PDP-11/34 CPU PROMS

Noel Chiappa jnc at mercury.lcs.mit.edu
Thu Feb 10 19:17:23 CST 2022


    > From: Warner Losh

    > Do those chips have ROM numbers on them?

I have updated the:

  https://gunkies.org/wiki/KD11-E_CPU
  https://gunkies.org/wiki/KD11-EA_CPU

articles with the DEC part numbers for the i) microcode and ii) instruction
decode PROms.

That's not all the PROMs on the Control card - there are effing bazillions of
the damned things (I suspect they used them to reduce the amount of random
logic, so the CPU'd fit on two boards) - but it's most of them.

I have yet to triple-check them, so there might still be transcription error
or two.


    > From: Rod Smallwood

    > I am sure somebody will come up with the actual images either the
    > original files or derived from what we have.

I wouldn't be too sure of that; silence so far. I have reached out to Mike
Douglas, to ask where the microcode dump on DeRamp came from: perhaps the
originator can help with the missing bits. (Although perhaps I should ask Al
K; BitSavers also has the dump, and it's older, so perhaps that copy came
from the originator.)

    > We have narrowed the problem down.
    > Its the instruction decode ROM's that are the issue.
    > The images of those are whats needed.

All of them? Or is just one failed?

I'm wondering if you've just had a single one lose a bit or two; that's
somewhat common in old PROMs. The chip you reported as failing (E111) almost
certainly couldn't have taken out an instruction decode PROM, it's nowhere
near them.

I ask because we have absolutely nothing on those PROM's contents. With the
microcode PROMs, we at least have the contents in symbolic form (see pg. 15
of MP00082; alas, we don't seem to have the KD11-EA equivalent of Table 7-15
from EK-FP11A-TM-002), but for all the instruction decode PROMs - nada.
Absolutely nothing.

But if they're _mostly_ there, with the partial contents, and a description
of the failure mode (e.g. 'SETC doesn't set the C bit'), we might be able to
work out what bit got dropped.

Failing that, someone's going to have to volunteer to unsolder a set, and
read them out - at least, I assume that's what would have to be done. Perhaps
a logic analyzer could be attached to an instruction decode ROMwhile the CPU
ran diagnostics, and eventually a complete readout of the contents
accumulated.

	Noel


More information about the cctech mailing list