PDA

View Full Version : PET won't power on



commodorejohn
April 30th, 2017, 02:25 PM
So I picked up a PET 4016 that had been sitting in a storage unit for God knows how long yesterday. Opened it up and things look pretty good overall (lotta dust, but no rust or capacitor leakage.) But when I plugged it in and flicked the switch, nothing happened - even the monitor doesn't turn on. Checked the fuse and that's good, so I'm assuming the PSU bought it. What should I look for to troubleshoot this? I've heard advice that some older PSUs shouldn't be disconnected for troubleshooting as they'll have further issues without a load placed on them - does the PET fall under this category?

MikeS
April 30th, 2017, 02:40 PM
A lot of things could make it appear dead other than the power supply; unfortunately it's rarely an easy power supply fix.

The way it is integrated and split up with the capacitor off board you'll have trouble testing it while disconnected; assuming you have a meter you can check the raw AC voltages at the transformer while disconnected, but for the DC voltages you'll have to connect it.

m

AndyG
April 30th, 2017, 10:40 PM
Hi

Is the 4016 a 9inch or 12inch version. I ask as I have a 12 inch which didn't power up. Could be lost of things but found that the Europeans ones, at least, have a small silver mains filter (to reduce interference with TVs etc) that was mounted underneath the transformer and had popped. I bypassed it and all was well (apart from memory issues).

If not that, does the transformer output any voltage to the board ? Also check for blown capacitors on the board on the voltage regulator side. Zimmers website have great circuit schematics to help you.

Cheers

commodorejohn
April 30th, 2017, 10:58 PM
It's the 9" model. I haven't gotten around to serious testing, but I did pull the board out and take a better look at it - none of the caps on the main board look blown, but I don't actually know how to test them aside from a basic continuity check with a multimeter (which they passed.) Haven't checked voltages out of the PSU or anything yet.

KC9UDX
April 30th, 2017, 11:29 PM
You won't fry the power supply by running it unloaded; it's a good olde fashioned linear supply. Do check it first.

The monitor won't get a raster if the logic board (PET computer part) is not doing what it's supposed to, which it likely isn't. But do start with checking the power supply. In the meanwhile, it may be best to unplug the connector on the mainboard that leads to the monitor.

I wish I had time to write a summary of the procedures we end up using here for PET repair. Start with the power supply, next probably reseat all the socketed chips, then check the clock generator.

If you have close neighbors, for their sake, please don't run without that mains filter, if you can avoid it (that is, replace it when it dies), as running without it can cause unlawful (here in the States, but still most likely won't get you in trouble) at worst, and discourteous at best radio interference. These days, dwellings are teeming with devices that do this, but it's still best to squash it wherever you can. Depending what kind of wireless devices you have (including digital TV), generating RF interference could be wreaking havoc for yourself, and you might not even know why.

AndyG
May 1st, 2017, 02:24 AM
Ok the 9inch ones don't have the extra filter ... so guess it will be down to basic detective work. This forum has lots of examples of where to start looking and reseating chips and measuring the voltages are correct is a start .... wish could be more helpful.

I run my PET's off dedicated mains filters to reduce the effects of mains spikes and interference.

commodorejohn
May 1st, 2017, 05:58 AM
Well, the only socketed chip is the 6502 itself, and removing that and reseating it didn't solve the problem, so yes, I guess it's down to detective work...

If nothing else, this'll be a good learning experience :lol:

KC9UDX
May 1st, 2017, 06:20 AM
Must be a very late model!

commodorejohn
May 1st, 2017, 06:26 AM
Late enough that it's one of the ones where they punched holes out of the board by the additional RAM sockets because people were upgrading the RAM themselves rather than buy a 32KB model :D

KC9UDX
May 1st, 2017, 06:43 AM
Simple enough to get around that! I always wished (back when those were still commonly in use) that I could afford one so I could add RAM to it.

commodorejohn
May 11th, 2017, 09:57 PM
Okay, I've had a chance to poke around with it a little bit. Good news is that the power supply is at least working enough to get a stable raster on the monitor - no actual picture, but I can see the beam retrace patterns in the absence of one, and while it's somewhat distorted it's recognizable and stable. Of course, part of the issue might be the wasp's nest that's sitting atop the width coil :lol:

