It's been a hell of a day! HP 9845/2x i8008/Sage II/2x Grid/TI

[Christian R. Fandt]

Upon the date 19:29 16-05-05, Tony Duell said something like:
> > Tony, the HP250 CPU is said to be the same as the 9845* except for
> > different microcode. Reference this website if you can:
> > http://www.hp-eloquence.com/history/history.html
>
>YEs, but _which_ 9845 processor?

I know little about the 9845 hardware at present. I wish I had my own 9845 
but haven't found an intact one yet.


>An HP9845 has 2 processors (not counting the graphics accelerator that
>may be in the monitor). One handles I/O (the 'Peripheral Processor
>Unit'), the other runs user programs (the 'Language Processor Unit'). In
>most machines, these are those HP custom hybrid modules with a large
>die-cast heatsink on top, a bit like the processor in the 9825, but with
>different pinouts, etc.
>
>Soem machnies (and mine is one of them) have the high speed language
>processor option. This replaces the LPU board with its processor with a
>set of 3 boards linked by a little backplane on top. One of the boards
>plugs into the main backplane slot that takes the normal LPU board. The
>other 2 hang over the side of the cardcage. These boards contain (amongst
>otehr things):
>
>Interface PCB : Bus buffers, abritration logic, some processor registers
>
>Data path : ALU (4 off 2901), condition logic, microcode branch PLA, ALU
>decode PLA
>
>Control : Micorocode PROMs, sequnecer (2910), BCD adder and shifter, more
>registers

Based on that description I would bet that whomever stated the 250 CPU was 
the same, except for ucode, had intended the comparison to be to the LPU 
which is implied above to be equivalent to the "main" processor (vs the I/O 
processor which has less responsibility for calculations, logic processing, 
program control, etc.).


>I am pretty certain than the 2 processors are object-code compatible. But
>I am alos sure that schematics are totally different. I will take a look
>at the ose HP250 diagrams if I get a chance (no, I am not going to spend
>5 hours downloading them here!) to see what that did. But i suspect the
>hardware will again be different.

I always wanted to compare the two machines (9845 and HP250) once I 
discovered some years ago that the 250 CPU was (alleged to be) the same as 
the 9845 CPU. You will have a pinout of the 250 CPU in those drawings to 
determine if one of the '45 CPUs essentially match.

I think if you find nearly all pins match to the 250 processor on one of 
the '45 processors, based on functions extrapolated from interconnecting 
circuit board traces between the '45 PPU, LPU, and adjacent circuitry, then 
you have your culprit. If not many match, although the 250 CPU was *said* 
to be the same except for ucode (implying sameness hardware-wise), HP had 
therefore evidently wire bonded the three differently to accommodate 
circuit pinout and/or function requirements.

Device seems big enough to have separate die holding ucode bonded to the 
substrate next to the CPU die itself under that cover. Substrate could have 
different artwork to accommodate different pin location requirements for 
different uses (PPU, LPU, 250). If only contact could be made with an HP 
chip fab engineer or tech on these parts to understand this for sure.


> >
> > I feel this may give you something to work with as you reverse engineer 
> the

I stand by my statement :-)  Just a little enlightenment sooner is better 
than having to travel through darkness longer :-)


>FWIW, a reverse-engineered 9825/9831 schematic is on the HPCC schematics
>CD-ROM (along with similar diagrams for the 9100B, 9810, 9830, 9815, some
>late rmachines, and most of the handhelds)

Hmm, great (re: 9825 schema)! I will track down that CDROM again and buy 
it. I have a link stored somewhere.

Regards,  Chris F.

NNNN

Christian Fandt,    Electronic/Electrical Historian
Jamestown, NY  USA      cfandt at netsync.net
         Member of Antique Wireless Association
         URL: http://www.antiquewireless.org/