Re: DECstation 2100: power trouble?

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

On Wed, 15 Apr 2026, Mouse wrote:

> >> I tried turning it on.  [...lights acting odd...]
> > Lights at EF would be memory.  [...]
> 
> Where are these documented?  I guess I haven't found the right document
> yet.  Is this the PDF you mention below?  I'll have to look for a copy.

 No, that's for the PSU only, and the rest is different.  The code is from 
"DECstation 2100/3100 Operator's Guide".  It's available online.  I guess 
the maintenance guide referred there would be even more interesting, but I 
haven't come across a copy.

> > 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.
> 
> 2 ohms on 5V is 2.5A for 12.5 watts, hm, yeah.

 Yep, pretty chunky with a heatsink integrated and case-mounted.

> But then I noticed a jumper between two of the pins on the miniDIN-7
> marked with a mouse icon.  I removed that (thinking it may be making
> the ROM code think it's got a mouse and therefore a monitor) and, while
> it took some seconds from powerup, I got a KN01 V7.0 banner on the
> printer port serial line (nothing connected but power and serial).  It
> got to test 3 ("7..6..5..4..3..") and then I got a ?? getting erased
> and redrawn on the next line.  I eventually found that ^C stopped it,
> printing "exit(-1)" and giving me another banner and a language prompt.

 Yep, it's what you get with no keyboard or mouse connected, it stops at 
"3".  A VSXXX-AA or -GA mouse will do, as will an LK201, LK401, LK421, or 
a similar keyboard.  Unsure of a loopback.

 Next would be "2" for the network connection.  A dead battery pack will 
also cause you similar trouble (I have replaced the original one with a 
mechanically compatible three-cell NiMH cordless phone pack, which seems 
good enough, or at least have served me for 10+ years).

 The most reliable way to force headless operation is to remove the VSIMM.  
The system will then switch to the serial port console automatically.  Or 
you can set the "console" environment variable appropriately, but it needs 
a live battery pack to survive a power cycle.

> But I now think I may know why I didn't have it out where I could run
> it: after netbooting it a couple of times (took me forever to find the
> command - btw, is there documentation on the console command language?

 See the operator's guide for most commands.

> maybe in more vs31mmg2.pdf?) and in the process discovering I had a
> NetBSD/pmax netboot setup set up, it then seems to have decided to stop
> working:
> 
> KN01 V7.0 
> 08-00-2b-14-8c-7e
> 0x00000000

 NB the hex number above is the amount of RAM the console firmware thinks 
the system has.  At zero it may well be confused, you do want to complete 
the power-on diagnostics.  A successful power-up looks like:

KN01 V7.01

7..6..5..4..3..2..1..0

24Mb..........0

KN01 V7.01
08-00-2b-xx-xx-xx
0x01800000
>> 

(of course the amount of RAM will vary).

> >> boot -f tftp()
> 1845632+65024+374812-47c xfr addr: 0x80030000
> 
> Excptn:    <vtr=NRML>
> Excptn pc: 0x801c9670
> Creg:      0x1000001c<CE=1,EXC=DBE>

 A Data Bus Error exception.  Sadly little information is given here as to
the circumstances and it could be just about anything, from a software bug 
to all kinds of hardware faults, such as a bus timeout or a memory parity 
error (did I suggest cleaning memory modules?).

> Sreg:      0x20080000<CU1,CM,IPL=8>
> Sp:        0x8002ffb0
> exit(-1)
> 
> KN01 V7.0 
> 08-00-2b-14-8c-7e
> 0x00000000
> >> 
> 
> So I'm not sure where to go from here.  The power supply fans seem to
> be moving a reasonable amount of air, but on the possibility that it
> was something thermal, I opened it up and removed the disk mounting
> plate, so the PCB is in the open.  For two consecutive powerups it
> didn't even banner.  Then it bannered, but, on booting, gave the above
> "Excptn".
> 
> I am feeling I may have dead hardware on my hands. :-(

 So far I haven't seen anything truly fatal, these all seem usual issues 
with a piece of old hardware that hasn't seen use for a while and got a 
bit dusty, etc.  Try issuing:

>> t a

at the console prompt for extensive testing and do consider cleaning 
memory modules.  You may try leaving only the first bank occupied.

  Maciej