• Please review our updated Terms and Rules here

Pet 2001 strange characters

Excellent.

So, there is some 'science' to the NOP generator rather than just looking at pretty pictures.

For address line A0 on the CPU - i.e. ahead of the address buffers, the frequency should be 250 kHz. On your oscilloscope, this is 4 divisions at 1.0 us/div = 4 us = 250 kHz. Correct answer. See: https://www.unitjuggler.com/convert-frequency-from-µs(p)-to-kHz.html?val=4.

You also need to look at the HIGH and LOW voltage levels for the TTL signal. I generally work on the principle that a maximum LOW voltage level of 0.5 Volts is fine (i.e. a reading of < 0.5 Volts is good). and a minimum HIGH voltage level of 4.0 Volts is fine (i.e. a reading of > 4.0 Volts is good). Of course, the lower the LOW reading is, and the higher the HIGH reading is, the better.

Also, the frequency should be absolutely stable.

The above tests for any faulty address lines and short circuits between the CPU address lines.

As you measure higher address lines, the frequency should halve for each subsequent address line.

So, A1 = A0/2 = 250 kHz/2 = 125 kHz.

Science... If you have access to a spreadsheet, use that to ascertain what you should see on your oscilloscope.

After you have checked ALL of the address lines on the CPU (signals A0 to A15) you can then test on the other side of the address buffers (signals BA0 to BA11):

1762341714438.png

This check makes sure that the buffers for the address bus are working OK and that there are no short circuits. Again, the correct frequencies and acceptable voltage levels.

You can then check on the pins of the ROMS and logic gates D2, D3 and D4 (for the BAn signals). This checks for PCB track breaks or other related issues...

After that, I will teach you how to check a few other CPU signals, and we can move on to check the 74154 address decoder and buffer enable signals.

We can also perform a limited test on the data bus - but this is not science (at this stage) but a bit of magic!

Does this make sense?

Dave
 
From for first post in thread

>>> After removing all ram, rom via and pia we used the tynemouth ram rom board.

So you already have a RAM ROM board?


>>>There is some recognition in the boot screen but it's scrambled. I see the numbers and ready scrambled

Yes. It looks like it is mostly working, just the Video RAM, access to Video RAM or Character Generator having issues?

>>> I read somewhere that there must memory on the mainboard even using the ramrom board. Is this True?

YMMV but that is not normally true... Any decent RAM ROM board will substitute all System RAM (not video RAM).

FWIW when not substituting RAM you rapidly need some to provide the Stack and Zero Page but PETTESTER and VOSSI are smart enough to avoid memory enough to provide diagnostics.
 
The only slight 'niggle' is that PETTESTER is not working - so (possibly) something else has gone faulty between then and now...

Dave
 
A lot to learn :-) thx.

I will do this as soon as i have time.
The other way someone suggested, putting the 74LS154 on a breadboard and test.
This could also a nice way of learning in think or not?
daver2 said:
Excellent.

So, there is some 'science' to the NOP generator rather than just looking at pretty pictures.

For address line A0 on the CPU - i.e. ahead of the address buffers, the frequency should be 250 kHz. On your oscilloscope, this is 4 divisions at 1.0 us/div = 4 us = 250 kHz. Correct answer. See: https://www.unitjuggler.com/convert-frequency-from-µs(p)-to-kHz.html?val=4.

You also need to look at the HIGH and LOW voltage levels for the TTL signal. I generally work on the principle that a maximum LOW voltage level of 0.5 Volts is fine (i.e. a reading of < 0.5 Volts is good). and a minimum HIGH voltage level of 4.0 Volts is fine (i.e. a reading of > 4.0 Volts is good). Of course, the lower the LOW reading is, and the higher the HIGH reading is, the better.

Also, the frequency should be absolutely stable.

The above tests for any faulty address lines and short circuits between the CPU address lines.

As you measure higher address lines, the frequency should halve for each subsequent address line.

So, A1 = A0/2 = 250 kHz/2 = 125 kHz.

Science... If you have access to a spreadsheet, use that to ascertain what you should see on your oscilloscope.

After you have checked ALL of the address lines on the CPU (signals A0 to A15) you can then test on the other side of the address buffers (signals BA0 to BA11):

View attachment 1310516

This check makes sure that the buffers for the address bus are working OK and that there are no short circuits. Again, the correct frequencies and acceptable voltage levels.

