Commodore pet flickering screen

[daver2]

Nope...

A resistor is a passive device. This is being actively driven...

Each IC TTL input pin has a very weak pull-up resistor inside. The recommendation is to always have an external pull-up resistor on unused inputs (as is the purpose of R39) but even with it not there, you shouldn't see what you are seeing.

It is either 'conductive contamination' or a faulty input pin on either C5 or D5 connected to R39.

Something is 'driving' this network low.

One option is to use your oscilloscope and accurately measure the LOW voltage on each pin of C5 and D5 connected to R39 looking for the LOWEST voltage. This may indicate the faulty pin... But your oscilloscope may also not be up to the task.

Dave

 

[genobombino]

Nope...

A resistor is a passive device. This is being actively driven...

Each IC TTL input pin has a very weak pull-up resistor inside. The recommendation is to always have an external pull-up resistor on unused inputs (as is the purpose of R39) but even with it not there, you shouldn't see what you are seeing.

It is either 'conductive contamination' or a faulty input pin on either C5 or D5 connected to R39.

Something is 'driving' this network low.

One option is to use your oscilloscope and accurately measure the LOW voltage on each pin of C5 and D5 connected to R39 looking for the LOWEST voltage. This may indicate the faulty pin... But your oscilloscope may also not be up to the task.

Dave

So i was checking c5 and d5, and i couldnt really find any abnormalities with d5.

However.

C5/10- avg 4-5v while working, and when the problem is active it jumps down to .8 -- .79v

C5/6- avg 1.6v, and jumps to 4.2v while problem is active

C5/4- same thing as c5/6 as they are connected to the same trace

Im not sure if im reading the pins wrong, but it doesnt appear that c5/10 actually connects to anything on either side of the board

And for the previous message, i was mistaking c5/13 for c5/10. The resistor i was measuring was connected to d4/9 and c5/10,
(My appologies)

 

[Hugo Holden]

Im not sure if im reading the pins wrong, but it doesnt appear that c5/10 actually connects to anything on either side of the board

Most likely there is a track hiding under the IC body. The way to double check is to use the meter to see if it connects to the other IC's they show in the diagram.

There have been cases where a pin has been left disconnected on a pcb, due to a pcb design error, and didn't cause any trouble, because it assumes a logic high, so nobody found the missing trackwork. If this is the case and that pin 10 has signal on it, and goes low too, you are in luck, because it narrows it down to that IC C5 !

Your are sure how to identify the pin numbers on IC's ? Generally with the IC long axis orientated on the horizontal , the writing / label is the right way up when the notch in the IC body is on the left hand side, and there is often a dot or a dimple on the body where pin 1 is on the lower left hand side below the body notch, lower row of pins. And on the lower row of pins you count, starting from left to right. Many (not all) of the TTL logic chips are 14 or 16 pin, and pin 7 or 8 is therefore on the low row on the right. And pin 14 or pin 16 on the upper row is on the left.

Most (but not all) 74 series IC's have pin 7 or 8 as GND (for the 14 and 16 pin types respectively) and pin 14 or 16 as +5V. There are notable exceptions to watch out for such as the 7490 and 7493 with non standard power supply pins, that really put a spanner in the works, so the manufacturers later created the 74390 and 74393 to fix this problem.

 

[genobombino]

Most likely there is a track hiding under the IC body. The way to double check is to use the meter to see if it connects to the other IC's they show in the diagram.

There have been cases where a pin has been left disconnected on a pcb, due to a pcb design error, and didn't cause any trouble, because it assumes a logic high, so nobody found the missing trackwork. If this is the case and that pin 10 has signal on it, and goes low too, you are in luck, because it narrows it down to that IC C5 !

Your are sure how to identify the pin numbers on IC's ? Generally with the IC long axis orientated on the horizontal , the writing / label is the right way up when the notch in the IC body is on the left hand side, and there is often a dot or a dimple on the body where pin 1 is on the lower left hand side below the body notch, lower row of pins. And on the lower row of pins you count, starting from left to right. Many (not all) of the TTL logic chips are 14 or 16 pin, and pin 7 or 8 is therefore on the low row on the right. And pin 14 or pin 16 on the upper row is on the left.

Most (but not all) 74 series IC's have pin 7 or 8 as GND (for the 14 and 16 pin types respectively) and pin 14 or 16 as +5V. There are notable exceptions to watch out for such as the 7490 and 7493 with non standard power supply pins, that really put a spanner in the works, so the manufacturers later created the 74390 and 74393 to fix this problem.

