Capacitor popping, not sure why?

[esprits300]

Hello, i have a situation that i do not completely understand. i have a capacitor, 35uF @ 950V. i have a regulated DC power supply connecting to it. when i set the voltage to anything above 700V the capacitor pops and sparks. when its below 700, it seems just fine. i was told via forums from the manufacturer that it can easily handle 900V. i have tried this with 3 different capacitors, all the same make/model so I am fairly certain its something with the regulated power supply but i don't know what. i measure the voltage at the capacitor. i am stumped as to why it is popping well below this voltage. forgive me if i left out any info. thank you!

 

[fleem]

Hello, i have a situation that i do not completely understand. i have a capacitor, 35uF @ 950V. i have a regulated DC power supply connecting to it. when i set the voltage to anything above 700V the capacitor pops and sparks. when its below 700, it seems just fine. i was told via forums from the manufacturer that it can easily handle 900V. i have tried this with 3 different capacitors, all the same make/model so I am fairly certain its something with the regulated power supply but i don't know what. i measure the voltage at the capacitor. i am stumped as to why it is popping well below this voltage. forgive me if i left out any info. thank you!

I take it you are watching the polarity? Putting backwards voltage on an electrolytic cap will make it get hot quickly.

 

[yungman]

I hope you did not buy them on ebay shipped from China or Hong Kong. I bought a 1% resistor kid and one value marked 68K read from 70K to 76K!

 

[es1]

How are you determining the voltage on the capacitor is only 700V max? 950V is 700V with about a ~33% overshoot. Depending on the supply particulars such a transient is possible. Say on start-up, for example.

Are you sure it is popping due to over voltage? Maybe it has too much power in it due to its ESR and ripple current.

 

[cmb]

I hope you did not buy them on ebay shipped from China or Hong Kong. I bought a 1% resistor kid and one value marked 68K read from 70K to 76K!

I have similar experience with silicon discretes also...

But electrolitic capacitors tend to just explode in a fuzzing steam of colourful electrolyte if they go bad, rather than 'spark' (yeah.. sometimes you just end up forgetting to monitor the voltage rating on your parts when you're trying out different circuits! A 4400uF 40V cap made quite a mess of my room when I left it at 60V. It wasn't until a few seconds afterwars that I realized there should be no reason for red rain in my room! )

Just one other thing, is your regulated supply a linear or switch mode/buck-boost supply? I expect it is switch mode at that voltage, and your cap is not a film type? It is conceivable that there are kV transients passing out of the supply. Wrap the output leads (or 'live' line if one is earthed) around an EMC ferrite ring, see if there is a change.

 

[esprits300]

i am positive the polarity is correct, it is made here in the USA and i checked the voltage by using my multimeter at the capacitor. i know the ESR is 0.02 Ohm, but honestly that's about it. I am not familiar with how your getting the 950v is 700v with the 33% overshoot. i understand the basics of electronics and schematics but I am not too advanced. my understanding is that you supply 950v and the capacitor charges automatically. (one thing i don't know is if you leave the power supply on will it overcharge or damage it) i am not sure the popping is due to over voltage BUT when i increase the voltage it from ~650-700 it pops almost instantly. so what your saying is that on my power supply, 700v is equal to 950v due to other internal components and such? thank you for all the responses, I've racked my brain against this and I am out of ideas!

 

[cmb]

Can you please cover what this 'popping' sounds like and where it is coming from? Never heard a cap do that before. D'you mean you can hear it bubbling in the can?

 

[esprits300]

sure, beside the obvious LOUD pop noise, balls of blue/white light appear on the negative cable of the capacitor and appear to arc to the casing. its a capacitor surrounded in a gel, that's in a an aluminum enclosure. the wire eventually gets burned and burn marks are on the aluminum enclosure, so I am assuming its trying to get to ground. this help?

 

[cmb]

Well, all I can tell you is that in one home-built application I run >10 kV potentials within a cm or two of each other and never had such effects (all I get is a little bit of corona hiss and ozone), so one has to conclude that additional material is entering the space between the leads and the case to perpetuate the spark. As it is a capacitor, the inevitable conclusion is that the capacitor is venting through the legs. Not only do the caps appear to be unable to take the voltage, but the pressure safety feature at the top is failing to do its job.

