Capacitors in PSU are dangerous?

kony wrote:

.... snip ...

Power-on, charged voltage of cap (Rubycon 680mF) was 158V.
Within a couple hours it had lost over half of it's charge, below
70V. Nearly 8 hours later it was below 10V and right now, 14 hours
later, is at about 7.5V and draining so slowly that further
measurement might better be made in days rather than hours.

Are you leaving the meter connected or just sampling? Most DVMs
have a constant input impedance of about 10 Meg. Multimeters can
be expected to have anywhere from 1k to 50k ohms per volt full
scale.

It all falls out of Ohms law and Q = CE.

--
Churchill and Bush can both be considered wartime leaders, just
as Secretariat and Mr Ed were both horses. - James Rhodes.

kony wrote:
On Thu, 15 Apr 2004 23:49:24 +0100, Tim Auton
tim.auton@uton.[groupSexWithoutTheY] wrote:
[snip]
I'm assuming the worst case - ie component failure, where the only
discharge of the caps is self-discharge. In that case though, 24 hours
may not be enough. Hmmm, I think we need some experimental data.

Well the data is a bit slow in coming but I did a sloppy test yesterday...

I took a board from a power supply, that is quite typical, the basic
filtering components between the AC socket and the rectifier, NOTHING
after the rectifier... no bleeder resistors, no further power supply
components at all beyond the rectifier By attaching the capacitor
directly to rectifier output we would have an absolute worst case
scenario, there is no way for the cap to drain slower than that no matter
what else had failed in a power supply.

Power-on, charged voltage of cap (Rubycon 680mF) was 158V.
Within a couple hours it had lost over half of it's charge, below 70V.
Nearly 8 hours later it was below 10V and right now, 14 hours later, is at
about 7.5V and draining so slowly that further measurement might better be
made in days rather than hours.

Most interesting, thanks for the data.

I've a dead 400W ATX PSU lying around so I may duplicate your
experiment, but I've a telescope mount to reassemble before doing that
so don't hold your breath.


Tim
--
Love is a travelator.

kony wrote:

On Thu, 15 Apr 2004 23:49:24 +0100, Tim Auton
tim.auton@uton.[groupSexWithoutTheY] wrote:


It's beyond overkill to advise waiting 24 hours. Even if you didn't know
what you were doing you should've known that there's another very obvious
way an ATX power supply drains besides the bleeder resistors.

I'm assuming the worst case - ie component failure, where the only
discharge of the caps is self-discharge. In that case though, 24 hours
may not be enough. Hmmm, I think we need some experimental data.


Tim


Well the data is a bit slow in coming but I did a sloppy test yesterday...

I took a board from a power supply, that is quite typical, the basic
filtering components between the AC socket and the rectifier, NOTHING
after the rectifier... no bleeder resistors, no further power supply
components at all beyond the rectifier By attaching the capacitor
directly to rectifier output we would have an absolute worst case
scenario, there is no way for the cap to drain slower than that no matter
what else had failed in a power supply.

Power-on, charged voltage of cap (Rubycon 680mF) was 158V.
Within a couple hours it had lost over half of it's charge, below 70V.
Nearly 8 hours later it was below 10V and right now, 14 hours later, is at
about 7.5V and draining so slowly that further measurement might better be
made in days rather than hours.

You're measuring the time constant of the internal, (parallel), resistance
of the capacitor. This can vary widely with the type of capacitor and even
its previous charge state. From a safety point of view its energy
content before its terminal voltage has dropped to a safe level may be more
important. In other words does its effective capacitance remain the same
during the whole of a self discharge cycle?

Given that this was only a single cap and that I now have a better idea of
time inteval for measurements, I may try another cap soon of higher
capacity.

Do you know what value of capacitors are used in typical PSUs today? My
sample is limited to those with 470 mfd capacitors.

I made some measurements on a cheap L&C LC-235ATX supply. The CB inside
was marked "DEER", not one of the best quality makers.

