2 Oct
2026
2 Oct
'26
6:59 p.m.
Yes, you are right - I should have said 11/15 not 11/05. If that's the
worst mistake I make today, I'm doing alright :-)
I thought the explanation at the beginning of the document was excellent,
explaining what the KH11-A does and why. Fundamentally, they were solving
the problem of a system with lots of DMA devices running out of Unibus
bandwidth to complete all the DMA requests. Roughly, this mod allows more
opportunities during the operation of the KA11 CPU for DMA requests to get
acknowledged and get a turn on the Unibus. (For a better description, read
the first 20 pages of the documentation!)
Martin
On Fri, Oct 2, 2026 at 1:56 PM Paul Koning <paulkoning(a)comcast.net> wrote:
Wow. I've never heard of this. The O/M list shows it as a regular
product, not a CSS special which was my first guess. I wonder how many
they actually sold. The 11/20 I used in college was a rather large system
(it had 16 serial lines, each requiring a KL11 or DL11 single line serial
interface module) plus RK11, RF11, and TC11. So I'm sure it had one or two
bus extenders, but as far as DMA load is concerned it probably wasn't too
crazy. Though the RK11 is pretty fast for an 11/20 to keep up with
(faster, in transfer rate, than the RF11).
For all I know it may have had one of these, but I never heard that.
paul
On Oct 2, 2026, at 4:38 PM, Martin Eberhard via
cctalk <
cctalk(a)classiccmp.org> wrote:
I just finished scanning the DEC documentation for the KH11-A "NPR
Expansion" option for the PDP11/20 and the PDP11/05 machines. (I just
bought this drawing set on eBay...)
The 1975 option/module list says it's for the KA-11 and KC-11, which are
the 11/20 and 11/15 CPU respectively. Not 11/05, that is an entirely
different design.
As far as I can tell, this is the only
documentation for this CPU
modification that's online - aside of course for my own
reverse-engineering
drawings from earlier this year.
You can find the drawings here.
<
https://deramp.com/downloads/mfe_archive/011-Digital%20Equipment%20Corporat…