• Please review our updated Terms and Rules here

Commodore 3016 scrambled screen

Check the address lines from the CPU then the BA (Buffered Address) lines via the address buffers with the NOP generator installed.

Dave
 
Strange, It was working. I had the normal CBM screen. Used pettester to check. Everything was fine.
Placed original rom back and now this:
 

Attachments

  • IMG_7702.jpg
    IMG_7702.jpg
    1.5 MB · Views: 1
Did my PETTESTER give the the correct checksums for the ROMs you have fitted?

The usual problem is fitting the EDIT ROM back in the wrong way round!

That is the screen indicating the 6502 is not running properly. Check CPU pin 7 (SYNC) for activity.

Dave
 
First time the pettester gave everything correct. I will check CPU pin.
Now pettest goes through the screen with everything on G and then the next screen appaer and then it freezes
 

Attachments

  • IMG_7703.jpg
    IMG_7703.jpg
    1,018.6 KB · Views: 3
There is still a fault in the DRAM.

It is good enough to pass the 'simple' tests but then fails on the full memory test.

However, some memory is needed to run the full memory test. A chicken and egg situation I am afraid.

PETTESTER V5 will have a more thorough 'simple' memory test...

Dave
 
I made this memory test system which does get around that chicken & egg thing, but it may not be suited to all PET boards, it works well on the dynamic pet with the 4116 DRAM array, basically it deactivates the lower 1k or 2k of DRAM and electronically inserts a good SRAM so the computer can boot to BASIC and then that can be used to help diagnose troubles by examining the remainder of memory either by using PEEKs & POKES or running short prigrams to fill the memory with various bytes. Depending on the exact PET board it may need a different connector. It also eliminates the requirement of all of the DRAM support circuits from the fault equation, but they still run normally, so they can be checked with the scope to see if they are working properly.

www.worldphaco.com/uploads/DRAM%20MEMORY%20TEST%20SYSTEM%20FOR%20THE%20DYNAMIC%20PET%20COMPUTER.pdf

The first one I made was on a hand wired proto pcb. Then I had a few pcb's made. It wasn't a project to sell, but I have a spare unpopulated pcb or two, I could send one over for $20 if you got stuck.
 
Last edited:
daver2 said:
There is still a fault in the DRAM.

It is good enough to pass the 'simple' tests but then fails on the full memory test.

However, some memory is needed to run the full memory test. A chicken and egg situation I am afraid.

PETTESTER V5 will have a more thorough 'simple' memory test...

Dave
Click to expand...
I will first try V05 then. Where can i find this version?
 
Hugo Holden said:
I made this memory test system which does get around that chicken & egg thing, but it may not be suited to all PET boards, it works well on the dynamic pet with the 4116 DRAM array, basically it deactivates the lower 1k or 2k of DRAM and electronically inserts a good SRAM so the computer can boot to BASIC and then that can be used to help diagnose troubles by examining the remainder of memory either by using PEEKs & POKES or running short prigrams to fill the memory with various bytes. Depending on the exact PET board it may need a different connector. It also eliminates the requirement of all of the DRAM support circuits from the fault equation, but they still run normally, so they can be checked with the scope to see if they are working properly.

www.worldphaco.com/uploads/DRAM%20MEMORY%20TEST%20SYSTEM%20FOR%20THE%20DYNAMIC%20PET%20COMPUTER.pdf

The first one I made was on a hand wired proto pcb. Then I had a few pcb's made. It wasn't a project to sell, but I have a spare unpopulated pcb or two, I could send one over for $20 if you got stuck.
Click to expand...
This is also i nice to have ;) Is this working on a 4016? I see you are using the expansion port. The 4016 has different positions
 
>>> I will first try V05 then. Where can i find this version?

V05 is in the future tense "will have"; so (at the moment) it is on my iMac in development...

If you have a 32K machine (two banks of 16K) you could try swapping over the two banks of 16K memory. I have posted on this method in a couple of threads already.

Dave
 
Before you go unsoldering DRAM chips, let's do a little bit of thinking shall we...

You say that there was a bad connection on UA7. Isn't this one of the MC3446 IEEE488 buffers? What was the bad connection itself? Unless there was a short circuit here somewhere, it is unlikely that this was the actual cause of the 'fix'.

Have you tried the "piggy-back" method (one DRAM at a time)? Not a fool-proof method, but it can yield some results.

Dave
 
Last edited:
There was some kind of sortcut on UA7. After i fixed that, i was getting the normal CBM screen. Then is used the pettester to see everything was ok. It run correct for like 2 hours.
Then i placed the rom back and it was not working anymore. So placed the Pettester back and was getting the screen i posted.
How do i do the 'piggy-back' method?
 
daver2 said:
Have you tried the "piggy-back" method (one DRAM at a time)? Not a fool-proof method, but it can yield some results.

Dave
Click to expand...

I discussed this on pages 14 to 17 of the article I posted.

Piggybacking does sometimes work, other times it cannot, but it is still worth a try as a method to diagnose defective DRAM, without actually having to remove the chips from the pcb.

But, I don't think the hit rate is high. I have never relied on this as a "diagnostic methodology", though that does not mean that others sometimes have found the method successful.

The basic problem with the method is that it only really works if the fault inside the DRAM IC is one where there is something wrong with the tristate buffers in the IC's output circuits and they are simply open circuit or not responding. If they are in fact ok and responding, the chip will impress either a high or low logic level onto the output pin, which in a faulty state, due to other defects in the IC, may be incorrect.

I see the method as "low yield" for fault finding, though, occasionally, the "piggybacking thing" can yield a useful result, so it is not totally without some merit.
 
>>> It run correct for like 2 hours.

Do you mean it performed the DRAM test for 2 hours and increased the PASS counter on a regular basis?

This would indicate to me that the DRAM is actually OK.

Either your 'fix' was not really the fix - and you still have a bad connection somewhere. Replacing the ROM caused the problem to occur.

Or you have just had another fault occur by powering OFF and back on.

Dave
 
daver2 said:
>>> It run correct for like 2 hours.

Do you mean it performed the DRAM test for 2 hours and increased the PASS counter on a regular basis?

This would indicate to me that the DRAM is actually OK.

Either your 'fix' was not really the fix - and you still have a bad connection somewhere. Replacing the ROM caused the problem to occur.

Or you have just had another fault occur by powering OFF and back on.

Dave
Click to expand...
Do you mean it performed the DRAM test for 2 hours and increased the PASS counter on a regular basis? <-- Yes!

Or you have just had another fault occur by powering OFF and back on.
I think that this is the case. because a had other failure before.

First i had this screen. Then i replaced UC5 (2114)
Then found the bad connection and had normal screen and pettester was working
 

Attachments

  • Pet.jpg
    Pet.jpg
    633.7 KB · Views: 0
Last edited:
The "piggy-back" test involves very carefully "piggy-backing" a known, good, working DRAM on top of a DRAM under test (with the power OFF). Make sure that all of the pins mate correctly without shorting out.

Power ON and test.

Power OFF before changing the DRAM location.

If you get a scenario that now works, power OFF and remove the piggy-back chip. Test again to make sure it FAILS and then place the piggy-back chip on again and re-test to make sure it PASSES again. If this is so, replace the associated DRAM.

If you piggy-back test all eight (8) devices and have not observed a working machine at the end, either the piggy-back test method is not working for you OR you have more than one (1) faulty DRAM.

Always, always, power OFF before adding or removing the piggy-back chip.

Dave
 
You must log in or register to reply here.
Back
Top