More mystery recycler boards - DEC, Fujitsu(??), Cipher, Emulex

[Pete Turnbull]

On 07/09/2016 13:02, Noel Chiappa wrote:
>> From: Pete Turnbull
>
>> all microPDP-11/73 machines had an M8190. The M8192 was mostly sold
>> as an OEM board.
>
> That's so bizarre (although the "Supermicrosystems Handbook", which
> covers the 11/73, confirms it used the KDJ11-B).

As does the MicroPDP-11 Systems Maintenance Manual, the Micro-PDP-11 
Handbook, some Micronotes, the MICRO/PDP-11 Technical Manual, the IPB, ...

> So the KDJ11-A
> (M8192) was not used in any 'PDP-11/xx'?

Unless DEC sold something like a PDP-11/73S in a BA11-S chassis (like a
PDP-11/23plus), it was only used for OEM systems and customer-fitted
upgrades.

> The other thing that makes no sense is that the KDJ11-B (M8190) has
> all that extra circuitry on it to support PMI, etc - all of which is
> unused in the 11/73 application! Why not just plug in a (presumably
> cheaper) M8192? In the /73 application, the two are basically
> equivalent. (OK, there are two built in serial lines on the M8190 -
> big whoop.) Both have 8KB caches (although the one in the M8190 has
> slightly fancier tagging, IIRC), etc, etc. Maybe it's the ROM (which
> the M8190 has, but not the M8192)?

Don't forget the LTC :-)  All in all it saves a decent amount of 
backplane space, makes field service easier, and follows DEC's attempts 
to integrate as much as possible.  Otherwise, you'd need an additional 
bootstrap card such as an MRV11-D with -B2 boot ROMs, a DLVE1 (DLV11-JE) 
for the SLUs, something with an LTC, and termination.  That's at least 
twice as many slots in a Q-Q backplane and four slots in Q-CD.  A BDV11 
wouldn't work as it doesn't have the ROM capability (well, one of mine 
does but it's been seriously modified, well beyond the ECO for 22-bit 
:-)).  OK, you could use an MXV11-B with -B2 boot ROMs, but that's an 
expensive way unless you just want a very small (and slower) system, and 
you might still need termination.

And don't forget that DEC sold the microPDP-11/73 as a lower-cost 
alternative to the microPDP-11/83 which didn't appear until slightly 
later, or looking at it another way, as a fancier and faster 
microPDP-11/23.  The very first boards had some ASIC/J-11 problems that 
meant they wouldn't work with an FPJ11, and the first J-11 CPUs wouldn't 
run nearly as fast as intended so they were fitted with 15MHz crystals 
(as-sold-by-DEC 11/83 systems have 18MHz crystals).  I couldn't possibly 
imply they were finding a way to sell the inferior parts.  After that 
it's probably all marketing differentiation.

-- 
Pete
Pete Turnbull