I've seen this before, about a year ago, then it happened again in
another BA356: Disks go offline and into mount verification, then come
back after a minute or so. The process repeats every ten minutes. Of
course, being VMS, the cluster continues and practically nothing else is
affected, but it is annoying.

Replacing the power supply solves the problem.

I understand that a power supply can fail, but why is it failing in this
manner? Maybe overheating and shutting off then coming back on?

Being VMS, I dismounted the shadow-set members in the box with the flaky
power supply. Thanks to minicopy, things were quickly back to normal
after replacing the power supply. To do that, I turned off the box,
replaced the power supply, and turned it back on. The node whose
system-disk shadow set was in the box didn't crash.

I like VMS.

Now to find a few more replacement power supplies. When collecting
SBBs, I concentrated on disks, and I think that I have enough of them.
Since the first power-supply failure wasn't until about a year ago, I
hadn't explicitly looked for them, but fortunately got a few with BA356
and other SBBs (mostly disks).

I need the Top Gun Blue (that's really the official colour!) 180W
versions. If anyone has any in Scandinavia, Germany, Benelux, England,
Switzerland, Austria, Slovenia, or northwestern Croatia I could probably
collect them sometime this year, probably before autumn. Rest of the
Balkan: next year.

I also plan to be in the southeast USA soon so would be happy to collect
there as well. (Apparently they run on 50 or 60 Hz and anything between
100 and 240 volts.)

It's mainly the power supplies I need, but I'll take any blue SBBs.

Den 2022-03-28 kl. 11:24, skrev Phillip Helbig (undress to reply):
> I've seen this before, about a year ago, then it happened again in
> another BA356: Disks go offline and into mount verification, then come
> back after a minute or so. The process repeats every ten minutes. Of
> course, being VMS, the cluster continues and practically nothing else is
> affected, but it is annoying.
>
> Replacing the power supply solves the problem.
>
> I understand that a power supply can fail, but why is it failing in this
> manner? Maybe overheating and shutting off then coming back on?
>
> Being VMS, I dismounted the shadow-set members in the box with the flaky
> power supply. Thanks to minicopy, things were quickly back to normal
> after replacing the power supply. To do that, I turned off the box,
> replaced the power supply, and turned it back on. The node whose
> system-disk shadow set was in the box didn't crash.
>
> I like VMS.
>
> Now to find a few more replacement power supplies. When collecting
> SBBs, I concentrated on disks, and I think that I have enough of them.
> Since the first power-supply failure wasn't until about a year ago, I
> hadn't explicitly looked for them, but fortunately got a few with BA356
> and other SBBs (mostly disks).
>
> I need the Top Gun Blue (that's really the official colour!) 180W
> versions. If anyone has any in Scandinavia, Germany, Benelux, England,
> Switzerland, Austria, Slovenia, or northwestern Croatia I could probably
> collect them sometime this year, probably before autumn. Rest of the
> Balkan: next year.
>
> I also plan to be in the southeast USA soon so would be happy to collect
> there as well. (Apparently they run on 50 or 60 Hz and anything between
> 100 and 240 volts.)
>
> It's mainly the power supplies I need, but I'll take any blue SBBs.
>

PSUs are aging. The electrolytic capacitators are drying out. This will
make the voltages to become more unstable and causing intermitent
issues before going totaly "off". Quite normal, I'd say.

Some makes a buiness out of replacing old capacitors in power supplies
for TV sets and other home/kitchen appliences. I guess that is equially
possible in these PSUs. They are not too modern and should be fine to
work on without special soldering equipment.

