• Please review our updated Terms and Rules here

CBM3032 Mainboard Repair

Most likely a chip on that /INIT signal is dragging the voltage low.

What is the actual voltage (relative to 0V) of G5/11?

Also (out of interest) is your character generator ROM (F10) a 'real' ROM or an EPROM. The two are different when it comes to the /INIT signal on F10/21...

Dave
 
So in other words I will have to check every chip having something in common with R12 if it is pulling the voltage low(?), which are essentially the G and H rails.

The voltage on G5/11 is 0.65V ( the same as on R12. The "input" voltage of R12 is 4.97V.

I am using an EPROM for the moment in F10, but even when removing it entirely I see no difference.

Joe
 
>>> So in other words I will have to check every chip having something in common with R12 if it is pulling the voltage low(?), which are essentially the G and H rails.

Correct. /INIT appears on schematic sheets 6, 7 and 8. This is an awful job...

By taking very accurate voltage measurements of the voltage on each IC pin that /INIT connects to it MAY be possible to find the pin with the lowest voltage. This COULD be the offender.

>>> I am using an EPROM for the moment in F10, but even when removing it entirely I see no difference.

Just a thought. F10/21 on a ROM is a logic signal. The same pin on an EPROM is a power supply pin. It has been observed on some machines (when an EPROM is fitted) it drags the /INIT line low enough to cause problems (lower than a TTL logic '1' level into the grey area). If you remove the EPROM - and the /INIT signal is still low - this is not the problem.

0.65V is a very LOW level - it is being actively pulled down. I would look on the board (both solder side and component side) for any signs of foreign material causing short circuits. A naff job, but someone has to do it...

Dave
 
Last edited:
+ 1 Daver2.

On the topic of the /INIT line: One interesting thing about it, is that when the original mask style ROM is used for F10 character ROM, the pin 21 connected to the /INIT line sinks very little current. As soon as a UVeprom is put in place of F10, it sinks too much current and drags the /INIT line down to an indeterminate logic level. That is when odd things happen such as errors in the V drive timing that cause the image to bounce up and down on the VDU.

To correct this problem, if pin 21 of F10 is disconnected ,one way is to follow the track leading to it, it passes through a plated through hole. Rather than cutting the track, which is very messy, if a small drill is run through the plated trough hole, just big enough to clear the connection, it disconnects pin 21 of F10 from the /INIT line. Then you simply link pin 21 to pin 24 (+5v) of F10 to tie it high with a small length of wire. Later if you wanted to you can re-establish the connection of the plated through hole by soldering a small length of wire through the drilled out hole to the pads on each side.

The /INIT line is a strange one, in that it is not driven by a gate output. It represents a common pullup for many logic IC's. Interestingly, if you tie it high directly to +5V, it can cause the computer to malfunction on start up.

Also, if you have a logic circuit where there are multiple IC's tied to a common pullup, some very interesting things can happen if one of the IC's on the common pullup line malfunctions and injects a signal onto it. I know of only one instance where this happened , on a logic board with multiple IC's causing a complex and difficult problem. Fortunately this sort of fault is very rare.
 
Gents,

Thank you very much for your valuable inputs! I might be able to spare a few hours tomorrow to start the search for the holy grail...in the worst case I have a day or two after the weekend, will report back.

Joe
 
I started a bit of troubleshooting today, but it definitely ain't easy to find this fault. A couple of resistors (R39, R34) not having any voltage (powered through G1/13 respectively H1/6), and the cause probably being the INIT line being pulled low somewhere. After the weekend I will start measuring individual voltages of pins from ICs connected to the INIT line.

Joe
 
Yes, if the /INIT line is LOW then this will mess up most of the circuitry on this schematic sheet!

Yes, to find this fault will be a right royal pain I am afraid - so gather yourself for a fight with the PET!

Dave
 
I have a second board ready to take some comparative measures if needed; I wish I could take a week off from work to tackle this beast!

Joe
 
There is something you could do, perhaps not a textbook recommended method, but I will explain how it works by analogy.

Many years ago fixing AM transistor radios, it was noticed that sometimes some sort of conductive material would get between the aluminium vanes of a tuning capacitor. It could be nearly impossible to see where it was , especially if the fragment was between extremely closely spaced vanes, and at times shorting them out.

Somebody came up with the idea of connecting the capacitor across a power supply with a low internal resistance, when the particle shorted the vanes out, it got blown to pieces, much like an insect in a bug zapper, I am sure you can see where this is going.....

Any IC's that are normal, connected to the /INIT line, won't draw excessive currents on any of their input pins, and won't mind being temporarily pulled to hard 5V supply. But the failed IC will, it will likely draw more current and probably the short or leakage pathway in it might get vaporized, or at least it might heat up and you could spot the defective IC that way. If connecting the /INIT line to +5V directly (ie short out the pull up resistor) cleared the problem, at least then, most of the computer would likely start working and you could fault find it better to locate the defective IC. If it doesn't clear the abnormal leakage pathway, nothing is lost with the experiment as no normal chips connected to that line will be harmed.

