PDA

View Full Version : A2000CR Rev 6a Motherboard - intermittent boot issues



dhoelzer
October 2nd, 2014, 07:22 PM
Hello!

Yet another A2000 in need of assistance. If I let it sit for a few days it will happily boot to a floppy, possibly even more than once... Next time it starts I will leave it running and see if it will run for a long period of time; that is unknown at this time.

After it boots properly once or twice, it switches to black screen only; no drive clicks. Keyboard caps lock still flickers and Ctrl-A-A causes the keyboard to flicker again so there's someone home but no floppy, black screen.

Checked/done so far:

* All custom chips and CIAs check out just fine in a 500 and another 2000.
* I just completely recapped the board - Thought that fixed it when it booted three times but I'm now back to a black screen.
* I have tried multiple known good power supplies
* At the moment it is a bare board on a workbench - No aux cards connected.

Any ideas? I was trying to locate the reset circuit as a possible source of problems but I can't seem to figure out what _RST is wired to on the 68000 based on the schematics. It seems like there's a page or two missing.

dhoelzer
October 2nd, 2014, 07:55 PM
A perusal of the 68000 docs and a voltmeter reveals that both _RESET and _HALT are being held low. Any pointers to what drives the reset circuit on these? The other Commodore stuff I've worked on all had 555 circuits driving it.

dhoelzer
October 3rd, 2014, 05:56 AM
I'm looking at the LM339 and here's what I'm finding:

Pin 1 is +5.02 (Output #2)
Pin 2 is 0 (output #1)
Pin 4 is +5.02 (- #1)
Pin 5 is +2..86 (+ #1 - this causes the low pin 2)
Pin 6 is +0.74 (- #2)
Pin 7 is +2.86 (+ #2 - this causes the high pin 1)
Pin 8 is +4.3 (- #3 - causes pin 14 low - this is the RESET signal as far as I can tell)
Pin 9 is +2.86 (+ #3 - With 8, causes pin 14 low)
Pin 10 is 0 (- #4)
Pin 11 is +3.6 (+ #4 - causes pin 13 high)
pin 13 is +4.3 (output #4)
pin 14 is 0 (Output #3 - RESET signal to 68000?)

What I can't see is what would drive RESET high. Is Pin #6 actually connected to the keyboard clock line? That seems strange if it is because it seems like that would turn this into a huge buffer for that clock line. Unfortunately, I seem to be too braindead to see where that _KBCLK is going on the other sheets.

- - - Updated - - -

Ok. If I tie pins 6 and Pin 1 together (driving pin 6 high with the output from pin 1) I can actually cause the system to function.

I'm a bit uncomfortable with this because I feel like there's a reason that reset is staying low and jimmying the reset circuit like this feels dangerous. Especially since I don't know what I'm doing here, I'm really nervous about just soldering in a jumper.

Please, any thoughts or advice?

dhoelzer
October 3rd, 2014, 11:16 AM
Just an update.

Jumping pins 1 and 6 will allow the system to boot very reliably. All this is really doing is pinning the reset line high by overriding the comparator.

Doing so, however, results in behavior that looks like the quote key is mashed down. -that- is likely the actual problem.

This happens with a keyboard connected or disconnected. So, the new question is, where can this problem come from, the CIA chips?

geoffm3
October 3rd, 2014, 11:41 AM
Just an update.

Jumping pins 1 and 6 will allow the system to boot very reliably. All this is really doing is pinning the reset line high by overriding the comparator.

Doing so, however, results in behavior that looks like the quote key is mashed down. -that- is likely the actual problem.

This happens with a keyboard connected or disconnected. So, the new question is, where can this problem come from, the CIA chips?

IIRC the CIA is part of the keyboard interface on the A2000. Why not swap the two CIAs and see what that gets you?

dhoelzer
October 3rd, 2014, 11:46 AM
Never mind. The quote is spurious and only happens occasionally. Here's where we are:

I can boot it every time if I hold pin 6 high in the reset circuit comparator. However, the keyboard clock is definitely going through here too and holding the pin high prevents the keyboard from functioning properly.

I can sometimes send pin 6 high to flip the reset/halt lines high and they will stick high, allowing me to boot the system and the keyboard functions just fine.

The trouble is that, most of the time, the system will lock up (light grey screen), taking me back to the point of having to power off and on.

dhoelzer
October 3rd, 2014, 11:47 AM
Swapped them both out. No change.


IIRC the CIA is part of the keyboard interface on the A2000. Why not swap the two CIAs and see what that gets you?

geoffm3
October 3rd, 2014, 12:02 PM
I just had a look at the schematic. So the comparator is only looking at the keyboard clock as an input, presumably because the keyboard will stop the kb clock from wiggling for some set time to trigger the reset condition. You are shorting out the kbclk by shorting 1 and 6, so that may be affecting the data being clocked by the CIA.

Perhaps the issue are the two electrolytic caps C318 & C319 in that comparator circuit.

geoffm3
October 3rd, 2014, 12:06 PM
The purpose of this circuit is to both provide a keyboard reset function, and to also act as a voltage monitor to hold the system in reset long enough for the power supply to become stable.

dhoelzer
October 3rd, 2014, 03:00 PM
The board was recapped completely last night. This problem existed before that.

I'm seriously thinking about what the effects would be of cutting the trace out of the reset circuit and letting the pull-ups keep HLT and RST high. I'm trying to research whether that would break anything other than the keyboard reset shortcut.

If I tie that line high the system will boot and run.. It will typically continue to run for a few minutes after that too, until the reset circuit goes low again and holds HLT and RST low, stopping everyone and giving me a gray screen.


