Re: DECstation 2100: power trouble?

["Maciej W. Rozycki" <macro%orcam.me.uk@localhost>]

On Tue, 14 Apr 2026, Mouse wrote:

> I've got a machine.  The case calls it a DECstation 2100 and the bottom
> sticker says "CPU KN01".  It does have (what is marked as) a MIPS CPU
> chip and the motherboard silkscreen has, among other things, the
> |d|i|g|i|t|a|l| logo and the text "KN01 54-19463-01", so I'm reasonably
> sure it is indeed a 2100.

 Yes, KN01 is either a DECstation 2100 or a DECstation 3100.  The 
difference is in the CPU clock rate only, 12MHz vs 16.67MHz.

> I tried turning it on.  The lights stuck at EF.  I pushed the reset
> switch and they changed tp 7F, then back to EF.  I pushed the reset
> switch again and they flickered in a way that to my eye said "flaky
> connection to the LED common, or flaky power" (they flickered
> irregularly, all exactly in sync as near as my eye can tell).

 Lights at EF would be memory.  I'd suggest cleaning and reseating.  Mind 
that the sockets were specified for 25 cycles only and the recommended 
maximum was as little as 5.  As to flickering, LEDs on this machine do 
seem to behave oddly, but then I haven't seen yours.

 These systems are quite sturdy, though as anything made by humans can 
fail from time to time.  Mine's reached:

# uptime
 05:17:35  up 686 days,  2:04,  2 users,  load average: 0.33, 0.07, 0.02
# 

now, after the replacement of a faulty SCSI buffer SRAM chip.

 NB you do need to have all the required peripherals (keyboard, mouse, 
network) connected or the system won't boot to the console monitor prompt.

> I opened it up and there is nothing obviously wrong to my eye.  Nothing
> looks overheated to the point of damage and none of the caps in the
> power supply are visibly bulging, nor is there obvious mechanical
> damage apparent.
> 
> Given the thing's age, though, flaky power seems more likely than
> something like a bad etch run giving an irregular connection to the LED
> common.  Just because caps aren't bulging doesn't mean they're certain
> to be good, after all!

 Some revisions of the H7821 PSU do suffer from leaking capacitors (the 
quaternary ammonium salt system issue), though the PSU has them soldered 
leads-up, so unless yours was stored in the wrong orientation, the seals 
should be fine.  Have a careful look at the PCB and watch out for 
Chemi-Con SXF series parts.  They do need to be replaced right away!

> But the power supply connector has something like eight different
> colours of wires across 14 pins.  Does anyone know the pinout of that
> connector - that is, what voltages I should expect on what pins?  Also,
> is the power supply capable of running with no load?  (I've run into a
> few that aren't, including some that let out the magic smoke upon being
> powered up with no load.)  I've got a fairly decent multimeter (a Fluke
> 87), so I can measure, but I'd like to know what I *should* get so that
> I can tell whether what I *do* get is good.

 The pinout is given in "VAXstation 3100 Model 76 Maintenance Guide"; 
search for vs31mmg2.pdf if you don't have a copy.  Mind the isolated -9V 
circuit.

 The same PSU is used with the SZ12/BA42A expansion box, where a load 
resistor at 2Ohm is wired between +5V and GND, so I'd be wary running the 
PSU with no load.

> If the voltages are compatible and the PS appears to be nonworking, I
> could hook it up to a boring peecee supply just to get the silly thing
> working; while I'd rather use the original supply, for my purposes I'd
> prefer a working FrankenPMIN to a dead original-equipment PMIN.

 I reckon you might get away without the -9V line.  I'm not sure if it was 
actually used for anything with the KN01.

 HTH,

  Maciej