PDP-11/45: RK11

Finished up the cleanup and rebuild of the H720E for the RK11-C controller: reformed the big electrolytics over a couple of days, and swapped out the tantalum filters on the regulator board (one of which had started to leak) for some replacements. Tantalums of the correct form and value are crazy expensive, and since these caps (C11-C17, C23) are banked resevoirs on the regulated rails their exact values are not critical. Went with some aluminum electrolytics instead that worked out to a little more aggregate capacity on each bank.

My H720E is missing its top cover plate, so there was about 40 years worth of accumulated dust in there -- a lot to clean up! I'm not sure if the cover was customarily left off to facilitate cooling of the regulator board, or if it is just genuinely missing. Will see if the forums/lists have any wisdom here. Should not be hard to fab a replacement if it is really supposed to be there.

Routed the inter-cabinet power control wiring, racked the RK11, and cabled everything up. This included moving the M9301 bootstrap terminator over to slot 0, then rummaging through a box of BC11 bus cables for one that was both long enough and in decent enough shape to connect the CPU and the RK11. Some of these bus cables had period-correct labeling (see picture below :-)) Got a good tip off the cctalk mailing list (thanks, Jerry!) to drape the BC11 with a service loop from the top-rear of the CPU rack, thus avoiding any entanglement with the rack slides while moving the CPU cabinet in and out.

A 2000pF cap that was flown over the RK11 backplane (+5V to DR BUS DC LO L) had broken free of its pin connectors; jury-rigged this with some arduino jumpers I had lying around, until I can track down some appropriate replacement connectors.

Good news is that after this the machine still booted the M9301 bootstrap, and was still able to run and pass diagnostics loaded via PDP11GUI. So, Unibus still working with the BC11 cable in place and termination out at the RK11.

Started in with diagnostic ZRKJE0.BIC, which is the controller-only static test. This indicated:

REGISTER NOT CLEARED
  PC   REGADD    RECVD
002560  177416  040000

This is the RKDB register, implemented by two M203 R/S flip-flop modules in slots A21 and B21 (see sheet RK11-C-10 in the RK11-C engineering drawings). Swapped these, and the stuck bit moved to the other half of of the register, so looks like failed gate. Pulled, socketed, replaced, and diagnostic no longer reports any stuck bits. Next diagnostic fail is:

UNEXPECTED TIME OUT AT PC=004300

This is test #21 of the diagnostic, testing controller reset. Will need to do some reading up on the design of the controller and the diagnostic source to understand how to troubleshoot it further. All for now!