I just had a look at the schematic. So the comparator is only looking at the keyboard clock as an input, presumably because the keyboard will stop the kb clock from wiggling for some set time to trigger the reset condition. You are shorting out the kbclk by shorting 1 and 6, so that may be affecting the data being clocked by the CIA.

Perhaps the issue are the two electrolytic caps C318 & C319 in that comparator circuit.

dhoelzer
October 4th, 2014, 03:05 PM
Just confirming that the resistor pack is fine (RP802).

I really am stumped as to where to go from here to identify the fault. It's clearly something in the reset circuit and the only outside source that I see in it is the keyboard clock, which is mystifying to me; there is no other outside source into this circuit except for the pull ups and pull downs.

What exactly is the purpose of the 1n4148 in that arrangement? It's pretty much the only component left in the circuit that hasn't been pulled out.

dhoelzer
October 4th, 2014, 06:06 PM
How many ohms would I expect the 1n4148 to read as forward resistance? Reversed it's open (correct), forward it's reading about 12k ohms... If I use the diode test mode that drops down to about 660 ohms (best I can tell... it actually reads .660 and a 1k resistor on diode mode reads .980, so...)

Agent Orange
October 4th, 2014, 07:58 PM
How many ohms would I expect the 1n4148 to read as forward resistance? Reversed it's open (correct), forward it's reading about 12k ohms... If I use the diode test mode that drops down to about 660 ohms (best I can tell... it actually reads .660 and a 1k resistor on diode mode reads .980, so...)

Rule of thumb is @ 10 to 1 - front to back.

dhoelzer
October 5th, 2014, 06:51 AM
Rule of thumb is @ 10 to 1 - front to back.

Err... Can you translate that for the electronically challenged?

dhoelzer
October 5th, 2014, 08:25 AM
I'm still interested in what Agent Orange was trying to tell me, but...

I may have managed to repair the board. It's up and running right now... Time will tell if I've nailed it.

Apparently the 1n4148 and the filter cap (glass encapsulated ceramic .1 uF) were both marginal, leading the keyboard reset circuit to hold low incorrectly.

I'll let the system run for a few hours and see what happens, but so far it's started up properly twice, keyboard reset functions and shows now signs of locking up.

Thanks to those who tried to help out. :)

Agent Orange
October 5th, 2014, 01:07 PM
Err... Can you translate that for the electronically challenged?

With the diode disconnected or out of the circuit, read the forward resistance (in ohms) through the diode. Don't load the circuit with your body (hands). Record the resistance. Now reverse the leads on the diode and read the reverse resistance (you may need to increase the resistance scale on your meter). The difference in the resistance (front to back) should at least be a ratio of at least 10:1. Same can be said for most PNP/NPN junctions; i.e. diodes, transistors, etc.

dhoelzer
October 5th, 2014, 01:26 PM
Ah! Thank you!

Alas, it is not fixed. Behavior improved but has degraded again.

I think I've narrowed it down to a very simple circuit that runs between the LM339, the 8250 and the Keyboard Clock line. The kb clock line sits at about 1 volt with the CIA removed. The KB clock line bounces a bit between 1.6 and 1.9 but is consistently below 2 volts.

With the CIA, every now and then the KB clock line will decide to come up to 4.8 to 4.9 volts and the system will instantly begin to function properly. This can happen with or without the keyboard connected. Not surprisingly, the system will not run without this CIA connected (it can run without the other but can't get to the floppy, of course)

Question: What could be holding this down even with the keyboard disconnected?? The only thing left in the circuit is the output of the comparator, the CIA on Pin 40 and a resistor pack that's a pull-up to 5 volts. I've checked the RP and it seems to be just fine.

Agent Orange
October 5th, 2014, 02:49 PM
Ah! Thank you!

Alas, it is not fixed. Behavior improved but has degraded again.

I think I've narrowed it down to a very simple circuit that runs between the LM339, the 8250 and the Keyboard Clock line. The kb clock line sits at about 1 volt with the CIA removed. The KB clock line bounces a bit between 1.6 and 1.9 but is consistently below 2 volts.

With the CIA, every now and then the KB clock line will decide to come up to 4.8 to 4.9 volts and the system will instantly begin to function properly. This can happen with or without the keyboard connected. Not surprisingly, the system will not run without this CIA connected (it can run without the other but can't get to the floppy, of course)

Question: What could be holding this down even with the keyboard disconnected?? The only thing left in the circuit is the output of the comparator, the CIA on Pin 40 and a resistor pack that's a pull-up to 5 volts. I've checked the RP and it seems to be just fine.

It sounds like you may have an intermittent issue with a component somewhere. You may want to try a can of circuit freeze in an attempt to isolate the faulty part (Mouser or any electronics supply). I would go over any suspected circuits with a soldering iron in order to reduce the possibility of cold solder joints. Also, if possible, swap out your keyboard just to eliminate it as problem source. While troubleshooting, reduce your system components to bare minimum.

dhoelzer
October 5th, 2014, 04:40 PM
Thanks Agent Orange. :) I've already eliminated the keyboard as the source of the trouble... The problem exists whether the keyboard is connected or not. :)


It sounds like you may have an intermittent issue with a component somewhere. You may want to try a can of circuit freeze in an attempt to isolate the faulty part (Mouser or any electronics supply). I would go over any suspected circuits with a soldering iron in order to reduce the possibility of cold solder joints. Also, if possible, swap out your keyboard just to eliminate it as problem source. While troubleshooting, reduce your system components to bare minimum.