Another PET in the repair queue.. 2001-8 with 28-pin rom sockets

[bitfixer]

Hi all,
Doing some repairs on yet another PET, this time it's a 2001-8, the earlier board with 28-pin ROM sockets.
Before doing some more in depth investigation, here's a quick report of the symptoms.
The CRT has some issues, looks like the raster is messed up in some way. But I unplugged the CRT completely for now, will look at that issue after the board issues are resolved.
For display I hooked up a homebrew adapter to the user port which then uses an RGBtoHDMI for video output. And that seems to work.
On power on, first there is the usual garbage screen, then the screen clears. After about a second an "ILLEGAL QUANTITY ERROR" appears on the screen, but seems to be written out slowly. It takes about half a second for the message to write out, letter by letter.
Then "READY." beneath it, blinking cursor. There is a single garbage character beneath READY, and the keyboard seems non-responsive. Tried with both the original chiclet keyboard, and a keyboard from a 2001-N, same effect.

Any obvious ideas before starting to do some more in depth investigation? Thanks.

 

[Hutch]

Which board revision? 6540 ROMs or 2316B ROMs? 6550 RAM or 2114 RAM? Not that it makes much of a difference really.

One thing to consider with those symptoms is that there is a 60Hz interrupt generated from the video vertical refresh signal.
This interrupt is responsible for keyboard IO since the keyboard is only polled during this interrupt.
On the older boards like this, the kernal also waits for the vertical refresh before printing to the screen in order to avoid screen corruption that happens when the CPU writes to video RAM.

The 6520 PIA#1 is responsible for this interrupt and the same chip is responsible for reading the keyboard and the cassette.
If this interrupt was not happening or was happening intermittently instead of 60 times/sec, that might slow down drawing text on screen.
I'm assuming you have a vertical sync signal since you're getting video but check that it's reaching pin 18 (CB1) on PIA#1 at G8
You might try swapping the 6520 chips to see if that changes things. The other 6520 is used only for the IEEE port.

The screen corruption I mentioned is what looks like static or snow. Random pixels get turned on for one raster as the CPU writes to vRAM and asserts the data bus while the video circuits are busy drawing some other location. The "killer poke" was used to speed up text output by forcing the system to not wait for vsync to output to the screen.