HP9836C colour alignment (grey scale tracking)
ard at p850ug1.demon.co.uk
Sat Oct 4 15:10:15 CDT 2008
> > That reminds me, in passing, of the HP98x0. The 'service manuals' are
> > boardswapper guides with PSU schematics only. But HP patented the
> > machines, and the patents (I have a list of the numbers, I think they're
> > also on Eric Smith's site) are several hundred pages each, and include
> > such interesting things as schematics (albeit of pre-production
> > machines), interface schematics, the machine language instruction set,
> > ROM sources (sysrem firmware and many of the expansion ROMs, etc). In
> > other words what _should_ have been in the service manual...
> They actually put ROM sources in the patent? Wow. I wouldn't have expected
Oh yes. With explanations of what's going on too !
> that. What I would like, if not stuff quite so detailed, is something
> along the lines of "this is how it's supposed to act" for a lot of stuff out
> there. Like all those VCRs that I encountered that simply wouldn't turn on,
> where I went looking for power supply and all sorts of other issues initially
> only to find that if the controller chip in there didn't get what it
> considered to be a sensible response from the mechanism when it tried to move
> things it would give up. Causes were as simple as bad belts, etc.
Oh don't get me started on that. If we go back to the machine that
started this thread, the HP9836C. One of the faults in mine was that it
reported 'Floppy Disk Missing' in the power-on test. In other words it
hadn't found the floppy controller card. Now, the normal way that add-ons
are detected in these machines is by seeing if said device assets DTACK/
or whether there's a bus timout, but the DTACK/ for the floppy controller
is producted on one of the other boards (basically all intenral I/O gets
the DTACK/ produced there) So it wasn't that. In fact it didn[t take me
long to rrealise that it detected a floppy controller iff it could do a
read/write to the first byte (I think) of the 256 byte sector buffer on
the card. As soon as I spotted this was implemented with 2114s s, I
knew what to do. I remplaced them, and put the card back in, not
conencted to any drives, and tired again.
A great improvement. It then reported 'Disk drive not responding' or
something like that. Which made sense as there were no drives. So I
cabled up drive 0  and tried agqin. I then got a 'Floppy Drive Failed'
error.. After much testing, putting the drive on an exerciser, looking at
signals round the FDC chip, etc, I realised that being a 9836 (and not a
9826), it was expecting 2 drives. If you gave it 2 drives, it correctly
reported '2 Floppy Drives'. If yoy gave it 1, you didn't get '1 Floppy
Drive' you got 'Floppy Drive Failed'.
And this isn't in the boardswapper guide either!
 In this machine, the drives are jumpered to be permanently enabled,
and have separate cables back to the controller board . This means the
controller board can continually monitor the Wprot/ line to detect when
disks are changed.
 The first version of the controller, which I've never seen 'in the
flesh' had one 34 pin cable soldered to it, and could only be used with
one drive. Thus this controller is used in the 9826 only. The second
version has a 34 pin cable soldred to it for drive 0 (in 9826s and 9836s)
and a 34 pin header plug, into which a second cable is pluged for drive 1
in the 9836.
You would ahve liked the way I had the machine set up for this. I wanted
to be able to work on the PCBs easily, which essentially meant not having
any case parts in the way. So what I had was :
The monitor on the bench, with the computer's motherboard on top of it.
Monitor cable plugged into the back of the motherboard. Video boards (and
brightness control cabled up in the normal way.
CPU board in its normal slot. Floppy controller in its normal slot. 1M
memory board in the slot for the expansion backplane (I didn't need any
more slots, the pinputs are similar enough -- jsut 2 signals, not used on
memory boards different -- for this to work
Keyboard connected by normal ribbon cable and placed between the edge of
the monitor and the top of a bench PSU (see later).
PSU PCB in the normal motherboard slot, but not connected to the HP
transoformer/rectifier bits. Instead I wired a spare edge connector to
take power from that bench PSU to the regulator board input.
> > I must admit I've never had any success with these third-party flybacks.
> > I had an Amstrad VGA monitor with flyback trouble (the voltage divider
> > block for the focus and A1 supply was breaking down),
> Snap, Crackle, Pop! :-)
Actually not. It was just that the picture got ever brighter and more out
of focus as the monitor warmed up. Checking the CRT pin voltages revealed
the obvious (A1 going up, and I think focus coming down, but I wouldn't
bet on that).Further tests showed the problems was alsmost certainly the
flyback and not leakage in the CRT, replacing the flyback proved it.
More information about the cctalk