• Please review our updated Terms and Rules here

Commodore pet 2001-8 no video from logic board

Given we have disconnected pin 2 the counter should be free running... the counters will only be HI HI HI for a single period i.e. 64us so the output of B1 pin 8 should only be low for a period of 64us... B1 must be knackered?

1669053846798.png
 
Nivag Swerdna said:
Given we have disconnected pin 2 the counter should be free running... the counters will only be HI HI HI for a single period i.e. 64us so the output of B1 pin 8 should only be low for a period of 64us... B1 must be knackered?

View attachment 1249148
Click to expand...
My brain has melted too. It was a good test.
I shall revert the A1 chip and await the B1 replacement if we are all in agreement.
 
RetroGadgetMan said:
My brain has melted too. It was a good test.
I shall revert the A1 chip and await the B1 replacement if we are all in agreement.
Click to expand...
I think we are all in agreement to replace B1. Though, just in case I would test the connections between A1 and B1 with the meter.

( One thing you could do, to get a recording where A1 was still soldered in and you could not disable the reset so easily, that would probably work to get a good recording of B1s output on pin 8; is to put the scope on A1 pin one as you have done that shows a nice stable clock pulse. Then use another probe on channel 2 of the scope to look at the output of B1, you would also need to slow the timebase down a few steps to see a few B1 output pulses, and we could measure their frequency from that).
 
Last edited:
Hugo Holden said:
I think we are all in agreement to replace B1. Though, just in case I would test the connections between A1 and B1 with the meter.

One thing you could do, that would probably work to get a good recording of B1s output on pin 8; is to put the scope on A1 pin one as you have done that shows a nice stable clock pulse. Then use another probe on channel 2 of the scope to look at the output of B1, you would also need to slow the timebase down a few steps to see a few B1 output pulses, and we could measure their frequency from that, but I think they will be 4x faster than they should be.
Click to expand...
I have already ordered a 2nd probe that has a X1 and X10 setting.
I did check all the connection between these two already and all was ok. Post #337
 
Hugo Holden said:
I was thinking, since Commodore did not really provide an adequate description of this creative circuit design, when we figure it out, we should write up a detailed description of it. Maybe they wanted to keep it "proprietary"
Click to expand...
This might be a good time for me to mention that I have made a Verilog simulation of this circuit, basically all of page 320008-3. I originally did it to just learn what the exact timings and relationships of the video signals. It was a quick hack but it seems to be correct for all the pins I have probed in my machine. I have returned to it once in a while to try to learn how it works. Recently, I fixed some data path problems while trying to learn exactly how "snow" happens on old PETs when you write the video RAM.

It's buried in a directory in github in my "Commodore PET in an FPGA" project:
github.com

Pet2001_Arty/petvid.v at master · skibo/Pet2001_Arty

A Commodore PET in an Artix-7 FPGA. Contribute to skibo/Pet2001_Arty development by creating an account on GitHub.
github.com github.com
 
RetroGadgetMan said:
Ok now have a visual on pet screen again.

Trigger set to norm
Pic 1 Pin 8 of B1 50us
A1
Pic 2 pin 9 50us
Pic 3 pin 8 50us
Pic 4 pin 11 0.1 time div
Click to expand...
Hmmmm.... There might be something rotten in the state of Denmark as the old saying goes, or perhaps a red herring.

Just looking at the recordings again here. There is something very wrong with the A1 pin 8 recording pic 3.

If we now have the A1 counter free from resets, then those three A1 output pins we are checking should only have square waves on them because they are the result of clocked flip flops, that is if there is a uniform 15.75kHz clock pulse and it is not changing frequency abruptly at times.

It looks therefore that somehow that A1 is still being reset, or the clock pulse must be changing frequency.

With the second scope probe it could be worth checking there are no pulses somehow on the other reset pin 3 of A1 which should not be there (which as I recall was tied by a resistor, I'll have to check the schematic again) and make a dual trace recording of A1 pin 8 and the clock on A1 pin 1.
 
Pic 2 (aka A1 pin 9) is squarish... Pic 3 (A1 pin 8) and Pic 4 (A1 Pin 11) have never really been explained, that's why I previously blamed A1.

To B1, or not to B1: that is the question.

@RetroGadgetMan you could remove B1 and replace with an empty socket... we could then see if the outputs (pins 9, 8 and 11 of A1) recovered; temporarily removing the character ROM again perhaps.
 
tsky said:
This might be a good time for me to mention that I have made a Verilog simulation of this circuit, basically all of page 320008-3.
Click to expand...
Wow, this looks like excellent work.

Having never gone much past using PALASM and ABEL in Hardware Description Languages, I do not know how to use this file in a simulation to see waveforms, etc. I am impressed that even old parts like the Commodore 6550 RAM and 6540 ROM have been modeled and are in a library.

I suppose one would have to use a powerful LINUX type machine and load up tons of special apps. Would all the required software be free or low cost?
 
Nivag Swerdna said:
Pic 2 (aka A1 pin 9) is squarish... Pic 3 (A1 pin 8) and Pic 4 (A1 Pin 11) have never really been explained, that's why I previously blamed A1.

To B1, or not to B1: that is the question.

@RetroGadgetMan you could remove B1 and replace with an empty socket... we could then see if the outputs (pins 9, 8 and 11 of A1) recovered; temporarily removing the character ROM again perhaps.
Click to expand...
Ok I will do this later. Shall I remove the bridge wire and re seat pin 2 of A1 aswell.
 
