection where no connection should be, then a short was introduced. E.g. I once had a tiny piece of wire that was stuck under an IC socket and connected two pins with each other.



mmldrm

Quote from: eto on 20:03, 16 January 252) check that all pins that should be connected indeed are connected. See attached one example: Pin 10 of all RAM ICs should be connected. With your multimeter you can put one probe to any IC and then check if all other ICs have a connection on pin 10. Finally also check if there is a connection to pin 7 of IC104 (resistance should show 47 Ohms). If any connection fails, there must be fault somewhere. Just an example what happened to me: When removing a RAM IC I accidentally removed a solder pad - and once I inserted the IC socket and didn't recognize that one leg was bent - and then it had no connection.

Thanks once again for the very clear instructions. I've done continuity checks on all 8 data pins (5,6,7,9,10,11,12,13) on all RAM ICs, and the appropriate pin on one of the 74LS153. The resistance showed 50-52 Ohms in all cases. I also checked the resistance across one or two of the resistors, and it was also 50 Ohms.

In short, all the connections between RAM ICs, and the 74LS153s, are intact.

Quote from: eto on 20:03, 16 January 253) check that there is no connection between pins that should not be connected. As you already checked the correct connections in step 2 it should be sufficient to test
a) on each RAM IC: check all pins - start with one probe on pin 1, then check Pin 2-16 with the other probe, move on to pin 2, check pin 3-16 ... and so on
Well this was fun!  ??? My eyes did cross a few times. But definitely worthwhile. In short, there are no shorts between pins on any of the RAM ICs.

Quote from: eto on 20:03, 16 January 25b) on one bank: check all D-Out pins. start on Bank 0, RAM IC 1, put probe on Pin 14 and check Pin 14 on all RAM IC 2-8 of bank 0 with the other probe, move on to IC 2 and check PIN 14 on RAM IC 3-8
The connections between pins on the same data line were all intact, and I checked all 8 data pins on each IC. Maybe worth mentioning that on Bank 0 I was checking sockets, as the RAM ICs are currently removed.

Happy to do the same tests with the RAM ICs in place, if that would rule out any disconnect between socket and IC. In fact, the guy that socketed the Bank 0 RAM did say that he thought the sockets were a bit crap. Some RAM ICs don't sit well in the sockets, so it might be worth me doing these tests again with the ICs in place. Even though I am pretty sure that I already checked the signals with a scope. Might be safer if I do that, just to rule things out.

Well, I learned a bit more from this and also got peace of mind that the RAM connections in both banks seems to be in good shape.

Still puzzled that the current situation seems to be:

  • RAM seems to pass all manual tests with multimeter/oscilloscope
  • PAL chip seems screwed
  • PAL override (bank switching trick) doesn't solve the problem

This makes me wonder about the other ICs involved in RAM management (the four 74LS153s) or the data latch / buffers (74LS244 and 74LS373).

Then again, as you know, I have no idea what I'm doing.  ;D I'd love to suggest a Dandanator to the owner of this board, so I can run the diagnostic ROM and he could use it to load games in the future, but I'm still not convinced that this board is working enough for the diagnostic ROM to give correct feedback.

eto

Quote from: mmldrm on 23:34, 19 January 25, but I'm still not convinced that this board is working enough for the diagnostic ROM to give correct feedback.
If you get a white screen with black border it will be enough to run the diagnostics ROM. "white screen with black border" means that ROM can be executed.

Quote from: mmldrm on 23:34, 19 January 25This makes me wonder about the other ICs involved in RAM management (the four 74LS153s) or the data latch / buffers (74LS244 and 74LS373).
Yes, also possible. More likely though that it's one of the RAM ICs of bank 1.

Quote from: mmldrm on 23:34, 19 January 25PAL override (bank switching trick) doesn't solve the problem
E.g.
1) If just one IC of bank 1 is fried in a way that its data out is always high, then also bank 0 cannot work properly as the lanes of the data pins are shared. the HIGH from bank 1 will always set the respective bit to 1 and the CPU will not read correct data.
2) if one of the other ICs is malfunctioning, either the address is not corrcetly calculated or the data out will not be properly passed though to the CPU.

In all cases the observed effect can be the same: white screen + black border.

Bryce

Quote from: mmldrm on 23:34, 19 January 25
Quote from: eto on 20:03, 16 January 252) check that all pins that should be connected indeed are connected. See attached one example: Pin 10 of all RAM ICs should be connected. With your multimeter you can put one probe to any IC and then check if all other ICs have a connection on pin 10. Finally also check if there is a connection to pin 7 of IC104 (resistance should show 47 Ohms). If any connection fails, there must be fault somewhere. Just an example what happened to me: When removing a RAM IC I accidentally removed a solder pad - and once I inserted the IC socket and didn't recognize that one leg was bent - and then it had no connection.