Now one thing to consider before doing anything is to make the theoretical statement and then test it: "All chips connected to the /INIT line are perfectly normal" . Now if this is true, what could be wrong that could cause all/many of the chips to draw current from the /INIT line ? The answer to that is power supply failure, if one or more chips is not getting its +5V supply. So before doing anything else, I would recommend identifying every chip on the /INIT line and checking on the actual power pin entering its package, with the meter, that it is definitely being powered. If they all definitely powered, only then move to the bug zapper method. Of course an internal failure in an IC, if the die loses its power, will cause the IC to start sinking significant current at its input too. Especially check any IC in a socket, because if the IC's power pin failed to connect that would do it.

(If a chip loses it + 5V power, it can get powered indirectly from its input pins via substrate diodes or diodes intended to prevent the input pins going significantly above the +5v rail)
 
Last edited:
There is something you could do, perhaps not a textbook recommended method, but I will explain how it works by analogy.

Many years ago fixing AM transistor radios, it was noticed that sometimes some sort of conductive material would get between the aluminium vanes of a tuning capacitor. It could be nearly impossible to see where it was , especially if the fragment was between extremely closely spaced vanes, and at times shorting them out.

Somebody came up with the idea of connecting the capacitor across a power supply with a low internal resistance, when the particle shorted the vanes out, it got blown to pieces, much like an insect in a bug zapper, I am sure you can see where this is going.....

Any IC's that are normal, connected to the /INIT line, won't draw excessive currents on any of their input pins, and won't mind being temporarily pulled to hard 5V supply. But the failed IC will, it will likely draw more current and probably the short or leakage pathway in it might get vaporized, or at least it might heat up and you could spot the defective IC that way. If connecting the /INIT line to +5V directly (ie short out the pull up resistor) cleared the problem, at least then, most of the computer would likely start working and you could fault find it better to locate the defective IC. If it doesn't clear the abnormal leakage pathway, nothing is lost with the experiment as no normal chips connected to that line will be harmed.

Now one thing to consider before doing anything is to make the theoretical statement and then test it: "All chips connected to the /INIT line are perfectly normal" . Now if this is true, what could be wrong that could cause all/many of the chips to draw current from the /INIT line ? The answer to that is power supply failure, if one or more chips is not getting its +5V supply. So before doing anything else, I would recommend identifying every chip on the /INIT line and checking on the actual power pin entering its package, with the meter, that it is definitely being powered. If they all definitely powered, only then move to the bug zapper method. Of course an internal failure in an IC, if the die loses its power, will cause the IC to start sinking significant current at its input too. Especially check any IC in a socket, because if the IC's power pin failed to connect that would do it.

(If a chip loses it + 5V power, it can get powered indirectly from its input pins via substrate diodes or diodes intended to prevent the input pins going significantly above the +5v rail)
Dear Hugo,
Thanks for your input!!

I can very well visualize the idea of a bug zapper :-)
Funnily I just today took my schematics out again and am trying to formulate a plan regarding further troubleshooting of the (beasty) PET.
I will beg my wife today to allow me an hour or two in the basement with the PET and will definitely try your idea; the alternative would probably be to de-solder each IC eventually, but this would change the PET's originality considerably, so I will try other options first.

As I consider myself an absolute beginner with PETs and electronics, it gives me also an excellent opportunity to read schematics & visualize connections between the various ICs.

I will definitely report back with an update....

Joe
 
>>> the alternative would probably be to de-solder each IC eventually, but this would change the PET's originality considerably

And potentially damage the PCB in the process...

Dave
 
Dear Hugo,
Thanks for your input!!

I can very well visualize the idea of a bug zapper :)
Funnily I just today took my schematics out again and am trying to formulate a plan regarding further troubleshooting of the (beasty) PET.
I will beg my wife today to allow me an hour or two in the basement with the PET and will definitely try your idea; the alternative would probably be to de-solder each IC eventually, but this would change the PET's originality considerably, so I will try other options first.

As I consider myself an absolute beginner with PETs and electronics, it gives me also an excellent opportunity to read schematics & visualize connections between the various ICs.

I will definitely report back with an update....

Joe
.....don't forget to check all the IC's power supplies before trying the bug zapper method, there is still the possibility that no IC's are defective and one has simply lost power.

I don't think it will be necessary to unsolder all the IC's. If you got forced into that, ideally it would be with the pin isolation method; You solder suck just the single pins that are connected to the /INIT line, if done well, mostly you can move the pin back and forth in the hole and cause it to disconnect from the plated through hole, check with the meter. Applying fresh multicore solder helps prior to sucking. Whatever you do , don't apply any liquid fluxes to the pcb. The best Lead/Tin Multicore (5 core) solder is made by Ersin Multicore (Loctite in the USA). Don't use Lead free solder. That way it limits the potential of major damage to the pcb. I would not resort to global unsoldering of whole IC's, I agree with Daver2.
 
Last edited:
...Just returned from an overseas assignment, going to continue working on the mainboard over the next couple of days...unless there was some magic self-healing over the past couple of months:-)

Joe
 
Back
Top