Re: Non-working Commodore PET 3016
 

Absolutely so, but that is the area where I think the next fault is. The screen shows random stuff if we prevent the CPU from clearing it by taking out the PROM which has the clear-screen routine in it. It gets cleared if we put the PROM back in, but then it doesn't make it all the way through initialisation.

The chief suspect now is possibly the system RAM because the return-from-subroutine mechanism, widely used in the OS PROMs, can not work unless you have working system RAM.

It so happens that the clearing of the screen RAM and initialisation of the peripheral ICs takes place in the very first subroutine, so those tasks appear to be being carried out successfully, but the hypothesis is that when the system then tries to return-from-subroutine it fails because the stack, which is held in system RAM, is not working.

Re: Non-working Commodore PET 3016
 

Think it is worth trying the bank-swap dodge (resistor crossover) in that document that aj found for us?

Re: Non-working Commodore PET 3016
 

By the looks of it I could persuade the PET it only has 8K by using the jumpers on page 3.

Is that worth a try as well?

Colin.

Re: Non-working Commodore PET 3016
 

Yes and when I realised that I deleted my post. I'll try to engage brain before posting again.

Alan

Re: Non-working Commodore PET 3016
 

No problem aj, at least yours has synchromesh.

I've just remembered that the system RAM is separated from the CPU by not one but two lots of the dreaded 74LS244 buffers, one set being our old friends UE10/U11 which we never changed, and UI10 / UI11 which separate the RAM data pins from the buffered databus provided by UE10 / UE11.

A while back in #832-#833 you (Colin) did some while-you-wait resistance measurements around the system RAM area and you did find one possible anomaly where the circuit node of UI11 pins 2/3 - J2 pins 2/14 - I2 pins 2/14 had a noticeably different resistance to +5V and 0V than the other seven similar nodes. This could indicate a problem with UI11, UJ2 or UI2.

Could you maybe do those measurements again, power off of course, but this time use a different resistance range for all the tests to see if that amplifies or exaggerates the difference between that node and the others?

Then after that, make comparative scope observations of these pairs of connected pins (as they are connected toegether, you need only scope the first one in each pair)

UI10 pins 8-9
UI10 pins 6-7
UI10 pins 4-5
UI10 pins 2-3

UI11 pins 8-9
UI11 pins 6-7
UI11 pins 4-5
UI11 pins 2-3

The signals on those pins may look quite chaotic, but what you are really looking for is a significant difference in the height or shape of the signal on one or two compared to the others.

Re: Non-working Commodore PET 3016
 

Colin, sorry, the idea of switching out half of the RAM is a good thought, but if you can just try the above first, please.

Re: Non-working Commodore PET 3016
 

If the reference to page 7 in Alans deleted post was from my post #1052, then I was referring to page 7 of the pdf and not to page 7 of the pet memory.

It gets confusing when posts are deleted.

Re: Non-working Commodore PET 3016
 

I agree with Sirius, try #1065 first.

Switching out half of the ram might not work if the fault is caused by a stuck output on one of the ram chips, even if the bad chip were no longer selected it would probably still be pulling the data line and blocking the ram in the other bank.

Hopefully #1065 will point to only one bad data line on the system ram, which is then down to only two bad chips, and a fifty-fifty chance you can pick the right one to remove.

Re: Non-working Commodore PET 3016
 

Sorry for the confusion Mark. Ironically I deleted the post because I thought it would cause confusion. :'(

I did realise that you were referring to page 7 of the PDF by the way. Time for me to go away and play with my PET. :)

Alan

Re: Non-working Commodore PET 3016
 

To stir things a bit further I believe aj (and he has just confirmed it ) knew that but was pointing out that it was referring to system RAM, not video RAM. He thought for a moment that we were still musing over a potential video RAM problem.

Having read through that doc again it sounds increasingly like it must have been supplied with a diagnostic gadget of some sort rather than using a built in 'bootloader' feature. I would be interested to know what purpose the 'diag' connection (J9 pin 20 / User port pin 5) on sheet 3 serves, and how it might be used. It goes to pin 9 of UC7, 6520, and is indicated on the diagram as being both input and output.

