I did check voltages with multimeter AND scope, hence above I said 'with no visible ripple'. Just checked the 5V line again with two DMMs, one showing 5.26V, the other one 5.34V. They are using a L296 switching regulator for the 5V. The switching frequency is 85-110kHz according to the datasheet. I also just checked again for high-frequency ripple on the 5V line. When I check on the plug from the power supply I get <100mVpp of 'noise' (refer to attached pics). When I probe the Vcc of the RAMs directly, I get about 150-200mVpp, mostly noise and 20MHz clock, no spikes. I don't think that's too much of an issue with the ripple to be honest.
@RoGeorge
Now THAT's a great idea with the needles, thanks! Why I didn't think of that!? So I went to 4 different pharmacies/medical supply stores this morning, only 1 of them had them in stock and selling small quantities. The only sizes they had were 0.45mm and 0.9mm, so I took the 0.9mm ones, hoping they will fit through the holes on the board. Pretty cheap, just 0.10 EUR each. Took a file and made one of them flat already. Yes, I did put in that nice green IC socket.
And regarding the outputs, you are right. Those RAMs have a 3-state output - the 3rd state being high Z, so no problem to connect them in parallel to another RAM 'on top' as long as we make sure that all outpus are in fact in the high-Z state. Telling from the truth table of the HM6264, this is possible in 3 different ways (I'm learning a lot here, so bear with me). !CS1=H or CS2=L or !OE=H. So when cutting the !CS1 tracks from the RAMs I should ideally also pull them High (can be done by running a short wire to CS2 in this circuit as it's always HIGH). Still waiting for my RAM, will try as soon as it arrives. Take another look at the screenshot of the 4 traces that I took tonight. 1&2 looks different from 3&4, so we must assume the error happens in seperate stages of 1/2 and 3/4. In the block diagram that would be block 6.15 (Beamswitch & ADC) and 6.11 (Data Control Array, Max-Min, RAM) as they join together in the next stage 6.11 (Acqusition Controller, Beamswitch Control, Ramp Control) -> hence my reasoning: ADC first, RAM next (after I already made sure that the power supply is clean).
@stj
Haven't probed address lines yet, might be of limited value. Unless... there's a short or interruption on one of the address lines. OK, makes sense. Will do tmrw. Was a very busy day...
@Alex Eisenhut
Well, I assume the scope must have worked once, but I got it in this bad condition, pretty messy and full of issues, but still worth repairing. If only for the learning experience (I do own other scopes). You're right about the 5v issue, it should be 5.0v not 5.3v. No UV EPROMS in this scope, but I was already considering corrupt firmware, not sure.
ALRIGHT, things start to make sense to me now! I have probed R/W and CS1 lines of the RAMs. R/W was 500ns low/500ns high. As those 3 R/W pins are connected they read/write from all 3 RAMs at the same time. The 13 address lines are in parallel too, that's where the CS (chip select) pin comes in I suppose. On !CS1 I saw 100ns low (which means THIS Chip is Selected). I also have a 500ns cycle on top of it (see pics).
Attached what it looks like on startup.
Stage 1) Turn on: the screen top is too low with fuzzy readout characters (less than 1 minute)
Stage 2) The srceen suddenly pops into place, still fuzzy readout, menu and traces
Stage 3) The fuzzy stuff slowly goes away, gone after about 2 minutes
Stage 4) Scope becomes usable from about 2-3min, although the traces are still distorted in storage mode only (that's what I'm working on right now)
Good nite!
P.S. Today no missing/wrong characters in the readout.