I haven't checked the voltages, though - which pins on the power-supply connector should be supplying which voltages? I have the 4016/4032 reference manual, but it's not very clear on this.

iz8dwf
May 12th, 2017, 02:49 AM
Okay, I've had a chance to poke around with it a little bit. Good news is that the power supply is at least working enough to get a stable raster on the monitor - no actual picture, but I can see the beam retrace patterns in the absence of one, and while it's somewhat distorted it's recognizable and stable. Of course, part of the issue might be the wasp's nest that's sitting atop the width coil :lol:

I haven't checked the voltages, though - which pins on the power-supply connector should be supplying which voltages? I have the 4016/4032 reference manual, but it's not very clear on this.

Are you sure it's a 4016? To me looks like a 3016 PCB (2001N universal dinamic-RAM PCB with first bank of 16K installed only), the one without CRT controller.
If you see a completely black screen on that board, the CPU and kernel are at least partially working.
You should really first clean the main and monitor PCBs from all the dust and from all foreign stuff.
If I'm not wrong on the actual PCB revision, I've recently completely restored one, so I can guide you on troubleshooting. But you need to have some test equipment (at least an oscilloscope, eprom reader and good desoldering tools).

HTH
Frank IZ8DWF

dave_m
May 18th, 2017, 08:21 AM
You get a raster because, in your hardware, the video timing is hardwired and does not need the CPU running in program. So the problem could be bad ROM, RAM, clock, etc. With your meter, measure the voltage of the 1 MHz clock. If you get about 2.4 V then your clock is running.

Without a scope or at least a logic probe, your best bet may be a RAM/ROM Replacement board that fits in the 6502 socket. If you want to learn troubleshooting, get some test equipment and the guys here will help you through this checkout.

commodorejohn
May 18th, 2017, 09:43 AM
Are you sure it's a 4016? To me looks like a 3016 PCB (2001N universal dinamic-RAM PCB with first bank of 16K installed only), the one without CRT controller.
Well, the case says 4016, but I don't really know from PET internals. Are there any identifying markings I should loook for on the board?

I do have some test equipment (scope, EPROM/EEPROM reader/burner, $5 multimeter,) but I'm a novice at using an oscilliscope for anything meaningful.

MikeS
May 18th, 2017, 10:16 AM
Well, the case says 4016, but I don't really know from PET internals. Are there any identifying markings I should loook for on the board?
It can be confusing: a 4016 can be either one of two quite different models:
If it has a 12" screen then it is a 'FAT40', which uses a CRT controller to generate the video.
If it has a 9" screen (yours?) then it is a 3016 (aka 2001N-16) with BASIC 4 installed, which generates video with discrete TTL ICs.


I do have some test equipment (scope, EPROM/EEPROM reader/burner, $5 multimeter,) but I'm a novice at using an oscilliscope for anything meaningful.
You can check all the voltages (except the cassette motor) at the following pins of any one of the 4116 RAM memory chips:
pin 16: GND
pin 1: -5V
pin 9: +5V
pin 8: +12V

commodorejohn
May 18th, 2017, 10:25 AM
Yeah, I haven't taken a ruler to the CRT or anything, but I'm pretty dang certain it's the 9" model. Thanks for clearing that up - and thanks for the tip on where to check the voltages, I'll pull out the multimeter this evening and see what I can find out.

iz8dwf
May 18th, 2017, 11:18 AM
Well, the case says 4016, but I don't really know from PET internals. Are there any identifying markings I should loook for on the board?

Yes, it's a 3016 upgraded to 4016 (2001-N dynamic ram board, with upgraded basic 4 ROMs). I have the same PCB, with basic 2 and 32K RAM.
I know quite well this machine by now.



I do have some test equipment (scope, EPROM/EEPROM reader/burner, $5 multimeter,) but I'm a novice at using an oscilliscope for anything meaningful.