Re: Non-working Commodore PET 3016
 

Because you can. :)

Well done, by the way, and actually entirely achieved in spite of our advice, rather than because of it.

Re: Non-working Commodore PET 3016
 

There’s no pull up on diag, so I’m guessing it might be an output, so probably best if we don’t ask colin to short it to 0v or 5v to see what happens.

Re: Non-working Commodore PET 3016
 

On the circuit diagram, inside the chip outline, there are arrows helpfully indicating the mode / direction in which I/O pins are used. Some are INs, some are OUTs and some are IN/OUTs. DIAG is one of those.

I wouldn't ask anyone to do anything to an IC pin without a better idea of what it does. ;)

Re: Non-working Commodore PET 3016
 

Thoughts .....

Would it be worth getting a 62256 static RAM IC (together with a breadboard if needed, and some long pin-pin jumper cables that can be cut in half and the new ends tack-soldered to the PET) and patching that in temporarily in place of the RAM on the motherboard?

If we think the video RAM is fine, could we make a 2816 boot EPROM to do some sort of memory test? Have to be all inline code with no subroutine calls, because we can't assume a usable stack page, but we could use video RAM as workspace.

Re: Non-working Commodore PET 3016
 

Slothie has kindly provided the code for that very thing with all the precautions you have outlined, see posts #992 / #993. It sends its output to the video RAM / video display which might have been a problem before today but now we are fairly sure Colin has fixed that.

I'm guessing Colin does not have an EEPROM programmer or a suitable EEPROM-to-PET-PROM outline adaptor handy so that's not the first thing we are trying, but it is on the list.

Let's see what Colin finds for us first.

Re: Non-working Commodore PET 3016
 

OK - the voltages all check out OK when powered on. As do the resistances on the RAM chips.

However, the resistance measurements on UI10/11 do not.

For example (meter set to 20K at all times)

UI10 6/7 (0V) - 1.54
UI11 6/7 (0V) - 2.03

UI10 2/3 (0V) - 2.07
UI11 2/3 (0V) - 1.56

UI10 4/5 (5V) - 1.61
UI11 4/5 (5V) - 2.29

UI10 2/3 (5V) - 2.16
UI11 2/3 (5V) - 1.67

Am I making the right comparisons here?

I have not scoped anything yet - would that still be useful?

Colin.

Re: Non-working Commodore PET 3016
 

Slightly confusing because you have done half of them to +5V and half to 0V, unless I am misunderstanding your results. To clarify:

Measure from +5V (red lead on +5V) to each of

UI10 pins 8-9
UI10 pins 6-7
UI10 pins 4-5
UI10 pins 2-3
UI11 pins 8-9
UI11 pins 6-7
UI11 pins 4-5
UI11 pins 2-3

And report the results, then measure from 0V (black lead on 0V) to each of

UI10 pins 8-9
UI10 pins 6-7
UI10 pins 4-5
UI10 pins 2-3
UI11 pins 8-9
UI11 pins 6-7
UI11 pins 4-5
UI11 pins 2-3

..and report the results.

Regardless of what you find, could you please also do the scope comparison between these eight pin pairs as well. Doesn't have to be tonight, take your time.

Re: Non-working Commodore PET 3016
 

I should probably clarify 'scope comparison' - I just mean, to take captures from all eight of those points so we can see if there is anything suspiciously different about any of them.

Re: Non-working Commodore PET 3016
 

Will do.

I was trying to show the differences between what I thought should be similar readings on UI10 and UI11 - for example:

UI10 2/3 (5V) - 2.16
UI11 2/3 (5V) - 1.67

I'll do them again tomorrow in the order you suggest so we can et a better set of comparisons. Scope readings will also follow.

Colin.

Re: Non-working Commodore PET 3016
 

We'll look out for those.

It might not feel like it but with the restoration of the video RAM and video chain you've taken another major step forward today. If these next measurements don't point to any specific area we'll try the memory bank swap / memory halving tricks to see if they show anything useful. They might, might not.

We'll also work out a way to get Slothie's test code into the machine to test the system RAM in-situ, if possible.