SMPS Behaviour at low supply voltages...

[Cursitor Doom]

Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter. This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.

 

[Fred Bloggs]

On Sunday, September 10, 2023 at 5:25:22 AM UTC-4, Cursitor Doom wrote:

Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter. This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.

Nichicon specializes in very low leakage e-caps mainly intended for timing. Their reforming procedure is for testing/ quality control purposes, and not intended for general applications for which it is mostly inapplicable.

See the section on reforming:

https://www.tdk-electronics.tdk.com/download/185386/e724fb43668a157bc547c65b0cff75f8/pdf-generaltechnicalinformation.pdf

 

[John Larkin]

On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <cd@notformail.com>
wrote:

Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.

We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.




This gives the oxide layer a chance to

reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.

We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

 

[Fred Bloggs]

On Sunday, September 10, 2023 at 10:55:18 AM UTC-4, John Larkin wrote:

On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

 

[Cursitor Doom]

On Sun, 10 Sep 2023 08:08:31 -0700 (PDT), Fred Bloggs
<bloggs.fredbloggs.fred@gmail.com> wrote:

On Sunday, September 10, 2023 at 10:55:18?AM UTC-4, John Larkin wrote:
On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

I\'m afraid neither you nor John (undoubtedly august personages that
you are) have actually answered the question....

 

[legg]

On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <cd@notformail.com>
wrote:

Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter. This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.

You can refer to MIL-HDBK-1131 for electrolytic cap storage and
reformation recommendations.

They suggest limiting current to 5mA, below the rated voltage of
the part, as the control during reforming.

RL

 

[piglet]

On 10/09/2023 16:31, Cursitor Doom wrote:

On Sun, 10 Sep 2023 08:08:31 -0700 (PDT), Fred Bloggs
bloggs.fredbloggs.fred@gmail.com> wrote:

On Sunday, September 10, 2023 at 10:55:18?AM UTC-4, John Larkin wrote:
On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

I\'m afraid neither you nor John (undoubtedly august personages that
you are) have actually answered the question....

Isn\'t the app note talking about reforming individual capacitors? In a
completed piece of equipment you\'d need to do that to each capacitor.
Possibly even having to unsolder at least one lead. Imagine an
electrolytic bypassing a low value emitter resistor in an amplifier - a
10K resistor there would have no chance against that in-circuit load.

If you mean putting 10k in series with the mains supply then the short
answer is NO. The first capacitor after the rectifier may see a little
voltage but I doubt the supply would startup and certainly no other
downstream capacitors would get reformed.

piglet

 

[Cursitor Doom]

On Sun, 10 Sep 2023 17:28:50 +0100, piglet <erichpwagner@hotmail.com>
wrote:

On 10/09/2023 16:31, Cursitor Doom wrote:
On Sun, 10 Sep 2023 08:08:31 -0700 (PDT), Fred Bloggs
bloggs.fredbloggs.fred@gmail.com> wrote:

On Sunday, September 10, 2023 at 10:55:18?AM UTC-4, John Larkin wrote:
On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

I\'m afraid neither you nor John (undoubtedly august personages that
you are) have actually answered the question....

Isn\'t the app note talking about reforming individual capacitors?

Yes.

In a
completed piece of equipment you\'d need to do that to each capacitor.
Possibly even having to unsolder at least one lead. Imagine an
electrolytic bypassing a low value emitter resistor in an amplifier - a
10K resistor there would have no chance against that in-circuit load.

If you mean putting 10k in series with the mains supply then the short
answer is NO. The first capacitor after the rectifier may see a little
voltage but I doubt the supply would startup and certainly no other
downstream capacitors would get reformed.

piglet

Oh bugger. I hadn\'t thought of that.Thanks, Piglet.

 

[John Larkin]

On Sun, 10 Sep 2023 16:31:59 +0100, Cursitor Doom <cd@notformail.com>
wrote:

On Sun, 10 Sep 2023 08:08:31 -0700 (PDT), Fred Bloggs
bloggs.fredbloggs.fred@gmail.com> wrote:

On Sunday, September 10, 2023 at 10:55:18?AM UTC-4, John Larkin wrote:
On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

I\'m afraid neither you nor John (undoubtedly august personages that
you are) have actually answered the question....

Demand a refund!

