What I've done so far:
- using two different bench power supplies to power it (I don't have the original power supply),
- replaced all electrolytic capacitors (all removed were ok),
- tested and socketed bus buffers U22, U23, U26, U27 (as I pulled one or two pads doing this, I've tested continuity of all connections around),
- swapped socketed chips (SHIFTER, MMU, GLUE) with a known working ones - no change in mine, "donor" still running fine on mine chips,
- tried freezing main chips when it heated and started to restart - no results.
Most test I was running on a single 5V supply, but 12V is only used for powering the floppy (not all require 12V, newer SONY drives don't need it), and is present on a monitor out. Before I was sucessfuly running 520ST on single 5V (but there was required 12V/-12V from PSU to power UART and MIDI ports, but everything except these ports was working on a single 5V).
I've also noticed that when it reboots the upper row of RAM is somewhat warmer than lower. Also the CAS signal on upper row of chips is nice packs of pulses, while on lower row it looks random, similar with RAS - on upper row there is a constant signal, on lower row, some random pulses are missing. Once something happened while I was probing the lines and pulses at bottom row CAS lines disappeared (CAS were high all the time), so did the noise on a picture.The pulses in each case are 2MHz. I don't have the right schematics for it, it is the C070523-001 board.
Does anyone have any suggestions what to check next?
Shouldn't it boot with CAS lines from bottom row disconnected from MMU and pulled high?