On Monday, March 28, 2022 at 5:24:53 AM UTC-4, Phillip Helbig (undress to reply) wrote:
> I've seen this before, about a year ago, then it happened again in
> another BA356: Disks go offline and into mount verification, then come
> back after a minute or so. The process repeats every ten minutes. Of
> course, being VMS, the cluster continues and practically nothing else is
> affected, but it is annoying.
>
> Replacing the power supply solves the problem.
>
> I understand that a power supply can fail, but why is it failing in this
> manner? Maybe overheating and shutting off then coming back on?
>
> Being VMS, I dismounted the shadow-set members in the box with the flaky
> power supply. Thanks to minicopy, things were quickly back to normal
> after replacing the power supply. To do that, I turned off the box,
> replaced the power supply, and turned it back on. The node whose
> system-disk shadow set was in the box didn't crash.
>
> I like VMS.
>
> Now to find a few more replacement power supplies. When collecting
> SBBs, I concentrated on disks, and I think that I have enough of them.
> Since the first power-supply failure wasn't until about a year ago, I
> hadn't explicitly looked for them, but fortunately got a few with BA356
> and other SBBs (mostly disks).
>
> I need the Top Gun Blue (that's really the official colour!) 180W
> versions. If anyone has any in Scandinavia, Germany, Benelux, England,
> Switzerland, Austria, Slovenia, or northwestern Croatia I could probably
> collect them sometime this year, probably before autumn. Rest of the
> Balkan: next year.
>
> I also plan to be in the southeast USA soon so would be happy to collect
> there as well. (Apparently they run on 50 or 60 Hz and anything between
> 100 and 240 volts.)
>
> It's mainly the power supplies I need, but I'll take any blue SBBs.

I have spares and can send you one. Can you send me the failing one? I can check to see the failure cause and hopefully repair it.
Contact me offline.

Dan

On 2022-03-28, Phillip Helbig (undress to reply) <helbig@asclothestro.multivax.de> wrote:
> I've seen this before, about a year ago, then it happened again in
> another BA356: Disks go offline and into mount verification, then come
> back after a minute or so. The process repeats every ten minutes. Of
> course, being VMS, the cluster continues and practically nothing else is
> affected, but it is annoying.
>

Careful Phillip. How do you know that you are not getting silent data
corruption if your power supplies are in this state ?

>
> I need the Top Gun Blue (that's really the official colour!) 180W
> versions. If anyone has any in Scandinavia, Germany, Benelux, England,
> Switzerland, Austria, Slovenia, or northwestern Croatia I could probably
> collect them sometime this year, probably before autumn. Rest of the
> Balkan: next year.
>

Any part of Europe you have _not_ driven through ? :-)

> I also plan to be in the southeast USA soon so would be happy to collect
> there as well. (Apparently they run on 50 or 60 Hz and anything between
> 100 and 240 volts.)
>

That could make for an interesting discussion when you try leaving
the US again. :-)

Simon.

--
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Walking destinations on a map are further away than they appear.

In article <t1rut2$uku$1@gioia.aioe.org>,
Phillip Helbig (undress to reply) <helbig@asclothestro.multivax.de> wrote:
>I've seen this before, about a year ago, then it happened again in
>another BA356: Disks go offline and into mount verification, then come
>back after a minute or so. The process repeats every ten minutes. Of
>course, being VMS, the cluster continues and practically nothing else is
>affected, but it is annoying.
>
>Replacing the power supply solves the problem.
>
>I understand that a power supply can fail, but why is it failing in this
>manner? Maybe overheating and shutting off then coming back on?

Yes, the supply is likely crowbarring itself. Replace all of the
electrolytic capacitors on the secondary side of the board. While you
are at it, you might pre-emptively replace the kickstart cap since they
are prone to fail in those cabinets.

>Now to find a few more replacement power supplies. When collecting
>SBBs, I concentrated on disks, and I think that I have enough of them.
>Since the first power-supply failure wasn't until about a year ago, I
>hadn't explicitly looked for them, but fortunately got a few with BA356
>and other SBBs (mostly disks).

Just fix it. No reason to replace the whole thing.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

On Monday, March 28, 2022 at 2:55:25 AM UTC-7, Jan-Erik Söderholm wrote:

(snip)

> PSUs are aging. The electrolytic capacitators are drying out. This will
> make the voltages to become more unstable and causing intermitent
> issues before going totaly "off". Quite normal, I'd say.