Oh well then if thats how to count them, (counter clockwise) luckily ive been doing it right

 

[Hugo Holden]

Oh well then if thats how to count them, (counter clockwise) luckily ive been doing it right

Excellent. Now you just have to find out if PIN 10 of C5 is connected anywhere or not, before making the next move. If it really is not connected anywhere, and you confirm that pulses are appearing on it and or it is going low, the IC is internally defective, replace IC C5. If that is the fault, it is pretty rare, I only know of one other case where a 74 IC failed causing pulses to appear on one of its input pins. So not unheard of, but pretty rare, I'm sure Daver2 would agree.

Some tips: I'm not sure how you are for de-soldering IC's. These 74 series IC's are common, plentiful and cheap. The idea is to sacrifice the IC in favor of any pcb damage. So if you remove it, I would recommend using small sharp needle nose cutters and cut each pin one by one close to the IC body. Then apply fresh solder to the pins on the solder side of the board and IC pads, this improves the thermal coupling of the iron to fresh solder. Grab each pin one by one on the component side of the board with small pliers or toothed forceps, don't pull on it at all, until you are 100% sure, after applying the iron, on the solder side of the board, that the solder is fully melted right through the hole and around the pin and withdraw the pin (also check underneath, on the solder side, that none of the pins are bent over before you remove them). After that clear the holes with the solder sucker and clean up the surface flux with contact cleaner or IPA and a cue tip. You may as well fit a socket. Fit a dual wipe type as they have thinner pins than machined pin sockets, they can also be removed, one by one, if the socket gets damaged. Machine pin sockets are more tricky to remove and more difficult to solder suck, if it turns out for some reason, they need to be replaced later.

 

[genobombino]

Most likely there is a track hiding under the IC body. The way to double check is to use the meter to see if it connects to the other IC's they show in the diagram.

There have been cases where a pin has been left disconnected on a pcb, due to a pcb design error, and didn't cause any trouble, because it assumes a logic high, so nobody found the missing trackwork. If this is the case and that pin 10 has signal on it, and goes low too, you are in luck, because it narrows it down to that IC C5 !

Your are sure how to identify the pin numbers on IC's ? Generally with the IC long axis orientated on the horizontal , the writing / label is the right way up when the notch in the IC body is on the left hand side, and there is often a dot or a dimple on the body where pin 1 is on the lower left hand side below the body notch, lower row of pins. And on the lower row of pins you count, starting from left to right. Many (not all) of the TTL logic chips are 14 or 16 pin, and pin 7 or 8 is therefore on the low row on the right. And pin 14 or pin 16 on the upper row is on the left.

Most (but not all) 74 series IC's have pin 7 or 8 as GND (for the 14 and 16 pin types respectively) and pin 14 or 16 as +5V. There are notable exceptions to watch out for such as the 7490 and 7493 with non standard power supply pins, that really put a spanner in the works, so the manufacturers later created the 74390 and 74393 to fix this problem.

Should I replace the chip entirely, or should i try to lift it and see if i got animal crackers under there

 

[Hugo Holden]

Should I replace the chip entirely, or should i try to lift it and see if i got animal crackers under there

Double check with the meter that the pin 10 is not connected (as you suspect) anywhere to the other IC's, that the schematic shows it is connected to. If it is connected the plans change, because one of the other IC's could be defective.

If that is the case you need to remove the chip, the best way, is the way I suggested, is by destroying the chip (as it is faulty if your testing and inspection is reliable). So you have to get some spare IC's. If there was debris under the IC causing trouble, you would have to remove it anyway for cleaning.

Of course, if you are very good at de-soldering and have good tools for this, and you can completely free each IC pin, all of them, in the plated through holes, you could lift the whole chip out preserving the chip, but, if it is not perfectly free on each pin, you may tear out plated through holes and rip up pcb tracks and damage the board, which is a big no-no.

So for the learning or less experienced, I would go with the first suggestion of cutting the IC pins close to the body and then removing the pins from the pcb one by one, as less chance of damage to the pcb.

There are plenty of 74LS107's on ebay and elsewhere. If I had a working Teleportation machine, I could send you one right away, but its currently broken after a Fly got into the mechanism.

