KH11-A Notes M. Eberhard 6 August 2025 (2:14PM) According to The documentation for the DT03 Unibus Switch option, the KH11-A option is required to use the DT03 with the KA11 processor. The KH11-A is described in the DT03 description as the "Large System Capability Option." (PDP11 Peripherals Handbook, 1976, page 4-178) The list of boards that belong with the KH11-A comes from the Spare Modules Handbook, March, 1976. It lists the following modules for the KH11-A: {M724-YA, M725-YA, M726-YA, M727-YA, M728-YA, M822-YA, M824-YA, M825-YA, M7216-YA, M8251} The M7216-YA replaces the M7216 in a KC-11 processor (in the PDP11/15). The M728 in my system is not labeled as a M728-YA. However, it does have some rework on it that corresponds to the reworked wiring on the PDP11's backplane. Further, the rework mostly matches the M728 in pbirkel's M728 in a KH11-A. (His is also not marked as an M728-YA.) There is supposed to be a trace cut beneath E21, between pins 1 and 13, confirmed by checking pbirkel's board. This cut had not been made on my board, which resulted in a signal conflict. I removed E21 so that I could cut this trace, before replacing E21. The M725-YA appears to be an unmodified M725 (plain). However, one signal input changes name due to a modification on the backplane. All of the other -YA boards (except maybe the M7216-YA) are made from the non-YA equivalent boards, with small modifications (mostly a few jumpers and cuts, with one resistor-capacitor pair added on one board). The M8251 is a new board, not replacing any KA11 board. It belongs in slot 7B. Most of the added wiring on the backplane goes to this board. I have reverse-engineered the M8251 board, as well as most (all?) of the backplane modifications. It is possible that there are mistakes in my work. This directory has reverse-engineered schematics and photos of the KH11-A boards in my system, as well as a photo of the backplane wiring. (The blue wires are all part of the KH11-A option.) ***** After looking at pbirkel's KH11-A boards, I am a little suspicios of the rework on my boards that does not match his. In both cases, the rework wires are different than the other rework wires (yellow rather than greenish). Also these wires are not glued to the board the way the other rework wires are. These suspicious reworks are: * M727-YA: The rework on E32 pin 13 * M822-YA: The added RC on E11 pin 4 ***** From pbirkel on vcfed.org, 30 July 2025 I researched this last year -- it's not clearly documented anywhere; I had to cross-correlate fragmentary references in multiple sources in order to assemble a clear(er) picture of what happened BITD.. Basically the initial CPU design wasn't quite "fully-baked" and had an early board-set upgrade (KA11 vs. KA11-YA; the latter included the KH11-A). In the initial release NPR was supported but slow, and there were only two priority levels. The KH11-A upgrade changed how the basic instruction-execution cycle worked so as to speed up NPR to what we've come to expect as "normal"; doing so necessarily touched a lot of the CPU modules as well as required revisions to the wire-wrap backplane. If one has a bunch of *-YA modules then you have the KA11-YA, else you have the KA11 with the slower NPR and original instruction-execution cycle. The big change between the KA11 and KA11-YA was that the initial implementation didn't honor a NPR request until the end of a complete Read-Modify-Write cycle. The revised implementation honored a NPR request as soon as the Read phase completed, so reduced latency and potentially greater throughput.