• Please review our updated Terms and Rules here

PET 3032 diode repair

carlsson

Veteran Member
Joined
Jul 30, 2003
Messages
6,274
Location
Västerås, Sweden
One of my remaining PETs gives me sparkles and smoke. It is the one I previously lifted ROMs and big chips from to verify they all seem OK. I spotted the sparkles come from the diode rectifier (?) bridge in the power supply department, and planned to replace at least the four 1N5402 diodes that make up the bridge. One of them has even visual cracks in the plastic!

However after closer inspection I saw there has been quite a bit of corrasion on the board, and it didn't get any better after I desoldered the old diodes. My question now is if this is salvageable, and if so how do I move on? Are those motherboards just dual layer? I have attached an image to illustrate the damage as it lies right now. Assuming rest of the board might work, it'd be a bit of a shame to dump a PET just due to this issue.
pet-diodes.jpg
 
I'd say that's salvageable. I burned up a PET 4032 in a similar manner, and was able to just tack the leg onto a trace. In your case, you might be able to jump them back and forth, the lands look to be vcc or gnd planes? And I'm fairly well positive the boards are simple double-sided.

new-coil.jpg
 
These boards are dual layer boards only. You could scratch some solder mask away with a cutter knife to get access to some "good" copper again.
Maybe it would be useful to replace the 1N5402 diodes (3A max) by MUR420 (4A max).
 
Doesn't look bad at all, just clean it up and add a couple of wire jumpers to replace the burnt traces. Probably was just a failed diode, but wouldn't hurt to check that the regulators are all good.
 
Diodes replaced (and it became a rather ugly job) but no dice. Well, no more cracks but I get a display of garbage that goes away after a second. I've brought my known good PET and will run a series of chip swaps later this week to ensure it isn't a ROM or "big chip" issue. But as MikeS suggested, there might be other components in the power supply section that would need replacing as well. The fact it displays garbage for a second and then goes black makes me wonder if e.g. the big capacitor wants to be replaced?
 
The fact it displays garbage for a second and then goes black makes me wonder if e.g. the big capacitor wants to be replaced?

Anders,
You should check on all DC power, but it may be OK. The fact that the screen blanks after an initial garbage screen means that the CPU is running in program for a little while and clears the screen properly before going 'south'. That may mean a bad ROM or bad lower RAM.
 
I've performed the following tests:

1. Move the complete Basic 2.0 ROM set + all five big 65xx chips from the broken 3032 to the working 3032. It works like a charm, computer boots up with Basic 2.0 as expected.

2. Power on the broken 3032 without any ROM or 65xx chips at all. It gives me the garbage screen, but it remains stuck.

3. Insert the 65xx chips but leave out the ROMs. The garbage screen remains.

4. Re-insert the ROM chips. It makes the garbage screen display for a second, then go blank.

Thus, I suspect RAM problems. I'll go searching this forum for some small test program that can replace the $F000 ROM and at least write intelligent values to video RAM $8000, perhaps some more qualified RAM testing.
 
Eudimorphodon's RAM test shows a screen of "b" (as in bad) followed by a screen of the character set repeated. I take it as a RAM fault somewhere. Now I just need to figure out where and how much.
 
Yup, will do.

While discussing voltages, I'll take the advantage to hijack my own thread. The other 3032, the one that works most of the time, has a tendency to blow fuses. On the back side it is specified 0.5A, so I put in a slow blow 500 mA fuse. Usually it is good for many power cycles but sometimes like yesterday the fuse goes after just a few cycles. On the malfuctioning 3032, there is a slightly different power supply also labeled as 0.5A but its glass fuse is rated at 750 mA. While I have a rather good supply of slow blow fuses, I thought it is a shame if I run out of 500 mA ones while all the other dimensions remain unused. Yesterday I took a risk and inserted a 630 mA one and the PET seems to power on like normal. I understand this is the maximum power spike (?) before the fuse kicks in order to protect the motherboard and monitor.

Would you think I'm putting it to a risk by using a slightly beefier fuse? I understand these power supplies (or whatever one wants to call it) are 30+ years old and have been through a lot including possibly stored in a cold garage for a decade or more. Is there any simple tips to test the condition and spot any possible problem? Should there be a matter that they once were manufactured for 220V but wall outlets now rather are nominal 230V?
 
Would you think I'm putting it to a risk by using a slightly beefier fuse?

As you can imagine, it is difficult to say it is OK to use a fuse that is rated at a higher value than specified, but the .63A one you chose may be OK. As you know, 30 year capacitors get leaky and the input surge goes up. Realize that one day all old caps will go 'poof'. Keep an eye out for a good price on replacement capacitors.


Should there be a matter that they once were manufactured for 220V but wall outlets now rather are nominal 230V?

No issue there.

You can minimize power cycles by adding a Reset Switch or performing a SYS command to the cold start routine. Find the 16 bit Reset Vector address that is located at $FFFC and $FFFD. On my BASIC 4 machine it is SYS 64790.
 
Yep. I routinely SYS or CALL the reset vector on all my 6502 based computers. However in case the machine froze, the (often not yet installed) reset button would be the only choice.

Anyway, I'll get back to this machine and make some measurements later on, now that it appears that either every single RAM chip except the video RAM is toasted as a result of the busted diodes, or something completely else is wrong that makes it behave like RAM is not connected.
 
After checking voltages (the dynamic RAM chips are the only things that use the -5 and +12v regulators so if one of them blows everything else will still work) if you have a 'scope handy I'd poke around I11 and I10, the data buffers in front of the whole RAM bank, and make sure you're seeing interesting activity when you're running the RAM tests. (With my PETtester, and presumably the one you wrote, you should in particular see regular bursts of action on pins 1 and 19, the "read" and "write" enable lines, respectively. On PETtester they'll correspond with the instant before the screen fills with "b"-s.)

On the subject of voltages and RAM chips, the friend that worked on my strange Dynamic-board 2001 ended up replacing the bank of bad 4108s with 64k 4164 chips, after very carefully disabling the -5 and +12v regulators. (They did a lovely job. You have to look really sharp to notice anything non-stock.) Apparently the 64k chips draw only about a 1/4 of the power of 4116s so it's not a terrible idea to consider if you want to be extra kind to the power supply.

(Of course, I've been pondering the difficulty of modding the circuitry so I can use 32k of the single bank instead of just 16k of it, but... If I want the full 32k in this system I'll probably resort to hanging an SRAM chip off the expansion bus.)
 
Last edited:
Back
Top