One thing you could do, before anything else, as an experiment, is to force the pin 10 high, with a low value resistor like 100 Ohms or similar to the +5V rail and see what effect that has. That would likely overcome any resistance from conductive debris hiding under the IC.

 

[genobombino]

Double check with the meter that the pin 10 is not connected (as you suspect) anywhere to the other IC's, that the schematic shows it is connected to. If it is connected the plans change, because one of the other IC's could be defective.

If that is the case you need to remove the chip, the best way, is the way I suggested, is by destroying the chip (as it is faulty if your testing and inspection is reliable). So you have to get some spare IC's. If there was debris under the IC causing trouble, you would have to remove it anyway for cleaning.

Of course, if you are very good at de-soldering and have good tools for this, and you can completely free each IC pin, all of them, in the plated through holes, you could lift the whole chip out preserving the chip, but, if it is not perfectly free on each pin, you may tear out plated through holes and rip up pcb tracks and damage the board, which is a big no-no.

So for the learning or less experienced, I would go with the first suggestion of cutting the IC pins close to the body and then removing the pins from the pcb one by one, as less chance of damage to the pcb.

There are plenty of 74LS107's on ebay and elsewhere. If I had a working Teleportation machine, I could send you one right away, but its currently broken after a Fly got into the mechanism.

One thing you could do, before anything else, as an experiment, is to force the pin 10 high, with a low value resistor like 100 Ohms or similar to the +5V rail and see what effect that has. That would likely overcome any resistance from conductive debris hiding under the IC.

So, i removed the chip, and while there was no debri under the chip, unfortunately i found that c5/10 and c5/13 are actually connected. I wasnt able to see it since the trace was under the chip.

Luckily its been socketed, and a brand new chip has been put in its place, so that takes c5 off the list

Anywhere i should start next?

Also i have found that not only does the horizontal line go high, but so doesnt the vertical AND video line go high as well. Dont know if that helps with anything, but i thought id mention it.

 

[Hugo Holden]

So, i removed the chip, and while there was no debri under the chip, unfortunately i found that c5/10 and c5/13 are actually connected. I wasnt able to see it since the trace was under the chip.

Luckily its been socketed, and a brand new chip has been put in its place, so that takes c5 off the list

Anywhere i should start next?

Also i have found that not only does the horizontal line go high, but so doesnt the vertical AND video line go high as well. Dont know if that helps with anything, but i thought id mention it.

This is why I suggested checking with the meter that pin 10 was or was not connected to the other pins on the other IC's, but obviously you did not do that, it can be difficult to see on visual inspection under an IC, sometimes bright light transillumination of the board can help too.

With IC C5 eliminated from the equation now, and I gather that was not the cause of the fault, it suggests that it must be IC D5 as that is the only other IC connected to that pull up line.

But, before you think of removing it: Visually inspect all of the tracks between C5 and D5, just to make sure it all looks normal. Also, find the 1k pull up resistor R39 connected to those. There is a small possibility that the +5V line feeding it could be going, not open circuit, but going hard low. So check that when the fault occurs that there is still a solid +5v applied to that resistor.

 

[genobombino]

This is why I suggested checking with the meter that pin 10 was or was not connected to the other pins on the other IC's, but obviously you did not do that, it can be difficult to see on visual inspection under an IC, sometimes bright light transillumination of the board can help too.

With IC C5 eliminated from the equation now, and I gather that was not the cause of the fault, it suggests that it must be IC D5 as that is the only other IC connected to that pull up line.

But, before you think of removing it: Visually inspect all of the tracks between C5 and D5, just to make sure it all looks normal. Also, find the 1k pull up resistor R39 connected to those. There is a small possibility that the +5V line feeding it could be going, not open circuit, but going hard low. So check that when the fault occurs that there is still a solid +5v applied to that resistor.

The IC located in D5 appears to be an SN74177N from Texas instruments. It seems as though its a discontinued part, any idea on where i can find a drop in replacement that will arrive in a couple days?

 

[Hugo Holden]

The IC located in D5 appears to be an SN74177N from Texas instruments. It seems as though its a discontinued part, any idea on where i can find a drop in replacement that will arrive in a couple days?

On my schematic the IC D5 is another 74LS107 and the diagram of its internals matches that part. A dual JK flip flop. I think IC's D6 & D7 are '177.

