First is an HP9810A. I just powered it
up...Several LEDs turn on
immediately, but the display is blank and the keys are unresponsive.
Does anyone (Tony?) have any troubleshooting hints? Are schematics
for this machine floating around anywhere?
However did you guess that I've worked on this machine :-).
FIrst go to
http://www.hpmuseum.net/ There are some manuals there,
including te HP servive manual, which is, alas, a fairly pointless
boardswapper guide. But if you can stand my handwriting, there's also a
full set of scheamtics.
Now, the machine is really a bit-serial computer. What I mean by that is
that there's a 16 bit bit-serial processor with ROM and RAM hung off it.
The display, keyoard, printer and magnetic card reader are peripherals to
this processor. This means, alas, that since the display is scanned by
the processor under the control of the firmware, just about any fault
results in a blnak display. The keyboard LEDs are driven by an 8 bit
latch on the keyboard cotnroller PCB, and in fact some are driven by the
normal and inverted outputs of the same latch section (e.g. the fixed and
floating LEDs), so if the 5V PSU is working, then at least some of the
LEDs will be on.
To get inside, first remove the top cover. It's held on by 4 screws under
the printer flap. With those out, you can spring the cover out of the
side castings and take it out. Inside the unit is divided into 4 main
secitons. On the left is a metal box contianing PCBs. It's held down by 4
screws, then flip the ejectors (red and purple IIRC) to free its
backplane from the main backplane, and take it out. The side comes off
that box with 4 more screws. Inside are the memory and memory control
PCBs. From the bottom (I thinK) : ROM (firmware), memroy timing, memory
address register, mrmory data register, RAM, optional expansion RAM
Back in the main chassis, the next section is 4 PCBs plugged into the
main backplane. That's the processor. From the left, the data path (ALU +
reguisters), control (micrcoder ROMs, sequencer, Q register), clock (and
I/O sequencer) and I/O. Note that all tghe handles are colour-coded with
the guides that edge of the board slots into, and that the handles also
encode the last 2 digits of the PCB part number in the obvious code.
The next section in the machine is basically peripherals. The I/O
backplane at the back, the (optional) printer and the magnetic card reder
interface.
Finally, on the right are the 3 PSU boards under another cover. This
machine has mmany supply rails -- +5V for the TTL logic, +12V for the
card reader, -12V for the card reader and the ROMs, +24V for the printer,
and +16V, +19V for the RAMs.
Now some intiital debugging. After removing the cover, take off the
keyboard. It's held on by 4 screws on the bottom. This is one place where
I've found it best not to fit/remove them in a diagonal sequence. Take
out the front 2 first, they go into bushes in the front plastic strip.
Then remove the 2 rear screws. Unplug the keyboard logic harness under
the front of the printer. The other cabel, with the in-line connector at
the right hand side, carries mains to/from the on/off switch. Leave that
for the momemnt.
With hte keyoard out of hte way, power up and test the PSU voltages. The
+5V lione is easy to find on the display PCB (now clearly visible), the
others have marked testpoints on the PSU oards, accessile through holes
in the cover. PSU problems are uncommon, but it's worth checking.
Take out the card reader controller by unplugging the edge connector on
its front edg (motor/sensor signals), releasing the PCB from the
backplane, and freeing the locating stud from the PSU metalwork. Uplug
the head connector from the PCB. Take out the printer (4 screws) and the
I/O backplane (just pulls out). Take out the memory box and remove any
optional expansion RAM oard (very top slot, may be empty). In the main
machine, there's a cable harness plugged into the left edge of the main
backplane under the memory box. Unplug it (it carries the parallel
address/data buses to the ROM module slots) and put the memory box back in.
That's now the minimum configuration to get a display (processor, memory,
PSU, dipsly board). Power up. If still no display (and alas that's what I
suspect), you need to really get into the machine.
There are basically 2 ways to debug the machine. The one that's sure to
work (but which takes a lot of time and a logic analyser) is to grab the
M (memroy address), T (memory data) and microcode PC from the test
connectors Then work out what the machine is doing and compare it to what
it should be doing. The simpler one which works quite a bit of the time
is to rememebr that this is a bit-seiral machine and that as bits are
shifted through the various registers, many signals should be changing.
On the display PCB connectore there's a strobe signal. I think I called it
DispStb. Stick a logic proe on that first. If it's toggling, the CPU is
trying to access the display. The fault might be on the display board (on
the other hand, the CPU might be writing blanks to the display). If that
signal is not doing anything, look at the M reigster outputs (memory
address) on the test connector at the top of the memory box backplane. If
the high-order its are changing and the low-order ones aren't, there's a
fault in the M register (the bits are shifted in from the MSB end).
The next test is to be sure the clock is running properly. With just the
clock PCB and PSU boards in the backplane (the other 3 CPU boards and the
memory box out), check the master clock. It should be 8 MHz. Then look at
the muClk (micorcode clock) and bitclock signals. Both should be
toggling. If you have a 'scope,. there should be 16 bitclocks between
each muClk.
It now gets even more difficult. I've found it helpful to look at the I/O
flip-flop (on the clock PCB) to make sure the machine is not stuck in an
I/O operation. And then to look at the micorocde PC lines (on the test
connectoro n the CPU control oard) to see what the CPU is actually doing.
If you ask me nicely, I can send you an annotated source listing of the
98x0 CPU microcode...
The first thing that sticks out about the 11/05 is
that the
keylock switch has a key broken off in it. At least it's in the
"power" position. Does anyone have a spare key for this machine? If
so, can I talk you out if it? It's not one of the ubiquitous XX2247s
of which I have quite a few; it's a standard "slot" key.
If you don't mind making hte machine a little non-original, I found it
quite easy to take the lock apart (take the front panel off, remove the
lights/switches PCB and the microswitch assembly, and it's quite obvious)
and them to pull out the tumblers. Put it back without the tumblers, and
you can turn it with a screwdriver.
-tony