It seems to me that this is true for everything. I have replaced
capacitors in fluorescent lamp ballasts, though usually try to buy
new ones. Computers are usually built with better ones, but some
things last about a year. Digital TV converter boxes are some of
the worst in terms of capacitor life.

> Some makes a buiness out of replacing old capacitors in power supplies
> for TV sets and other home/kitchen appliences. I guess that is equially
> possible in these PSUs. They are not too modern and should be fine to
> work on without special soldering equipment.

It is nice to have a hot air soldering unit, though that might not do
everything that is needed today. You need something fancier for
BGA chips, though.

Otherwise, I would probably replace both primary and secondary
side capacitors, though I suspect less problems primary side.

In article <t1s8ml$baf$1@dont-email.me>, Simon Clubley
<clubley@remove_me.eisner.decus.org-Earth.UFP> writes:

> On 2022-03-28, Phillip Helbig (undress to reply) <helbig@asclothestro.multivax.de> wrote:
> > I've seen this before, about a year ago, then it happened again in
> > another BA356: Disks go offline and into mount verification, then come
> > back after a minute or so. The process repeats every ten minutes. Of
> > course, being VMS, the cluster continues and practically nothing else is
> > affected, but it is annoying.
>
> Careful Phillip. How do you know that you are not getting silent data
> corruption if your power supplies are in this state ?

I replace them when I notice it. My guess is that HBVS knows what to do
here.

> > I need the Top Gun Blue (that's really the official colour!) 180W
> > versions. If anyone has any in Scandinavia, Germany, Benelux, England,
> > Switzerland, Austria, Slovenia, or northwestern Croatia I could probably
> > collect them sometime this year, probably before autumn. Rest of the
> > Balkan: next year.
>
> Any part of Europe you have _not_ driven through ? :-)

Ireland, Luxembourg, the Baltic states, Poland, Slovakia, Bulgaria,
Romania, Greece, Monaco, Vatican City, San Marino, Andorra,
Liechtenstein, Cyprus, Malta, Iceland, Bosnia and Herzogovina, Kosovo.
So I've been to 22 of 42. (Defining "Europe" here to include former
Soviet-bloc countries but not any from the USSR (except the Baltic
states) nor Turkey.) Of the 20 I haven't visited, 6 are (almost)
microstates, 4 are islands (one is both). Poland, Slovakia, Bulgaria,
Romania, the Baltic states, and Greece are all at the Eastern border,
too far from the beaten path. I've had no reason to go to Kosovo or
Bosnia and Herzogovina, though I've been to the other Balkan countries.
Luxemburg is only a couple of hours from here, so that is an anomaly.
:-)

> > I also plan to be in the southeast USA soon so would be happy to collect
> > there as well. (Apparently they run on 50 or 60 Hz and anything between
> > 100 and 240 volts.)
>
> That could make for an interesting discussion when you try leaving
> the US again. :-)

I don't think that old DEC hardware is considered to be weapons export
anymore.

On 3/28/22 16:25, Phillip Helbig (undress to reply) wrote:
>
> Luxemburg is only a couple of hours from here, so that is an anomaly.
> :-)

It's so small you probably blinked and missed it. Worth a
visit sometime, though. They make some real good White
Mosel wine.

bill

gah4 <gah4@u.washington.edu> wrote:
>
>Otherwise, I would probably replace both primary and secondary
>side capacitors, though I suspect less problems primary side.

In the case of that specific supply, the problem is that the two big banks
of secondary side capacitors run hot because of all the high frequency
ripple. So they fail earlier than they might.

The filter stuff on the primary side seems to last much longer, except for
the kickstar cap and I don't know why that one seems to fail so often.

NOW... I will say that I have only experience with these supplies used on
120V and if you run them on 240V then the two big stab-in capacitors on
the high voltage input will have twice as much voltage across them and this
may shorten their life. So you may wish to replace those pre-emptively if
you are in a 240V country. I have never seen one fail in the US.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

On Monday, March 28, 2022 at 3:00:05 PM UTC-7, Scott Dorsey wrote:

(snip)

> The filter stuff on the primary side seems to last much longer, except for
> the kickstar cap and I don't know why that one seems to fail so often.

