Differences between PDP-11/83-84 M8190-AB/AC and AD/AE CPU cards

[Jerome H. Fine]

 >Noel Chiappa wrote:

>So I've seen some references which say that the M8190-AB and -AC (w-o/w FPJ11
>FPP chip) are 15 MHz "11/73" CPUs for QBUS use, and the M8190-AD and -AE
>(w-o/w FPJ11) are 18 MHz "11/83-84" CPUs for UNIBUS use (via the KTJ11-B
>convertor, of course). (Of course, the /83 is nominally QBUS, but let's ignore
>that for the moment... :-)
>
>(And I'm not even going to try to give the KDJ11- suffixes that go with
>these... :-)
>
>However, I'm wondering if this is correct... I have a CPU board out of a 11/84
>(one owned by DEC, no less!) which says M8190-AB on the handles, but the clock
>crystal says "18.432", and it has a 57-19400-09 J11 chip, which seems to be
>the 18MHz version (the 15Mhz seems to be -04).
>
>Comparing it to a 'stock' -AB, this one seems to have the same 74S472 PROMs
>(6 along the lower left-hand edge), PALs (top center and lower left), etc.
>But they also match the ones on a 'stock' -AE, too!
>
>So clearly an -AB/AC _can_ be _upgraded_ to an -AD/-AE... but are there
>actually any differences between an -AB/AC and an -AD/-AE, other than the
>speed?
>
>I don't have my /84 up yet, so I can't plug a stock -AB in and see if it
>works, but I'm starting to get the impression that except for the speed of the
>J11 (and the crystal), and whether or not it came with the FPJ11, all four of
>these boards are otherwise identical. Whether it's a 73, 83 or 84 depends
>entirely on whether it has PMI memory and the KTJ11-B UNIBUS converter (and
>the correct backplane, for the latter, of course).
>
>Anyone have an confirmation one way or the other? Thanks!
>
Short summary:  The M8190-AE is used in BOTH the PDP-11/83 and the 
PDP-11/84.
The crystal used is 18 MHz.  The latest revisions of both the CPU and 
the FPU must
be used to ensure that there are no floating point instruction errors.  
The FPU chip is
standard with the M8190-AE.

Longer explanation:  The PMI memory used in the PDP-11/84 is 
sufficiently different
from the PMI memory used in the PDP-11/83 that the magic smoke might be 
released
if the incorrect PMI memory is used.  The MSV11-JB (1 MB) and MSV11-JC 
(2 MB)
are used in the PDP-11/84.  The MSV11-JD (1 MB) and the MSV11-JE (2 MB) are
used in the PDP-11/83.  In addition, for at least the PDP-11/83, the PMI 
memory must
be placed into the backplane ahead of the CPU for the memory to function 
as PMI
memory.  That may be the only position allowed for the memory for the 
PDP-11/84.
I seem to remember that when two memory boards are used, they should be 
the same.
And for the Qbus, if the PMI memory is placed into the backplane 
following the CPU
(the normal position for the PDP-11/73), then the memory functions as 
standard non-PMI
memory.  Finally, most M8190-AA, M8190-AB and M8190-AC boards (PDP-11/73)
with the 15 MHz crystal will support using the MSV11-JD and MSV11-JE 
boards both
above and below the CPU.  If used above, they will function as PMI 
memory (as noted
by the RT-11 show command) and the system will also gain the extra 
speed.  If used below
the CPU, they memory functions as normal Qbus memory.

Speed Difference Between M8190-AC and M8190-AE:  When the PMI memory is 
below
the CPU on the M8190-AC and there is a 15 MHz crystal, the raw speed of 
the M8190-AE
with PMI memory above the CPU is about 33% faster.  If the PMI memory is 
placed above
both the M8190-AC and the M8190-AE, the raw speed of the M8190-AE is 
about 13%
faster.  By simple arithmetic, I conclude that placing the PMI memory 
above the CPU on the
M8190-AC adds about 20% to the raw speed.  The speed tests were done 
about 15 years
ago and with only done to obtain a ball park figure.  I compared the 
time is took for BINCOM
to compare a 7000 block VM: disk against itself, so no disk access was 
involved except to
bring BINCOM into memory and execute.


If I left anything out, please remind me.

Jerome Fine