• Please review our updated Terms and Rules here

Trs-80 model 1 flaky boot / possible crystal problem

M

mrthreeplates

Experienced Member
Joined
Jul 22, 2018
Messages
77
Hi everyone,

I have a M1 that I've been trying to get working a bit more reliably. The latest issue I'm chasing is a flaky boot issue. The M1 boots intermittently.

It is a rev-G board, with relatively few mods.

What I've noticed is that when it fails to boot, I get garbage on the screen (sometimes with the video sync'ed, sometimes not). When it fails to boot, hitting reset doesn't do anything. When it does boot, it stays running (reset works as it should). When it runs, I'm able to pass memory tests. I also have an EI, but that doesn't seem to affect the boot issue.

I've checked the voltages, and they are all good. (5v, 12v, -5v) - even when it fails to boot.

So, I chased the problem down to the cpu clock. Interestingly, when it fails to boot, the cpu clock is ~5Mhz(!) (should be 1.7). Chasing this back to the crystal, I found it was oscillating at 30Mhz (which makes sense as there is a divide by 6 feeding the cpu).

Obviously, when it boots correctly, the cpu clock is normal (per my scope).

I checked the resistors and load capacitor around the 10.6445Mhz crystal, and they are fine. Resistors measure nearly spot on at 910ohms, and the capacitor measures (out of circuit) 44pF (should be 47pF, but I assume this is close enough). I haven't checked the LS04 around the crystal, that is next on my list.

I know very little about crystals, but does it seem plausible that a bad crystal would mysteriously vibrate at a 3x higher frequency - only occasionally?

If anyone knows of ways to test crystals out of circuit for this, I'd love to actually prove the crystal is bad. Or perhaps I need to modify the values of the passive components to make the crystal more stable?

Otherwise, if it is a bad crystal, I'll have to hunt for a new one.

Thanks!

George
 
Thanks for the suggestion, and lead on the crystal. I'm continuing to debug the problem and will report back!

Much appreciated
 
I’m sure you’ve probably already checked this, but have you stuck your oscilloscope on the PSU outputs, by any chance? I have a basket case Model I I’ve been slowly poking away at, and the first big problem I found was the 5v supply was oscillating like mad because of some bad analog components in the voltage regulation. The main visible symptom was crazy horizontal sync problems, the CPU mostly ran “normally”… which is amazing considering the machine was basically running on high-frequency half-wave AC.

The voltages looked perfectly fine to a dumb meter, of course. Maybe a problem like this could do weird things to the clock circuitry as well.
 
Thanks again for the suggestions: I did some more poking around based on the suggestions above. The PSU outputs have very low ripple (a few mV), so I don't think that is the problem. I carefully reflowed the solder on the crystal (twice), but it didn't make any difference. As far as I can tell, there weren't any cracks beforehand.

With my scope on the output of the crystal circuit, I get 10.6Mhz (before the flip-flop which I think divides it in half). When I get 10.6Mhz it boots correctly. However, most of the time, I get ~31Mhz. Of course, the M1 won't boot and the screen just displays noise. If I slightly touch the crystal, the frequency snaps back to 10.6Mhz (and the M1 finishes booting). FWIW, the problem seems to be getting worse. I can only get it to boot maybe 1 out of 10 attempts. Touching the crystal seems to fix the problem every time. This is *after* I tried reflowing the solder on the crystal. The crystal is mounted close to the board, so there aren't any exposed leads on the PCB side.

Does this sound like a bad crystal? Or, I wonder, if there is something else causing the crystal circuit to not lock on its main frequency?

@Chromedome45, I'd love to get a spare crystal from you to try out.

Thanks!
 
Touching the crystal fixing it is a pretty bad sign it's your culprit. You could *try* messing with the other components of the resonator circuit, but it sounds like you've already checked them out and it's kind of a stretch to see how they would cause these symptoms.

I had no idea this part was so hard to get today. 10.7 mhz crystals (3x NTSC colorburst) aren't easy to find but are at least out there(*), this 10.6445 doesn't seem like a simple multiple or division of any "common today" speed. Here's an outfit in the UK that apparently custom ordered a batch of them.

(* I think maybe I'd actually just assumed this *was* a 3x colorburst crystal and hadn't done the math.)
 
Sounds like the actual crystal (inside the crystal case) is cracked in half, so vibrating at twice the frequency. Get a new one if you can. Even if it makes no difference it will clear up one possibility. Also, check all the components in the oscillator circuit, by replacement if necessary.
 
Well, I have a new crystal to test which is spot on at 10.6445 mhz (thank you Frank!). Unfortunately, it didn't fix the issue. When I removed the suspect original crystal, it checked out fine in my crystal checker. And replacing it with the new one made no difference.

So, I figured the next thing to try is to increase the capacitance of c43 (since it was reading a little low @ 44pF but probably still within tolerance). I had a 22pf capacitor handy and put it in parallel with c43. Well, after I did that my flaky boot issues seem to have largely gone away. After about 40 or so attempts, I only got it to fail once. Even then, the one failure looked completely different than when the crystal was oscillating too fast (which I can no longer replicate).

So, I think I I've made some improvements, but am not sure it is entirely fixed yet. More debug and testing in progress...
 
Getting there, one step at a time. Sounds like it's almost fixed.
 
Well, since adding the additional capacitor, my model 1 has booted reliably. No more failures after many attempts. I can only assume it was caused by c43. So, I'm calling it fixed for now and buttoning it up. Thanks for the help and suggestions!

On to the next project!
 
You must log in or register to reply here.
Back
Top