> NOW... I will say that I have only experience with these supplies used on
> 120V and if you run them on 240V then the two big stab-in capacitors on
> the high voltage input will have twice as much voltage across them and this
> may shorten their life. So you may wish to replace those pre-emptively if
> you are in a 240V country. I have never seen one fail in the US.

The popular design some years ago (as in when I was actually looking at
how they were built) is a bridge rectifier in 240V countries, and voltage
double in 120V countries. So the primary side capacitors run
about 300V in both cases. That is, for ones with a 120/240 switch.

Also, it used to be usual to run the oscillator at about 20kHz.
Some now might run much higher.

When working with the computer museum, we would replace all the
electrolytic capacitors before powering up the first time. That is,
for ones that might be 30 or 40 or 50 years old.

They used to make "computer grade" capacitors. I don't know
if they still do that.

In any case, the thing that electrolytics least like is being
reverse charged. I don't know about the kickstart, but it
is possible that it gets reverse charged, maybe just a little.

On 03/29/22 01:36, gah4 wrote:
> On Monday, March 28, 2022 at 3:00:05 PM UTC-7, Scott Dorsey wrote:
>
> (snip)
>
>> The filter stuff on the primary side seems to last much longer, except for
>> the kickstar cap and I don't know why that one seems to fail so often.
>
>> NOW... I will say that I have only experience with these supplies used on
>> 120V and if you run them on 240V then the two big stab-in capacitors on
>> the high voltage input will have twice as much voltage across them and this
>> may shorten their life. So you may wish to replace those pre-emptively if
>> you are in a 240V country. I have never seen one fail in the US.
>
> The popular design some years ago (as in when I was actually looking at
> how they were built) is a bridge rectifier in 240V countries, and voltage
> double in 120V countries. So the primary side capacitors run
> about 300V in both cases. That is, for ones with a 120/240 switch.
>
> Also, it used to be usual to run the oscillator at about 20kHz.
> Some now might run much higher.
>
> When working with the computer museum, we would replace all the
> electrolytic capacitors before powering up the first time. That is,
> for ones that might be 30 or 40 or 50 years old.
>
> They used to make "computer grade" capacitors. I don't know
> if they still do that.
>
> In any case, the thing that electrolytics least like is being
> reverse charged. I don't know about the kickstart, but it
> is possible that it gets reverse charged, maybe just a little.
>
>
>

A serious interest here is older workstation restoration and
power supplies are usually the first thing to look at. The
capacitors dry out over time, dues to high temperatures and many
do need to be replaced. Input capacitors often test ok, but the
output rail caps seem to be underspecified and get so hot that
they leak electrolyte all over the pcb as they fail. That can be
fixed by washing and drying, but only the best caps should be
used for replacement. Choose 105 C and low esr rated caps,
as these modern types are often far better than the
originals.

Many of the older systems were poorly designed for airflow and
internal sleeve bearing fans should be replaced with ball
bearing types of higher output if possible...

Chris

gah4 <gah4@u.washington.edu> wrote:
>
>The popular design some years ago (as in when I was actually looking at
>how they were built) is a bridge rectifier in 240V countries, and voltage
>double in 120V countries. So the primary side capacitors run
>about 300V in both cases. That is, for ones with a 120/240 switch.

Right, you don't see that very often anymore. You will still see a switch now
and then, but these days the PWM range is so enormous that there is no need.

>When working with the computer museum, we would replace all the
>electrolytic capacitors before powering up the first time. That is,
>for ones that might be 30 or 40 or 50 years old.

This is often a good plan.

>They used to make "computer grade" capacitors. I don't know
>if they still do that.

Computer grade capacitors hardly ever fail, but they also have incredibly
high ESR and very poor high frequency performance. They are designed for
60 Hz ripple filtering and they do very well for that... they can handle a
lot of ripple at 60 Hz. They are also expensive and died out pretty much
with linear supplies in the eighties.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."