Out of interest there are flip flops with different internal architecture. Some are D flip flop types, the others master-slave types and with J and K inputs. Some types are triggered by rising pulse edges, other require a full rectangular pulse and transfer data on the falling edge. It is an interesting area if you want to get into it.

If a J-K flip flop, like the 74LS107 has both its J & K inputs tied high, it is particularly reliable in changing states and performing a divide the clock pulse frequency by 2 function, better than a D flip flop with is /Q output connected back to its data input, which is another way to make a flip flop "toggle". So if you look with your scope you will see that the Q and /Q outputs of these flip flops are half the frequency of the clock pulses driving them.

 

[genobombino]

This is why I suggested checking with the meter that pin 10 was or was not connected to the other pins on the other IC's, but obviously you did not do that, it can be difficult to see on visual inspection under an IC, sometimes bright light transillumination of the board can help too.

With IC C5 eliminated from the equation now, and I gather that was not the cause of the fault, it suggests that it must be IC D5 as that is the only other IC connected to that pull up line.

But, before you think of removing it: Visually inspect all of the tracks between C5 and D5, just to make sure it all looks normal. Also, find the 1k pull up resistor R39 connected to those. There is a small possibility that the +5V line feeding it could be going, not open circuit, but going hard low. So check that when the fault occurs that there is still a solid +5v applied to that resistor.

Here is resistor R39,
Which conects to d4/8 and c5/13(which C5/13 immediately connects to c5/10)

When channel 2 goes low, thats when the screen cuts out

 

[genobombino]

On my schematic the IC D5 is another 74LS107 and the diagram of its internals matches that part. A dual JK flip flop.

Out of interest there are flip flops with different internal architecture. Some are D flip flop types, the others master-slave types and with J and K inputs.

If a J-K flip flop has both its inputs tied high, it is particularly reliable in changing states and performing a divide the clock pulse by 2 frequency function, better than a D flip flop with is /Q output connected back to its data input, which is another way to make a flip flop "toggle"

The ic in d4 is an sn74ls107an? Could that possibley be it? Its also connected to c5/10 and c5/13

 

[Hugo Holden]

The ic in d4 is an sn74ls107an? Could that possibley be it? Its also connected to c5/10 and c5/13

Interesting the recording 182437.jpg, showing the fault in evolution with the small negative going spikes, until it falls low on Ch2. It must be the IC D5 (or D4 4 on your board it would seem) doing that.

I'm not sure why the IC numbering on your pcb isn't matching the schematic, we might have the wrong schematic for your actual pcb.

Possibly the IC in position D4 is what is labelled D5 on the schematic. Check with your meter that it is definitely the one connected to the resistor R39, and the lines being pulled up on IC C5 pin 10.

If it is , then "your D4" must be the culprit.

 

[Hugo Holden]

PS: is this the schematic you have ?

http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001/320008-3.gif

 

[daver2]

Good work so far.

Hugo, I was guessing (but an educated guess nonetheless) that 320008-3 is the correct schematic (link posted in #51).

Yes, removing C5 may have been premature if you haven't done the check that both Hugo and myself suggested with a multimeter from the none +5V side of resistor R39... At least we know that the problem is not C5 now though!

Can you now do the check we asked for please and find out EXACTLY which pins of which ICs are connected to R39 (or C5 pins 10 or 13 if you like).

These early schematics DO contain errors. Sometimes by accident, sometimes 'deliberately'. So we sometimes have to do a bit of detective work ourselves BEFORE deciding to desolder anything; just in case we are being led on a wild goose chase by Commodore...

We now need to work out whether the other chip that is wired to R39 is D4 or D5, and whether there are ANY OTHER CHIPS WIRED TO R39 THAT ARE NOT SHOWN ON THE SCHEMATIC. You do this with the power to the PET switched OFF using a multimeter set to a low value resistance range. Clip one lead of the multimeter on R39 (the lead that us not connected to +5V) and use the other lead of the multimeter to probe all the pins of D4 and D5, checking for low resistance readings.

Then, follow the PCB tracks (as well as you are able) to see if any other IC pins are connected to this network.

This signal is upsetting logic that is applicable to the entire VDU sub-system; hence the observable effect elsewhere. I think you did mention this behaviour earlier on, and is why I homed in on this part of the circuitry, as it was common to all your observed issues.

Dave

 

[genobombino]

PS: is this the schematic you have ?

http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001/320008-3.gif

Good work so far.