What the cause of that is is less easy to tell. In addition to reverse polarity and inductive spikes the only other thing I can think of is that it is taking an undue level of RMS current (from a failing/imperfect regulation). Is there a load across the capacitor? Can you put a 'scope across the capacitor?

 

[esprits300]

no, i don't have a scope. but something you said makes sense. is it possible that in my particular setup that 700v is = 950v? so when i set higher than 700v on my power supply, due to components, that its blowing it due to over voltage? that seems to be a very reasonable answer. also, I've never been clear on this, is it safe to constantly keep my power supply on when connected to the capacitor? i usually turn it off once it hits say 650v. i wasn't sure if it would overcharge or damage it. thank you!

 

[likephysics]

no, i don't have a scope. but something you said makes sense. is it possible that in my particular setup that 700v is = 950v? so when i set higher than 700v on my power supply, due to components, that its blowing it due to over voltage? that seems to be a very reasonable answer. also, I've never been clear on this, is it safe to constantly keep my power supply on when connected to the capacitor? i usually turn it off once it hits say 650v. i wasn't sure if it would overcharge or damage it. thank you!

Are you connecting the capacitor directly to the power supply?
Try connecting thru a series resistor.
Try connecting 2 caps in parallel.

Maybe the regulated DC isn't regulated. What make is it?

 

[es1]

That does sound like over voltage to me. I've seen this when power supplies I designed had grounding issues.

Overshoot is the temporary peak voltage seen on the output of the supply when you turn it on. Depending on how the supply is designed it can be a good faction above the nominal output voltage for a short time.

See for example this supply going 30% above the nominal output voltage during start-up.
http://i.cmpnet.com/powermanagementdesignline/2007/04/TI_Overshoot_Fig3.gif

From this article:
http://www.eetimes.com/design/power...2157/Defeating-turn-on-overshoot?pageNumber=1

Is this for a tube amplifier? I ask because 700V is a common tube bias voltage. Is the power supply a step-up transformer that is wall powered and then diode bridge rectified? Is this cap part of an LC filter or maybe the cap is at the output of the bridge rectifier?

 

[cmb]

If your power supply is truly a DC signal of tolerable voltage to the capacitor, it should make no difference. There are some deterioriation effects on caps, but dependent on type and over 1,000's hrs operation.

What will kill the cap is if the supply is not a steady DC and you get a voltage ripple (esp if there is a high current load that may cause an excessive ripple). Any capacitor's spec should match, and comfortably exceed, the RMS current it will see. Those that do tolerate high RMS currents can usually do so for long periods of time, no worries, but if your situation is a regulated supply that is no longer regulating well, thus may have damaged or failed components, then, yeah, don't even operate it, let alone leave it on!

 

[esprits300]

it is a tube amplifier. i don't have many details about it, its very old and i got it off ebay. i believe it is diode bridge rectified, althought I am not 100% positive.it is powered via 120v wall outlet. I am not sure if your familiar with the ssy1 nd:yag laser but I am powering the Pulse forming network with this power supply. its a 35uF @950v capacitor that hits a flashlamp on the laser ~100ns. I am testing the voltage right before the laser head and i see it get to about 700v. i was starting to think my issue was the power supply, sadly I am nearly convinced now. I am not sure if this is correct or not but it charges almost instantly. i get an overload on my multimeter (only goes to 900v i think) and then it goes down to 700v. all is stable though, only when i go above that do i get a pop. i am looking at replacing the power supply now, it appears its either incorrect or operating poorly from what you guys are saying. is that a fair statement? thank you.

 

[jim hardy]

if it's an electrolytic capacitor that has sat on a shelf for a LONG time it is possible the plates have degraded. Perhaps you bought it at a surplus outlet?

The insulator in aluminum electrolytic caps is a very thin layer of aluminum oxide that's put on the foil plate by chemical process during manufacture. If it degrades from non-use then leakage current goes up and when it arcs through you'll get that popping. That degradation can happen when they sit on the shelf for years unused.

sometimes you can re-form the insulating layer by slowly, like over several days, raising the voltage up to full rated.

on lower voltage electrolytic caps they suggest placing a large resistor in series and applying voltage, the resistor limits current to a level low enough to re-form the insulating layer almost like electroplating.
But at 950 volts you're above the voltage rating of common resistors, which is just a few hundred volts apiece . As Yungman pointed out they'll arc across.

