HP 9000/216

Tony Duell ard at p850ug1.demon.co.uk
Wed Dec 5 17:52:31 CST 2007

> >> If someone in Europe wants one, I also have one that I do not use. ( 1MB 
> >> extension, 3x RS232, GPIO & HPIB cards , graphics extension . But I 
> >> would like my outlay back...)
> > 
> > I am wondering how on earth oyu get all taht into a 9816. Do you have an 
> > expansion chassis?
> > 
> No, got caught out with HP's naming : mine is actually a HP9826...

Ah, that expalins it. The 9826 and 9836 have 8 DIO slots, 4 of them can 
take PCBs with external connectors.

The 9826 and 9836 are pretty similar machines. The main difference is in 
the video system (9826 has 50*20 text and an internal 6" or so CRT, the 
9836 has 80*24 text and a 12" monitor that fits on top of the machine). 
The 9826 has 1 internal 5.25" drive with the CRT alongside it, the 9836 
has 2 intenral 5.25" drives. About half of the internal PCBs are the same 
in both machines.

I have a 9836 in bits on the bench at the moment, When I've sorted out 
the PSU problems and realigned the drives, I put it back together and put 
the 9826 on the bench.  My 9826 was dropped before I got it, and there's 
major cabinet damage mainly due to the mass of the mains transformer. 
Fortunately all the PCBs, the disk drive and amazingly the CRT are 
undamaged, so it should be repairable.

As well as the 9816, 9826 and 9836 I've already mentioned, I have a 9817. 
This is an HP-shoebox sized machine with 6 DIO slots. Inside the case is 
the DIO backplane (whcih also includes the +/-12V regulators, an SMPSU 
board, and the CPU board. The last contains a 68010 processor, MMU 
circuitry, HP-HIL port for the keyboard, HPIB port, RS232 port, etc. 
There's no on-board video, rather there/s a DIO video board in one of the 
slorts, and a graphics board in the 'internal' DIO slot above it. 

I have the higher-resolution video board in mine, which uses an special 
HP monitor. The output is composite video, but at a higher horixontal 
scan rate than RS-170. 

Yes, I have the monitor. One day, when sorting out an RS232 problem with 
the main machine [1], the monitor went bang and magic smoke came out. Of 
course I turend it off at once and pulled the covers. It turned out that 
one of the mains filter capacitors had failed (case split open, metalised 
paper coming out. Replacing that was easy enough, once I'd got to it. 
Getting the main PCB out of that monitor, actually a Samsung chassis, is 

Even removing the  back cover is harder than it should be. You have to 
unclip the CRT bezel, remove 2 screws through the hole the bezel came out 
of, then put the bezel back on, put the monitor face down and remove 4 
screws on the bottom. The case then lifts off. 4 very obvious screws hold 
a chassis plate undr the main PCB.

There is one pieece of very odd design in that monitor. The monitor 
circuit itself runs off 12V. This is provided by a mains 
transformer/rectifier/smoothing capacitor circuit and a 3 terminal 
regulaotr. There's one main PCB containing the monitor circuiry, the 
recrifier and smoothing cap and the mains filtering. The 3 terminal 
regulator is mounted o n seaprate heatsink, the transformer, mains swtich 
and maisn input IEC plug are separately mounted on the chassis.

Now for the odd bit. The regulator has a 3-pin Molek KK socket-thingy 
fitted onto the pins, so that can be easily disconnected. The mains input 
from the IEC plug goes to a 2 pin socket on the PCB, so that comes off, 
ditto the on/off switch (separarate conenctors fo the live and neutral 
sides...). Now the mains transforme has a single seondary winding, again 
conencted by a 2 pin socket, and 2 primary windings (to be put in 
parallel for 115V or series for 230V mains). There's a voltage selector 
swithc on the chassis 

Now many of the wires from that switch are soldered to the PCB. This is 
nopt a real problem, it's easy to remove the 2 screws holding the switch 
in place and remove it with the PCB. 3 of the 4 mains transformer wires 
go to a socket that plugs into the PCB. But the fourth is soldered to the 
voltage selector. In other words, you either have to desolder that wire, 
or remove the PCB and transformer together. Why they didn't use a 4 pin 
connector for the transofrmer primary wires and route this last wire via 
the PCB is beyond me. I almost suspect the votlage selector design was an 

[1] Back in the main machine, on the CPU board. The RS232 port was 
failing in odd ways, for example it would fail diagnostics at power-on, 
but would pass every subsequect diagnost. It turned out that all 4 of the 
RS232 buffer chips (1488s and 1489s) _and the UART_ (an 8250, suprisingly 
in a 68000 machine) were defective.


More information about the cctalk mailing list