Hugo, I was guessing (but an educated guess nonetheless) that 320008-3 is the correct schematic (link posted in #51).

Yes, removing C5 may have been premature if you haven't done the check that both Hugo and myself suggested with a multimeter from the none +5V side of resistor R39... At least we know that the problem is not C5 now though!

Can you now do the check we asked for please and find out EXACTLY which pins of which ICs are connected to R39 (or C5 pins 10 or 13 if you like).

These early schematics DO contain errors. Sometimes by accident, sometimes 'deliberately'. So we sometimes have to do a bit of detective work ourselves BEFORE deciding to desolder anything; just in case we are being led on a wild goose chase by Commodore...

We now need to work out whether the other chip that is wired to R39 is D4 or D5, and whether there are ANY OTHER CHIPS WIRED TO R39 THAT ARE NOT SHOWN ON THE SCHEMATIC. You do this with the power to the PET switched OFF using a multimeter set to a low value resistance range. Clip one lead of the multimeter on R39 (the lead that us not connected to +5V) and use the other lead of the multimeter to probe all the pins of D4 and D5, checking for low resistance readings.

Then, follow the PCB tracks (as well as you are able) to see if any other IC pins are connected to this network.

This signal is upsetting logic that is applicable to the entire VDU sub-system; hence the observable effect elsewhere. I think you did mention this behaviour earlier on, and is why I homed in on this part of the circuitry, as it was common to all your observed issues.

Dave

ERUKA!

I changed out d4, (with a nice new socket and chip) there was no debri that i could see, no crunchy or cut traces, no cracks. And afyer changing it out ive had it sat here for about a good 30 minutes w no flickering, jumps, or even a funny lookin character.

If anything decides to spite me I'll make sure to post about it, but for now i think we may be in the clear.

If this really was the fix, I can't thank you two enough. Thank you both for being so patient with me, you both were an amazing help, and this has been a great learning experience.

Hopefully ill know enough one day to pass the help on.

-Giovanni
Las Vegas, NV

 

[Hugo Holden]

ERUKA!

I changed out d4, (with a nice new socket and chip) there was no debri that i could see, no crunchy or cut traces, no cracks. And afyer changing it out ive had it sat here for about a good 30 minutes w no flickering, jumps, or even a funny lookin character.

If anything decides to spite me I'll make sure to post about it, but for now i think we may be in the clear.

If this really was the fix, I can't thank you two enough. Thank you both for being so patient with me, you both were an amazing help, and this has been a great learning experience.

Hopefully ill know enough one day to pass the help on.

-Giovanni
Las Vegas, NV

You are welcome and it looks like we have another fixed PET.

Thank you too because:

"It is the Students who help their Teachers".

When most would imagine it is the other way around. This is why teaching the great profession it is, it helps students & teachers both.

 

[daver2]

Well done for another saved PET!

I would give it a good test, as machines this old that haven't worked for many years can have a habit of developing further faults.

This particular problem is a low probability fault (in my experience) but it was a relatively simple one to diagnose from the good posts that you provided of the symptoms.

It is an interesting fault, as many people probably wouldn't have thought to test the unused inputs that are just pulled high...

You need to play a few games now. Sorry, I mean soak test the machine!

Dave

 

[Hugo Holden]

This particular problem is a low probability fault (in my experience) but it was a relatively simple one to diagnose from the good posts that you provided of the symptoms.


Dave

Yes very low, I mentioned that with this problem of pulses appearing on a 74 TTL IC's unused input pin (or any input pin for that matter) due to internal failure in the IC is very rare. I only know of the one other case of it that appeared in a Pong game. It hard to imagine what when wrong in there.

One possibility could have been a broken/intermittent connection to the power rail that supplied the IC die inside the package, causing all its inputs to draw current, and pull all those input lines close to zero volts. It seemed fairly predictable, after turn on, and that made it look like it happened regularly, when the IC heated up shortly after turn on perhaps.

We should have thought to try some freeze spray on D4 and C5 to see if the effect reversed and tried to identify it that way.

Though, there was the other interesting case where Desperado had an IC with two shorted input pins, which I have never seen before. That was a continuous short though.

The commonest input pin failure mode is where the pin simply goes open circuit and assumes a logic high inside the IC, which really fouls up the logic conditions of the external circuit, especially if it is a multi input gate looking to detect a certain state. Output pin faults are not uncommon, I have had a few where the output stage intermittently failed.