Firstly, it was almost impossible to get even my finger tips on any
high, (DC), voltage points. The switching transistors were pretty
well protected by the insulated body of the input capacitors. I had to
remove four screws to get to the point under the CB to measure the
capacitor voltage.

I shorted pin 14 to ground and applied AC power. The unit started
without any exterior load and the output voltages were within normal
tolerances. The voltage across each 470 mfd capacitor was 162 V.
It dropped to essentially zero in about a second. Removing the ground
to cause the unit to not start, did not affect this behavior. As you
previously noted, the +5V SB did the same.

I did not remove the 220 ohm bleeders, but this time constant
is about 100 seconds. These bleeders are needed to equalize the
voltage between the input capacitors, so it's unlikely they'd be not used
to save costs. It seems the switching circuitry rapidly discharges the
capacitors, even when the supply is inactive.

I still urge everyone to use normal safety precautions, as with any
electrical device. The line power input terminals were exposed, so
it's important to be sure the unit is unplugged. No one without good
electrical experience should trouble shoot a PSU with applied voltage,
though it almost a necessity to do so to get meaningful results.

Repairs other than replacing the fan are probably not cost effective.

Virg Wall
--

It is vain to do with more
what can be done with fewer.
William of Occam.

kony wrote:

On Mon, 19 Apr 2004 16:33:05 +0100, "half_pint"
wrote:

snip

I guess it depends on the internal resistance of the capacitor, which is
probably
pretty negligible ( I have never ever seen it on a circuit diagram).

Don't guess.

Don't be so condesending.


Instead of guessing, do some research first. If I+V/R does that mean
there is nothing else to know about capacitors? No.



You don't have a clue do you?
Try inceasing your vocabulary to that of a 5 year old.



It seems as though you've taken to pulling something out of your ass
every 3rd post, something that would take more than a few lines to explain
till you realized why your guess was wrong. I'm not going to write a 500
word essay every time you can't be bothered to do some prerequisite
research/learning before making a "guess".


He should change his name to "half_wit", but it still might be overstating
the capacity!

Virg Wall
--

It is vain to do with more
what can be done with fewer.
William of Occam.

"kony" wrote:
I took a board from a power supply, that is quite typical, the basic
filtering components between the AC socket and the rectifier, NOTHING
after the rectifier... no bleeder resistors, no further power supply
components at all beyond the rectifier By attaching the capacitor
directly to rectifier output we would have an absolute worst case
scenario, there is no way for the cap to drain slower than that no matter
what else had failed in a power supply.


What about the reverse leakage current in the rectifier? Doesn't
that amount to a bleeder resistor?

*TimDaniels*

On Mon, 19 Apr 2004 17:56:34 GMT, CBFalconer wrote:

kony wrote:

... snip ...

Power-on, charged voltage of cap (Rubycon 680mF) was 158V.
Within a couple hours it had lost over half of it's charge, below
70V. Nearly 8 hours later it was below 10V and right now, 14 hours
later, is at about 7.5V and draining so slowly that further
measurement might better be made in days rather than hours.

Are you leaving the meter connected or just sampling? Most DVMs
have a constant input impedance of about 10 Meg. Multimeters can
be expected to have anywhere from 1k to 50k ohms per volt full
scale.

It all falls out of Ohms law and Q = CE.

Just samping for a few seconds, though at the time I had forgotten about
it for a while, other work needed done, so I had no readings between 2 and
8 hour interval.

On Mon, 19 Apr 2004 11:26:12 -0700, "Timothy Daniels"
wrote:


"kony" wrote:
I took a board from a power supply, that is quite typical, the basic
filtering components between the AC socket and the rectifier, NOTHING
after the rectifier... no bleeder resistors, no further power supply
components at all beyond the rectifier By attaching the capacitor
directly to rectifier output we would have an absolute worst case
scenario, there is no way for the cap to drain slower than that no matter
what else had failed in a power supply.