You can then check on the pins of the ROMS and logic gates D2, D3 and D4 (for the BAn signals). This checks for PCB track breaks or other related issues...

After that, I will teach you how to check a few other CPU signals, and we can move on to check the 74154 address decoder and buffer enable signals.

We can also perform a limited test on the data bus - but this is not science (at this stage) but a bit of magic!

Does this make sense?

Dave
Click to expand...
This is a huge task

Started to trace a0 tm a11 to the input of the 74ls17's and 74ls08 following the schema.

Just a connection test.

Now try to trace the signals on the other side of the ttl ic. And then further.
Will take some time for me.

And yes looking at frequency, and high low.

This is a long journey for me that learns me a lot.
 
The tynemouth board, or at least some of them, has a tester... It might be worth trying that to work... From a Google it will do a RAM check and more importantly fill video RAM with a pattern.
 
Nivag Swerdna said:
The tynemouth board, or at least some of them, has a tester... It might be worth trying that to work... From a Google it will do a RAM check and more importantly fill video RAM with a pattern.
Click to expand...
Yes, when i had one. But that was an one time opportunity 200 km from here.
 
It's much, much easier to test the 74154 in-situ with the NOP generator and the oscilloscope...

You can (of course) change the testing procrdure to test (first) A0, then BA0, then all the places BA0 goes to. Then move on to A1, ... A15.

Tick off the pins you have tested on the schematic as you go.

Stop and report pins that look suspect!

Dave
 
Firstly, to look at what is happening on the CPU address bus, then the far side of the address buffers, then the outputs of the 74154.

I have just written a number of posts on this very topic.

Dave
 
First yesterday i tested with the nop tester the a0 till a15 on the cpu and that looks ok.

But in the afternoon my bitfixer emulator arrived. Build it and this morning put it in

Nothing. The same screen.

1000087507.jpg
 
So now the roms and the ttl logic 74LS154 and others can be excluded.
Vram and memory replaced with new ones and tested.

I ran the machine without cpu at all and that gave me the same screen also.

What i feel is that my cpu is very loose in the socket.

New observations. But what to do with it
 
Firstly, what was the results of the NOP test, and how did you actually test?

You have done an awful lot in the last two (2) posts, but provided no actual details of the configuration you used and the results. We are not (necessarily) just looking at the screen...

Running the machine without a CPU will give a random pattern on the screen (as it appears to be doing).

My $64k question would be (with the CPU in) is "is pin 7 of the CPU (SYNC) oscillating or not"? This pin indicates whether the CPU is fetching and executing instructions. One pulse = 1 instruction.

If you now have the ROMulator, how have you configured the switches (for example)?

Dave
 
Last edited:
Nop test on the cpu itself was ok.
With clips on the cpu pins, so above the white socket.
There is no resistance at all putting it in it out
 
That is not all we test for with the NOP generator. We test for the frequency of each individual address line (both before and after the address buffers). You can also check the 74154 and other parts of the address decoding 'glue'.

If the CPU socket appears loose, is it a white socket or not?

My suggestion is to cut a pin off a spare/sacrificial IC and solder it to a piece of stout copper wire. Test fit to each pin of the IC socket, and see which pins are loose. Test fit to other IC sockets to gauge what "good" feels like.

Can you take the IC socket apart? Some you can, some you can't. If you can't, it may be a socket replacement IF it is faulty. If you can take the IC socket apart, you may be able to repair loose individual sockets - or replace individual sockets.

Dave
 
I can take the cpu out and in without any resistance.
I will replace the socket with a new quality socket.

1000087523.jpg
 
You cannot be serious.

I have posted many times how to check sockets for yourself using a pin taken from a defunct IC soldered to a small wire handle and checking the socket tension for each pin, one by one.

My PET has all of its original white sockets, they are just fine. No connection problems and no intermittent problems.

But, sockets do get damaged especially if oversized pins are pushed into them from various adapters.

That CHATGPT was just "Global Advice" it might not relate to your sockets at all.

Use your own brain (if there is any of it left after the AI has done your homework) and check the sockets yourself. It cannot be done feeling the global pressure for the whole IC, it has to be done pin by pin.

Sure, if the socket is defective, go for it, replace it, but don't do that because somebody tells you to or an AI tells you to do it.
 
Dear Holden
Lets keep this thread informative.
For me the socket has to be ruled out.
So don't judge but stay with me. Thanks
 
Last edited:
You must log in or register to reply here.
Back
Top