Wonderful! You should first get the correct schematics here:
http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/index.html

Then, with the scope, check the clock input to the 6502 first (use a 10:1 probe with vertical scale set to 0.2V/div, time base around 100ns/div or so). Then check all the ROMs with your programmer, images are here:

http://www.zimmers.net/anonftp/pub/cbm/schematics/computers/pet/2001N/index.html

All but two are 2332 (configure your EPROM programmer for a 2532, it will read fine). Two are 2316 (EPROM programmer set for 2716): the character generator and the editor ROM.
You can quickly power off the on the machine to see if it starts with a screen full of random characters then it clears. If it does that, it's probably executing fine the kernel ROM at least.
You may want to look at what I did to repair my PCB:

https://www.youtube.com/watch?v=H5_mTETwWJk

The RAM diagnostic kernel that I used is this one:
http://www.zimmers.net/anonftp/pub/cbm/firmware/computers/pet/other/petester.bin
You can burn it into a 2532 (not 2732, or use an adaptor!)

In your PCB it should be perfect since it only tests the first 16K bank.
Keep in mind that the white sockets on that PCB are really unreliable. I had to change a couple of them because they wouldn't work even after pulling, cleaning and re-inserting the chips a few times. Then I changed almost all the ROM ones, just to be safe.

You may also want to cheat and install a RAM/ROM replacement board into the 6502 socket ;-)

HTH
Frank IZ8DWF

MikeS
May 18th, 2017, 02:13 PM
Oops; GND is on pin 16, not 14; fixed post #15.

commodorejohn
May 18th, 2017, 02:26 PM
Duly noted!

commodorejohn
May 18th, 2017, 07:38 PM
Okay, lesson learned: mud daubers' nests are really tough! I had to get a pair of pliers and bust the main body of the nest apart by crushing to avoid putting any undue strain on the wires. They're also kinda gross - it would've reminded me of Alien even without a couple old, dead larvae that came out with it :/

Anyway, I broke out the multimeter on it, and the voltages are somewhat below where they ought to be - the +/- 5V lines are slightly below 4V, and the +12V line is closer to +10V. Is this seriously out of tolerance for electronics of this vintage? Where should I start looking to troubleshoot this?

I also checked the clock signal as suggested, but it showed around -2.4V, not +2.4V. Is this correct?

KC9UDX
May 18th, 2017, 08:04 PM
Start by getting a different multimeter and verifying your measurements. Next, measure the AC voltage at the transformer secondary and tell us what you get.

It sounds like your clock is running but you had the meter leads reversed.

commodorejohn
May 18th, 2017, 08:15 PM
Black lead to the ground pin, red lead to the clock pin, right?

But yeah, I'll check with a different multimeter. I have a few to choose from.

Edit: okay, tried a different multimeter and this one shows the readings as more like +/- 4.8V and +11V.

Which pin on the transformer is the secondary? I admit, my knowledge of low-level electrical stuff is embarassingly limited...

KC9UDX
May 18th, 2017, 09:31 PM
4.8V is marginal (in my opinion) but generally OK for 5V TTL. I like to see it higher on older things like this, but I wouldn't worry about it at this point. 11V is good enough.

You can't really go wrong measuring any combination of pins on the transformer with an autoranging meter. The secondary will be the one that isn't 120 or 240V (nominal). I don't know which to tell you otherwise because I'm nowhere near mine, but I'm sure others here will be able to tell you. Either way, you'll find one pair of wires measures line voltage and the other pair is the secondary. At this point, I wouldn't bother, because it's going to be correct given how close your DC voltages are.

MikeS
May 19th, 2017, 07:17 AM
Check the voltage across the terminals of the big off-board capacitor; it should be 8V DC or more. The voltages you're measuring are all regulated; odds are better that your meter reads a little low than that all three regulators are off. Check voltages across C14, C16, C24 and C32; they should be very close to what you measured at the RAM chip.

I'm afraid that it looks like your voltages are OK (they usually are) and the problem is elsewhere; more investigation required. Good opportunity for some practice using that scope ;-)

KC9UDX
May 19th, 2017, 07:43 AM
Next up, check voltage at 6502 signals RESET, IRQ, NMI, and all the A lines, and report back.

When you check /RESET, it should be 0V when you first power on, and go to 5V after a few seconds. If not, the reset circuit is suspect.

If /NMI is 5V that's good, else you may have a bad 6502 or a short somewhere.

If /IRQ is 0V that's bad, pull out your I/O chips (6x2x) and check again. If still 0V, same prognosis as /NMI not 5V.

What your report at the address lines will give us an idea of what's next.