[rescue] SPARCstation 1 boot woes

[David McMackins II]

Alright, I've been doing some testing with the power supply and I am
not convinced that this is the problem.

I disconnected the floppy drive from power so I could attach probes to
its power connector on the main board (and therefore on the other side
of the potentially oxidized PSU connector). I then ran several tests
with the multimeter watching the +5V or +12V rails, with and without
the video card installed. The only thing I noticed was very early after
powering on, the +12 rail dips slightly below 12 volts (but only very
slightly) and recovers after a few seconds (well before the actual boot
sequence begins). Otherwise both rails seem to be rock solid.

This morning I also got out the oscilloscope to watch for any
transients that might have been too fast for the multimeter to see, and
again they appear to be very very stable right at 5 and 12 volts. I see
no fluctuation at all at the time of the "illegal instruction" message.

Before these tests, I also tested the hypothesis of failure under load
by writing some CPU-intensive Forth programs like prime factorization,
and they ran completely without issue.

I have not actually tried applying contact cleaner to the PSU
connection, but these results (as well as a visual inspection) do not
seem to indicate a problem there or with the power supply itself.

Upon further consideration of the symptom, a power supply issue also
seems unlikely because the boot firmware recovers cleanly from the
malfunction and returns to the "ok" prompt. I would expect a power-
related malfunction to send it off into executing nonsense and crashing
or just resetting the system.

Any other ideas?

-- 
Regards,

David E. McMackins II
www.mcmackins.org

On Tue, 2025-10-21 at 20:04 -0400, John Hudak wrote:
> An increased electrical load during boot was my thought as well.  Can
> you remove a board (or two) to reduce the quiescent load?  I'd also
> consider checking power and ground connections to the main board. 
> Oxidation buildup on power connections could drop the voltage just
> enough to have weird behavior.  It's a long shot but it is a cheap
> and easy task. Can you monitor the voltages with a volt  meter during
> a net boot?
>  
> 
> On Tue, Oct 21, 2025 at 7:11 PM David McMackins II via rescue
> <rescue at sunhelp.org> wrote:
> > It is with great dismay that I report that this netboot setup
> > exhibits
> > the same behavior as current NetBSD. It seems there is a subtle
> > hardware issue after all.
> > 
> > The only idea that I've come up with as a shot in the dark is that
> > it
> > could be a power supply issue of some sort. That is, during the
> > boot
> > sequence (or POST sequence on OBP 1.0), the power demand from the
> > system increases, and one of the PSU voltages drops below a
> > critical
> > threshold and causes malfunctions.
> > 
> > I don't have a spare SS1 power supply laying around, though, and
> > I'm
> > not really sure how I'd go about testing this hypothesis otherwise,
> > unless I could measure something very obviously wrong.
> >