Dave,

>>> I suppose one would have to use a powerful LINUX type machine and load up tons of special apps. Would all the required software be free or low cost?

The 'standard edition' tool is free to download from the Xilinx website (https://www.xilinx.com/products/design-tools/vivado.html). A (fairly) powerful Windows box is suitable. It is a large download though. If you want to 'program' an FPGA there is a cost to purchasing the requisite programming cable.

There are other 'freeware' Verilog simulators out there as well.

Dave
 
dave_m said:
Having never gone much past using PALASM and ABEL in Hardware Description Languages, I do not know how to use this file in a simulation to see waveforms, etc. I am impressed that even old parts like the Commodore 6550 RAM and 6540 ROM have been modeled and are in a library.
Click to expand...

I hacked those models together and I don't guarantee their accuracy but it all seems to work.
dave_m said:
I suppose one would have to use a powerful LINUX type machine and load up tons of special apps. Would all the required software be free or low cost?
Click to expand...

I run it on Vivado but I one idea I have is to dump the waveforms in a vcd file and then you could use gtkwave to browse it. I should make a top post about this and I can include a few screen shots.

-Thomas
 
Just a quick update. I am still awaiting the ic for B1. Postal strikes here have ment no post for several days now 😔
 
Nivag Swerdna said:
You can never have enough 74LS244 chips.... if you don't already :)
Click to expand...
Some years ago a local Electronics Supply shop in my area closed down. They had this massive stack of 74LSxxx TTL's many labelled drawers in a large cabinet. I wandered in there and said to the owner, what is up ? He said he was retiring and closing the the shop. I offered him $100 cash for all the IC's and the cabinets they were in, he agreed on the spot! What a lucky dog I was that day. There was also 4000 series cmos. They have come in so handy over the years for repairs and projects.

I also had a large stock of plain 74 series TTL's because back in the 1970's I got obsessed with Atari's arcade games of Pong & Tank and have to have all the spares for those. So usually, if my vintage computers break I have a spare. Not always Though, recently I had to buy spare IEEE-488 interface buffer IC's for my PET.

Also I got caught out on an IBM serial card without a spare IC, but I managed to temporarily repair it using part of the remaining IC that was still working, I called it The Lost in Space repair:


If it takes too long for that 74LS20 to turn up one option to test the theory it was faulty could be to cobble some other gates together to act as a 4 input NAND.
 
Hugo Holden said:
Some years ago a local Electronics Supply shop in my area closed down. They had this massive stack of 74LSxxx TTL's many labelled drawers in a large cabinet. I wandered in there and said to the owner, what is up ? He said he was retiring and closing the the shop. I offered him $100 cash for all the IC's and the cabinets they were in, he agreed on the spot! What a lucky dog I was that day. There was also 4000 series cmos. They have come in so handy over the years for repairs and projects.

I also had a large stock of plain 74 series TTL's because back in the 1970's I got obsessed with Atari's arcade games of Pong & Tank and have to have all the spares for those. So usually, if my vintage computers break I have a spare. Not always Though, recently I had to buy spare IEEE-488 interface buffer IC's for my PET.

Also I got caught out on an IBM serial card without a spare IC, but I managed to temporarily repair it using part of the remaining IC that was still working, I called it The Lost in Space repair:


If it takes too long for that 74LS20 to turn up one option to test the theory it was faulty could be to cobble some other gates together to act as a 4 input NAND.
Click to expand...
What a cool repair. I had actually thought about using different chips to use the gates I need but hoping it won't come to that yet. But going forward this may have to be the way if any of the logic becomes unavailable. At least one of the ics on the board is unavailable at the moment through eBay at least.
 
Don't forget that HCT series is a modern equivalent (and manufacturered).... so 74HCT20 would do you well. Also many devices have equivalents with differing part numbers.

This should stir a debate.... but generally a HCT should replace an LS.... and use less power, not get so hot etc... they are not identical... but close enough that in most cases you can use them interchangeably.

https://www.onsemi.cn/pub/collateral/an-368cn.pdf
 
Last edited:
Nivag Swerdna said:
Don't forget that HCT series is a modern equivalent (and manufacturered).... so 74HCT20 would do you well. Also many devices have equivalents with differing part numbers.

This should stir a debate.... but generally a HCT should replace an LS.... and use less power, not get so hot etc... they are not identical... but close enough that in most cases you can use them interchangeably.

https://www.onsemi.cn/pub/collateral/an-368cn.pdf
Click to expand...
Yes, the HCT cmos range had their inputs conform to TTL voltage thresholds. They are really good if you want to make low power (battery operated gear)

Back in the late 1970's Silicon Valley experienced a shortage of 74LS chips, so they reverted temporarily to plain 74 types (albeit a little more power hungry).

To give an idea of this, if you averaged this effect on a pcb, with over 60 plain 74 series IC's, with TTL, the current consumption was just over 1 Amp with the original 74 series. But only about 360mA with LS TTL.

Other options, are 74F TTL and 74S TTL if you don't mind the extra power consumption.

And yet another option too, if you are stuck for some 74 series TTL, search instead for the 54 prefix versions. These are the mil spec types and have ceramic packages. All very good quality. Here is a typical offering of these from the late 1970's:

www.ebay.com

2x FAIRCHILD 5420DMQB , 5420 IC CERAMIC , CDIP-16 | eBay

2 PCS FAIRCHILD 5420DMQB, 5420. IC CERAMIC, CDIP-16 . NEW PART, OLD D/C.
www.ebay.com
 
You must log in or register to reply here.
Back
Top