The same kind of project has come up to my mind.
Given how hard it is to find the original ROM modules, I am not suprised :-)
I was planning to build a ROM-board for the 9830 so
you can put all the roms
to 1 board.
I thought about doing soemthing like that, but there are problems.
The first is that so far I have found :
8 ROM modules for the HP9810
4 for the HP9820
11 for the HP9830.
I know of at least one more for the 9820 (PC2) and 7 more for the 9830
(DataComm 2, DataComm 3, and the 5 FastBASIC modules). Now, the 9810 nad
9820 have 3 slots for ROM modules, the 9830 has 8. The number of slots is
effectlive 'hardwired' int othe memory map of the machine. So there's no
way you could ahev lal the ROMs availalbe at onxe. You would need to hev
some way of selecting which ROM you wanted in each slot.
Then there is the problem that the singals on the ROM module connector
ocnsit of : Power lines, the 9 low-order address liens (A0 to A8, and
also A8/), the 16 data lines, a precharge line (CE) and 2 selects (for
the 2 sets of ROMs in a real HP module). You don't have access to the
highter order address liens or more module sleects.
So a multiple ROM odule would involve modifcations to the machine. For
the 9810 and 9820, it could conenct to tyhe card edge on the main
backplane in palce of the cable to the ROM backplane. For the 9830, I
think it would be sinplest ot replace the EXT ROM Selector board in the
machine (this has the higher address lines on its conencotr) and also one
of the itnernal expansion ROM boards. The 2 replacement board would have
ot be linked by an over-the-top cable.
In both cases there are rater more modifications to the machien that I
would like. You also can't use real HP moduls at the same time. And the
ROM sleection logic in a little complex. Not ridiculously so, but I try
to use the sort of deices that were avaialbel when the machien was in
use, so I would not want to use programmable logic here. It's out of
character.
So what I did was make a board that acts as one module. It plugs into the
module conector with no changes to the machine. You cna put HP modules in
thew other connectors if you want to.
The board cotnaisn a pair of EPROMs (2764/27128/27256). It's a lot
simplere to use 2 EPROms to get the 16 bit word than to have logic to
extract the word from 2 locations and latch it. My programmer doesn't
ahndle the more modern 16 bit wide EPROMs, so those were 'out'. THere's a
few passives, a couple of transsitors to bring the 12V sleect lines back
down to 5V, and a simple TTL chip to cenerate the ROM CS/ signal.
There's one board for the 9810/20 (same modules, electrically), another
for the 9830. They are both hand-wired at the moment, I intend to design
a PCB version sometime.
Each 1K of EPROM will hold a moduel image. So if you use 27128s, you can
store either all the 9810 _and_ 9820 iamges, or all the 9830 images.
There's a DIP switch on the board to select which image you want. Maybe
not elegant, but simple.
Some gotcha's... Decoupling is very important, particularly on the
9810/9820. You need 1k pullups on the address inputs and, for some
reason, on the output of the TTL logic feedign thr ROM CS/ signal. You
need to use one of the level-shifted moduel chip select signals to
provide the A9 signal to the EPROM so that the 2 halves of the module map
to different parts of the EPROM. The address and dat aliens at hte moduel
conencotr are active low (FWIW, the M and T registers in the 9800
machines are active high, but there are invertes on the address and data
lins to the module, look at the schematics). This is no proble, oyu jsut
flip bits around when you make the image to program into the EPROM.
I didn't came to at the moment, too much work and
too much to do..
But I'm very interested to your implementation of the home made ROM boards.
I don't suppose you're coming to the HPCC conference next month [1]. I
am presentign the module there. If not, I will get scheamtics to you
somehow. If there's ever a PCB version, I'll let you knwo about that too.
[1] Anyone here is very welcome to come along, It's 27/28 October 2012 at
Imperail College, London (England) Thsi is not really a classic computer
event, but there's a talk each day on classic HP desktop machines [2],
and many of the members have interests in old computers, calculators and
other related thigns. If you want to come, please let one of the HPCC
committee know.
[2] It is pointles to argue whether or not these machines are classic
comptuers or not. Many of them are comoters intenrally (processor
addressing memory + peripherals),. They are user-programamble. OK, they
are designed for calcualtions rather than, say, text processing, but that
comment could apply to someting like a KIM 1. They were not called
`computers' by the manufacturer, but that also applies to PDPs. If
they're not something that itnerest you, fine.
-tony