Thanks once again for the very clear instructions. I've done continuity checks on all 8 data pins (5,6,7,9,10,11,12,13) on all RAM ICs, and the appropriate pin on one of the 74LS153. The resistance showed 50-52 Ohms in all cases. I also checked the resistance across one or two of the resistors, and it was also 50 Ohms.

In short, all the connections between RAM ICs, and the 74LS153s, are intact.

Quote from: eto on 20:03, 16 January 253) check that there is no connection between pins that should not be connected. As you already checked the correct connections in step 2 it should be sufficient to test
a) on each RAM IC: check all pins - start with one probe on pin 1, then check Pin 2-16 with the other probe, move on to pin 2, check pin 3-16 ... and so on
Well this was fun!  ??? My eyes did cross a few times. But definitely worthwhile. In short, there are no shorts between pins on any of the RAM ICs.

Quote from: eto on 20:03, 16 January 25b) on one bank: check all D-Out pins. start on Bank 0, RAM IC 1, put probe on Pin 14 and check Pin 14 on all RAM IC 2-8 of bank 0 with the other probe, move on to IC 2 and check PIN 14 on RAM IC 3-8
The connections between pins on the same data line were all intact, and I checked all 8 data pins on each IC. Maybe worth mentioning that on Bank 0 I was checking sockets, as the RAM ICs are currently removed.

Happy to do the same tests with the RAM ICs in place, if that would rule out any disconnect between socket and IC. In fact, the guy that socketed the Bank 0 RAM did say that he thought the sockets were a bit crap. Some RAM ICs don't sit well in the sockets, so it might be worth me doing these tests again with the ICs in place. Even though I am pretty sure that I already checked the signals with a scope. Might be safer if I do that, just to rule things out.

Well, I learned a bit more from this and also got peace of mind that the RAM connections in both banks seems to be in good shape.

Still puzzled that the current situation seems to be:

  • RAM seems to pass all manual tests with multimeter/oscilloscope
  • PAL chip seems screwed
  • PAL override (bank switching trick) doesn't solve the problem

This makes me wonder about the other ICs involved in RAM management (the four 74LS153s) or the data latch / buffers (74LS244 and 74LS373).

Then again, as you know, I have no idea what I'm doing.  ;D I'd love to suggest a Dandanator to the owner of this board, so I can run the diagnostic ROM and he could use it to load games in the future, but I'm still not convinced that this board is working enough for the diagnostic ROM to give correct feedback.

The RAM tests you've done aren't conclusive. I suspect that there are one or more failed RAM chips.

Bryce.

McArti0

#318
So once again, only more consciously...
PAL wire
RAMDIS set Hi
ROMDIS set Hi

RESET one touch Lo

A short program is executed

38h read RAM FFh
write RAM 00h -Address-1
write RAM 39h -Address-2

out RAM D7-D0 0011-1001

CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

mmldrm

Quote from: McArti0 on 19:02, 20 January 25So once again, only more consciously...
PAL wire
RAMDIS set Hi
ROMDIS set Hi
Where is best to set RAMDIS/ROMDIS to high? It is not ideal to fix pins to the expansion port with tape.

Can I set RAMDIS high on pin 4 of the PAL socket, even with the wires? I also see it on pin 10 of IC112 (74LS32)... would that work?

Can I set ROMDIS high on pin 22 of IC103 (40025 ROM)?

McArti0

CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

Audronic

@mmldrm 

Where in the world are you Located ?

Keep Safe. Ray
Procrastinators Unite,
If it Ain't Broke PLEASE Don't Fix it.
I keep telling you I am Not Pedantic.

mmldrm


mmldrm

Quote from: McArti0 on 18:05, 21 January 25https://www.cpcwiki.eu/forum/hardware-related/cpc-6128-white-screen-black-border/msg238468/#msg238468

nothing new
Thanks. I thought I had better connection points but couldn't remember where. Your memory is better than mine (and my CPC's).

(Is there another for RAMDIS?)

McArti0

#324
R124 but ROMDIS is enough as I checked mine.
Of course you won't see vertical bars.
CPC 6128, Whole 6128 and Only 6128, with .....
NewPAL v3 for use all 128kB RAM by CRTC as VRAM
One chip drver for 512kB extRAM 6128
TYPICAL :) TV Funai 22FL532/10 with VGA-RGB-in.

Powered by SMFPacks Menu Editor Mod