HP9836C colour alignment (grey scale tracking)

Roy J. Tellason rtellason at verizon.net
Wed Oct 1 16:55:21 CDT 2008

On Wednesday 01 October 2008 16:59, Tony Duell wrote:
> > > Very few HP classic desktop computer service manuals contain
> > > scheamtics. You might, if you are lucky, get PSU schemaitcs, but not
> > > for the processor, etc, sections. I am, of course, trying to remedy
> > > this...
> >
> > That stinks.
> Waht does? The fact that HP desktop computer service manuals are anything
> but, or the fact that I am trying to do soemthing about it?

The former of course,  and I don't know why you'd ask me such a silly 

> > > Anyway, the HP9836 'service manual' is a boardswapper guide, and
> > > contains very little information that's not obvious from 10 minutes of
> > > looking at the machine. There's a large section on running the
> > > diagnostics, with what to do for each error, and alas the latter is
> > > 'repalce the <foo> board'. And very often those error messages give a
> > > lot more information once you've learnt to interpret them
> >
> > I've seen some diagnostics that were like that,  and some that were much
> > more
> THe HP ones are pretty specific. For example, this machine had a
> graphics board problem when I got it, the error would be reported as
> 'graphics failed Meory failed at <hex number> W : <hex number> R : <hex
> number>' 

Not bad.

> It was a good bet those 3 hex numbers were an address, the value written
> and the value read. THe only problem was that the W anmd R numbers were
> 32 bits each (8 hex digits), the 68010 processor used in this machine [1]
> has a 16 bit data bus. Worse than that, the graphics memory uses the low
> nybble of each byte only -- this is to allow writes to individual pixels,
> there being 4 bits per pixel. So even knowing which bits had failed it
> wasn't immediately obvious which memory chip to change.

At least it was possible to figure it out.

> In fact, after much testing, I discovered that 4 of the hex digits are
> always 0 (these correspond to the nybbles that don't actually exist), the
> other 4 nybbles are either all parts of high bytes or parts of low bytes,
> and these 4 nybbles correspod to different RAM chips on the PCB (there
> are 16 4416 RAMs on the graphics board...) having figured that out, I
> desoldered the chip I thought it was complaining about, replaced it (with
> one raided from an old EGA grapghics card) and was pleased to see that
> the graphics now passed diagnoistics

It's a nice feeling,  isn't it?  I miss that lately.

> And no, none of that is in the HP service manual either. If they'd
> explained how to decode the error message it would have saved me an hour
> or so's work
> [1] Yes, I know everyone claims it's a 68000 processor. This machine has
> a 68010 and an MMU circuit. It's the latest (AFAIK) version of the CPU
> board.
> > specific,  telling you what chip you might want to look at changing out. 
> > But not since the early days of the Zenith Data systems stuff.  The
> > trouble with that sort of diagnostics is that most of the machine needs
> > to be working in the first place,  or they don't load and run.  Printed
> > versions of the same
> Yes, I don't really trust diagnostics either. After all, the device
> producing the report (the computer you're working on) is probably not 
> working correctly (otherwise you'd not be repairing it) so how can you
> trust the diagnostic program output. 

I guess it all depends on the nature of the fault.

> I prefer to use a 'scope or logic analyser whenever possible (and when
> finding the memory problem I've just desicribed, I did look at the RAS/ CAS/
> etc signals first...) 

The scope  was my most-used tool for that sort of thing,  and I never had the 
use of a logic analyzer,  mostly working for myself I could never justify the 
expense.  Lots of positive aspects to being self-employed,  but there were 
occasional negative aspects too.

> [...]
> > I've often wondered why there wasn't some standard way of numbering
> > those. This sounds like the serial connectors that were out there (and I
> > have a pile of them) for different I/O adapters,  if you didn't happen on
> > the right ones the serial ports just wouldn't work,  and it wasn't
> > obvious why.
> The craziest one I've ever seen is on the PERQ. It uses a DB25 connector
> for a GPIB port (I am told this is an IEC625 interface or soemthing). The
> problem is that it simply maps the pin numbers pin-pin so pins 1-24 are
> used and pin 25 in a N/C. This puts one of the date lines on the top row
> of the connector after the shield pin (pin 12 is shield, pin 13 is a data
> line), it also means you can't simply crib an DB25 plug and a 24 pin
> microribbon socket onto a piece of 24 wire cabel to make an adapter. ARGH!

Even with the standard pin and serial connectors there are apparently enough 
different ways to put them together to cause problems.

> > > The dismantling procedures in that manual are a little odd too. Anyone
> > > who can follow the procedure for removing and replacing the internal
> > > CRT of the 9826 and not end up dropping it or using choice language is
> > > a lot cleverer than I am.
> >
> > I've thought for a long time that "using choice language" was a part of
> > being a technician.  :-)
> YEs, but I tend to reserve it for when it's really needed. Like when a
> 30kV supply flasehs over to my hand, or I drop the power transformer on
> my foot.