 

[Dan Green]

On Sun, 10 Sep 2023 11:01:01 -0700, John Larkin
<jjlarkin@highlandtechnology.com> wrote:

On Sun, 10 Sep 2023 16:31:59 +0100, Cursitor Doom <cd@notformail.com
wrote:

On Sun, 10 Sep 2023 08:08:31 -0700 (PDT), Fred Bloggs
bloggs.fredbloggs.fred@gmail.com> wrote:

On Sunday, September 10, 2023 at 10:55:18?AM UTC-4, John Larkin wrote:
On Sun, 10 Sep 2023 10:25:13 +0100, Cursitor Doom <c...@notformail.com
wrote:
Gentlemen,

An app note I was referred to recently (Nichicon source IIRC) stated
that even new electrolytic caps should have their oxide layer reformed
before being put into service if they\'ve been on the shelf for as
little as 12 months. They suggested doing this by charging them to 90%
of max stated voltage via a 1k resistor and leaving them fully charged
for 30 minutes. thereafter.
We don\'t do that in production, and we surely have caps in stock that
are several years old. I think elec caps are much better than they
used to be.
This gives the oxide layer a chance to
reform, they say. It\'s actually quite a rapid charge in most cases and
certainl much faster than we would use for bringing vintage/NOS caps
back to a serviceable condition. But for convenience you can\'t beat
it. Anyway, cut a long story short. Previously I\'ve brought old test
gear back to life by manually bringing the supply voltage up via a
variac and monitoring the current for any \'abberent behaviour\' which
is rather time consuming. In future I think I\'ll just use a 10k say
resistor in the hot wire and leave it to its own devices, plugged in
and switched on for 12 hours. That should be suitably kind to vintage
equipment. But then I thought - would this still work if the equipment
in question had an SMPS? It\'s fine for linear PSUs, but SMPS? Anyone
care to offer an informed opinion?

CD.
We don\'t do that in production, and we surely have caps that are
several years old. I think elec caps are much better than they used to
be.

I always wondered what that 30 second overvoltage rating was all about, then I powered up a largely vacuum tube product and watched that voltage zoom into overvoltage until the filaments heated up and pulled the old fashioned regulator output to operating voltage.

I\'m afraid neither you nor John (undoubtedly august personages that
you are) have actually answered the question....

Demand a refund!

Good idea. The Larkin\'s advice is worth exactly what you paid for it.

 

[Phil Allison]

Fred Bloggs wrote:
-------------------------------

I always wondered what that 30 second overvoltage rating was all about,
then I powered up a largely vacuum tube product and watched that voltage
zoom into overvoltage until the filaments heated up and pulled the old
fashioned regulator output to operating voltage.

** IMO allowing PSU electros to ever go over voltage is piss poor design..

At some stage during production testing or regular servicing, tube based items may need to be operated on test with some or all the tubes removed - in which case a 30 second time limit is absurdly brief.
Supply voltages, negative grid bias voltages and drive signals need to be verified to be within tolerance BEFORE expensive new tubes are installed. When some a problem emerges, one does not need the clock ticking before PSU electros turn into firecrackers to worry about during trouble shooting.


....... Phil

 

[Fred Bloggs]

On Monday, September 11, 2023 at 3:39:51 AM UTC-4, Phil Allison wrote:

Fred Bloggs wrote:
-------------------------------


I always wondered what that 30 second overvoltage rating was all about,
then I powered up a largely vacuum tube product and watched that voltage
zoom into overvoltage until the filaments heated up and pulled the old
fashioned regulator output to operating voltage.
** IMO allowing PSU electros to ever go over voltage is piss poor design.

That\'s the way they did at the time. The one piece of equipment I observed was 1940s broadcast band AM receiver.

At some stage during production testing or regular servicing, tube based items may need to be operated on test with some or all the tubes removed - in which case a 30 second time limit is absurdly brief.
Supply voltages, negative grid bias voltages and drive signals need to be verified to be within tolerance BEFORE expensive new tubes are installed. When some a problem emerges, one does not need the clock ticking before PSU electros turn into firecrackers to worry about during trouble shooting.

Aren\'t most of the small signal tubes biased for 10mA or something? Not going to take much to hang a load on the power supply to draw it down, something like a wirewound rheostat. The other tube electrodes are biased with uA, in effect nothing.

...... Phil