What about the reverse leakage current in the rectifier? Doesn't
that amount to a bleeder resistor?

*TimDaniels*

I suppose I could disconnect the rectifier and retest though the rectifier
is going to be in the theoretic power supply as a whole too, so it's
somewhat of a constant.

"VWWall" wrote in message
k.net...
kony wrote:

On Mon, 19 Apr 2004 16:33:05 +0100, "half_pint"
wrote:

snip

I guess it depends on the internal resistance of the capacitor, which
is
probably
pretty negligible ( I have never ever seen it on a circuit diagram).

Don't guess.

Don't be so condesending.


Instead of guessing, do some research first. If I+V/R does that mean
there is nothing else to know about capacitors? No.



You don't have a clue do you?
Try inceasing your vocabulary to that of a 5 year old.



It seems as though you've taken to pulling something out of your ass
every 3rd post, something that would take more than a few lines to
explain
till you realized why your guess was wrong. I'm not going to write a
500
word essay every time you can't be bothered to do some prerequisite
research/learning before making a "guess".


He should change his name to "half_wit", but it still might be overstating
the capacity!


Perhaps you shoild change your name to "Bucket of ****".
You would need to find a bucket first though.


Virg Wall
--

It is vain to do with more
what can be done with fewer.
William of Occam.

"VWWall" wrote in message
nk.net...

Repairs other than replacing the fan are probably not cost effective.

Virg Wall

The only PSU I had where the fan failed pretty much melted itself into a
heap of slag before the rest of it failed. It was only then that I found out
about the problem. Fortunately, it did not take anything else with it when
its soul went off to that great electronic warehouse in the sky.

On Mon, 19 Apr 2004 18:01:36 GMT, VWWall wrote:


You're measuring the time constant of the internal, (parallel), resistance
of the capacitor. This can vary widely with the type of capacitor and even
its previous charge state. From a safety point of view its energy
content before its terminal voltage has dropped to a safe level may be more
important. In other words does its effective capacitance remain the same
during the whole of a self discharge cycle?

Don't know and no capacitance meter here.


Do you know what value of capacitors are used in typical PSUs today? My
sample is limited to those with 470 mfd capacitors.

Depends a lot on make and wattage rating.
Typically the cheapies or lower wattage units, up to 250-300W, might be as
low as 220mF, but more commonly (perhaps most common) is 470-680, with
1000mF typically found only in some of the higher wattage models.

I made some measurements on a cheap L&C LC-235ATX supply. The CB inside
was marked "DEER", not one of the best quality makers.

Yep, I have some "guts" from a lot of those lying around, keep meaning to
throw them away but I might test one of those too, though I wasn't really
looking for a comprehensive analysis, just a rough idea of how long it'd
take to get under ~30V or so.


Firstly, it was almost impossible to get even my finger tips on any
high, (DC), voltage points. The switching transistors were pretty
well protected by the insulated body of the input capacitors. I had to
remove four screws to get to the point under the CB to measure the
capacitor voltage.

True, though it seems likely that one of the heatsinks is HV still.

I shorted pin 14 to ground and applied AC power. The unit started
without any exterior load and the output voltages were within normal
tolerances. The voltage across each 470 mfd capacitor was 162 V.
It dropped to essentially zero in about a second. Removing the ground
to cause the unit to not start, did not affect this behavior. As you
previously noted, the +5V SB did the same.

I did not remove the 220 ohm bleeders, but this time constant
is about 100 seconds. These bleeders are needed to equalize the
voltage between the input capacitors, so it's unlikely they'd be not used
to save costs. It seems the switching circuitry rapidly discharges the
capacitors, even when the supply is inactive.

I was considering just pulling the bleeders from a unit but already had
the misc parts I'd tried, lying out, "handy". Might get around to doing
that but it may be a while till I get a chance.