One more thing to fix... HP9000/380 power supply

[davidkcollins2 at gmail.com]

Carlos, from my experience in fixing a number of switch mode supplies in HP gear, if the supply is dead but you have full voltage across the big filter caps (~340v DC), it's typically a problem with the power supply to the regulator that drives the primary switching transistor.  That 'power supply' can be as simple as a high value resistor (maybe 470K), usually 1 watt or so, that has just gone open circuit with no external signs of failure. 

Assuming you are aware of the risks inherent in opening these supplies up and applying power, you might want to check if you have the high voltage DC and if so, do some basic checks of components on the mains side of the main transformer.   All the usual suggestions around checking for leaky/failed filtering caps on the low voltage DC outputs also apply of course. 

David Collins

-----Original Message-----
From: cctalk <cctalk-bounces at classiccmp.org> On Behalf Of Carlos E Murillo-Sanchez via cctalk
Sent: Friday, 12 February 2021 1:55 PM
To: General Discussion: On-Topic and Off-Topic Posts <cctalk at classiccmp.org>
Subject: One more thing to fix... HP9000/380 power supply


This decade seems to have increased the number of failing things in such a way that the "to be repaired" backlog is growing much faster than I can get to diminish it. Argh.  A month ago my trusty HP9000/380 ran just fine and I booted the different OS's in the SCSI and HPIB drives connected to it (this particular machine is interesting because the
9000/300 port of NetBSD was partly developed in it: it was Mike Wolfson's). Yesterday, it failed to turn on; the power supply is dead. So I unracked the pile of drives and the computer, checked for obvious things (the fuse is fine, and nothing in the power supply is swelled up or leaking, or browned by heat; visually, it looks new; the HV caps seem to hold a charge).  I need the schematics for the power supply (at least the output connector; I can work my way back from that)  and also those for the backplane in this hp9000/380.  A preliminary search at bitsavers and elsewhere did not help.  Does anybody have these?

In the meantime, I finally improved the mainboard (had the parts for a long while) from a 380 to a 385 by changing the clock generator, and replacing the 68040RC25 with an RC33.

I ran this machine as a web server continuously for ten years in the 2000's, totally exposed.  Many tried to hack it... and failed. Another personal connection to this architecture is that I used Apollos and
hp9000/300 at UW-Madison back in 1989-91.  Boy, did I crunch numbers...

carlos.