Yes,  that sounds like situations that would call for it all right.  But that 
may be a bit more reserved than I tend to be.  :-)

> > > There's one curious thing about that servic manual. It contains no
> > > schematics at all. Not even of the mains side of the PSU, which is
> > > trivially field-repairable.
> >
> > Then they really ought not to call it a service manual,  per se.  Or at
> > least not in my view of things.
> Exactly...
> For information, the PSU is bascially as follows. Mains comes in on a
> filtered IEC plug, goes through the mains fuse and then to the power
> switch under the keyboard. From there it goes to a little PCB screwed ot
> the (inide of the) back panel of the machine which contains a pair of
> ovltage sleector swithces. THese link to the primaries of the mains
> transformer (2 windings each 0-100-120V, the switches let you configure
> the machine for any of the normal mains voltages). There is a single
> centre-tapped secodnary on the transformer, this goes back ot the PCB
> which contains a pair of small diodes used to provide a power-fail
> indication to the optional battery backup unit and a large double-diode
> unit (TO3 can) for the main supply.

I don't think I've seen many diodes in TO-3 packages.  Plastic ones,  yeah.

> The output of this is smoothed by what a frield of mine would call a CFE
> (Capacitor F-ing Enormous) to provide about 30V DC. This is fused (15A) and
> fed to an edge connector socket screwed to the bottom of the case. One set
> of fingers on the PSU regualtor PCB plug in there -- this PCB contains
> step-down switching regulators providing +5V, +12V and -12V. The outputs of
> those go to more edge figners which plug into a connector on the
> motherboard. 

I notice that edge connectors are a lot less common than they used to be.  I 
wonder if there are reliability issues by comparison with other typies?

> The battery backup option was an 18V (?) NiCd (?) which fitted into a
> cavity on the bottom of the machine and which provided an input to the
> regulator board.

That'd be what,  15 cells?  Yikes!

> There was an extra control PCB that munted under the (RH) disk drive, I know
> little about this apart from the fact that it appears to have contained an
> 8041 microcontroller. The battery unit is one of the rew options I don;t
> have for this series of machines. 
> > > > Yup,  if you crank it way up.
> > >
> > > I _think_ the only misadjustment that will cause this is the A1
> > > ('screen grid') adjustment on the flyback transformer. Maybe setting
> > > the clamps way too low will do it.
> >
> > Screen controls are singular these days,  but the early stuff used to
> > have one
> Yes. In fact this CRT has separate cathode pins for each colour, but for
> all the other electrodes, corresponding ones are linked inside the CRT
> and brought out to one pin.

I'm pretty sure that was one of the changes when they moved away from the 
original delta arrangement.

> > for each color.  You'd flip a "service switch" which would give you just
> > horizontal lines on the screen,  and either adjust for all white or if
> > they
> Yes, ben there, done that.
> On this monitor, the 'clamp' presets on the video PCB prerform
> essentially the same function


> > weren't exactly overlapping for equal apparent brightness.  Or in some
> > stuff till they just went out.
> >
> > (Snip...)
> >
> > > > I wonder if it would be possible to get from some aftermarket
> > > > suppliers? I
> > >
> > > Als I doubt it. IT was not a common machine, so I doubt anyone bothered
> > > to make a nrw flyback for it.
> >
> > How certain can you be that they went and made one that was unique to
> > that machine?
> I cna't be. All it has on it is an HP part number (which is not a lot of
> help). I have no idea who actually made it, whether it was used in any
> other HP product, or whether it was actually a standard one that was
> re-labelled (although I think that is unlikely).

That's what I was wondering,  if there were any possible way to research that 
sort of thing.

> > Different strokes,  etc.  I suspect that with a little digging it might
> > be possible to find out who actually made that.  They likely didn't make
> > it themselves.  And an off-the-shelf product is much cheaper than one
> > that's custom-made for one specific product.
> THis is HP we're talking about... 

Which is not the same as saying it were say,  IBM.   :-)

> I would not be at all suprised if it was a custom part. And even if you
> could find who'd actaully made it,  they may well be under contract to to
> release any details. Of course now it has no real commerical importance, but
> trying to find somebody at HP who (a) realises this and (b) is prepared
> publically to say so is going to ba lot more work that designing and making
> a flyback from scratch. 

Maybe so.  And I could be totally wrong about this,  but I just feel like 
there should be some way to characterize those,  somehow.  Maybe not,  but 
maybe I'll play around with the idea at some point.

> > (Snip...)
> [Description of monitor horizontal deflction system snipped]
> > > Of course this information is not in the HP manual...
> >
> > Of course...
> >
> > Did you figure most of it out or come across it somewhere else?
> Now, given that I've had this machine in many bits over the last few
> months what do you think?

If I had that much of an idea I wouldn't be asking now,  would I?

Member of the toughest, meanest, deadliest, most unrelenting -- and
ablest -- form of life in this section of space,  a critter that can
be killed but can't be tamed.  --Robert A. Heinlein, "The Puppet Masters"
Information is more dangerous than cannon to a society ruled by lies. --James 
M Dakin

More information about the cctech mailing list