I went to power on my PET in a suitcase the other day and the boot screen appeared as usual, although after a few seconds strange things stared to happen. The cursor became unresponsive, and some of the characters became corrupt. I did a power cycle and the problem became worse. After a few power cycles the screen looked like this:

neoncluster.com

Corrupt screen on startup.

I could make out some parts of the CBM BASIC 4.0 splash screen. I have highlighted these in red.

After fixing the reset issue

Success after finding stuck pins on the input of D3.

My PET in a suitcase has a Nick Welte RAM/ROM board (I also have one in my full PET 2001). At first I thought maybe it was an issue with this, but I swapped the RAM/ROM units between my two  machines with no noticeable difference.

I then replaced the two socketed 2114 video RAMs but that didn't change anything either.

As a simple test I then pulled the video RAM completely, expecting to see the typical checkerboard, but instead I saw a constantly moving busy mess of pixels.

The next thing I looked at was the reset line on the 6502. Well it wasn't getting any reset pulse, and looked to be always stuck low.

I traced the line back through the 74LS04 at D9, then to the LM555. Pins 7 & 6 of the LM555 are tied together and testing with my logic probe they seem to briefly pulse low, then no discernible signal for a very short moment, then they jump high. Pins 4 & 8 both seem to be tied to 5V and read high all the time. The output at pin 3 is constantly high which certainly seemed wrong. I would expect this to be high for about half a second then jump low.

I then tried to jumper the 6502 reset line to 5V to do a manual reset, but I couldn't seem to get the CPU to reset this way. I also tried piggybacking a spare NE555 on top of the LM555, but still no change.

I ended up replacing and socketing a new 555, and also replacing the two capacitors (C5 and C14) in the reset circuit. Unfortunately that only improved things marginally. I was then able to get the board to reset about one in 10 attempts. I eventually traced the issue to the 1M resistor (R16), I could put in a resistor of a different (lesser?) value and get achieve consistent resets, but when I replaced it with another 1M resistor I was back to the same no reset problem. I just decided to use a different rated resistor and not stress about it too much as I obviously had bigger issues to solve...

Although I had solved the reset issue I was still seeing a lot of corruption on the screen. In places I could see fragments of the CBM BASIC 4.0 slash screen, but mostly the screen looked bad. I could type on the keyboard, so I knew that mostly the machine was up and running...Good news I guess.

I asked on the VCF forums about my problem, I was told I should look at the outputs of the D2, D3, and D4 multiplexers (74LS157) that switch between the CPU addresses and the screen scanning addresses. I couldn’t see much wrong with the outputs of these, but 2 pins on the inputs (B2 and B3) of D3 were both stuck high. These pins are usually fed from the 2114 RAM but as I have a RAM/ROM board installed I figured that the signals must now originate from A2 and A3 (pins 11 and 12) of the 6502. I tested these pins on the 6502 and I was seeing pulses. These pins are fed through a 7417 and then onto the 74LS157 multiplexer where I took the readings before. I decided to test the inputs and outputs of the 7417. It was a little tricky to get at these IC's as that are covered when the RAM/ROM board is installed, but I attached some jumper wires to the pins I wanted to probe and took readings from that...Finally here seemed to be my problem. Even though the inputs pins (pins 5 and 9) were pulsing, the outputs (pins 6 and 8) seems stuck high.

I have some spare 74LS07's which are 7417 equivalents, and decided to piggy back one
over the 7417.

Success!

It was then a simple case of removing the 7417, dropping in a nice machined socket, and replacing the dead chip with the newer 74LS07.

Thanks again to all the helpful folks on VCF.