So - you MIGHT try placing about five 1 meg resistors in series and connect your cap to supply through them. If you can get a measurement of the leakage current that flows and watch it for a few days, you'll be able to calculate its insulation resistance and observe whether you are helping it any. I'd place meter in low side wire so it's not sitting there at 900 volts daring you to touch it...

you said " i checked the voltage by using my multimeter at the capacitor..."

does your meter go that high? Did you actually read supply on that 750 volt setting? Perhaps as was suggested the supply is not regulating. Might open it up and check those high-ohm resistors in feedback circuit...

 

[cmb]

But at 950 volts you're above the voltage rating of common resistors, which is just a few hundred volts apiece . As Yungman pointed out they'll arc across.

I use regular 3W carbon resistors well above their rated voltage values. They tend to have ~20mm bodies and, frankly, I have never had one go below 10kV. Of course, I don't use them where it is 'safety cricitcal' or likely to damage kit if they were to fail, just in case they do, but so far, no failures or flash-overs observed. (I've burnt a few out, but that's easy if you've got kV on them and accidentally feed in a few extra kV on top!)

 

[jim hardy]

@cmb -

you are more experienced than i at that.

i ruined a nice little DMM on 1500 volt scale attempting to check the doubler in a microwave oven. ... i expected to find either no voltage or overrange indication but got smoke...

had it been manufactured with more robust resistors per your experience it woulda survived.

 

[cmb]

i ruined a nice little DMM on 1500 volt scale attempting to check the doubler in a microwave oven. ... i expected to find either no voltage or overrange indication but got smoke...

hah! That's funny! Yeah, I buy in a half-dozen of those $2 ebay meters from China every couple of months. Occupational hazard, burning out cheap meters.

I use two 100Mohm 15kV resistors, one in each plug of the meter. As the meter is 1 MOhm, so the reading is 1/200th of the voltage across the ends of those resistors. It's a very reliable means to measure up to a few kHz and 20kV.

The thing I forget to do is turn the bloomin' things on before cranking up the volts. If it is powered off, then the voltmeter's 1Mohm internal resistance is not in the circuit, meaning many kV ends up on either plug of the DMM and, pfff.. One day I'll remember every time, but still forget after burning out several that way!

Also, for some reason that has eluded me, DMMs fail real easy when measuring current at kV. Even after every effort made to isolate the DMM from where it is sitting, put it at the 'earth' end of the circuit, and never more than a few mA passing through, for some reason they work for a while, then go down like flies! Seem to need a moving coil ammeter for that application, but I don't really understand why.

 

[es1]

it is a tube amplifier. <snip> i get an overload on my multimeter (only goes to 900v i think) and then it goes down to 700v. all is stable though, only when i go above that do i get a pop. i am looking at replacing the power supply now, it appears its either incorrect or operating poorly from what you guys are saying. is that a fair statement? thank you.

I was thinking this might be it. Tube amp power supplies often have a huge internal L to make a filter at audio frequencies with a reasonable C. As it is tuned for audio I don't think this supply will have enough bandwidth for your application and I think you're seeing the voltage ring up due to the fast rising edge of the current load step when the laser is turned off.

As a quick patch it might work to put a slow (~10KHz) RC filter between the supply and the load. Assuming the load can tolerate the IR voltage drop on the RC this will slow down the edge seen by the supply and should significantly reduce any ringing.

However the high output voltage might be bad for the amp itself and it could already be damaged too.

 

[esprits300]

actually, yes the capacitor is several years old, maybe decades, its old military surplus that have been shelved for years. so what I am gathering is stick to what doesn't blow it and slowly rasie the voltage over a fwe days? leave it on non-stop? that kinda scares me leaving it unattended that long. for now i am satisfied with the results it has, i just wanted to know why, and assuming you guys are correct (which I am sure you are!) this explains it! thank you so much...

 

[cmb]

actually, yes the capacitor is several years old, maybe decades, its old military surplus that have been shelved for years.

I'd not waste my time. Just get a couple 450V and put them in series for 900V. This is a standard voltage (for mains rectifiers) and too cheap to bother risk other kit on old surplus.

I have a box of 250 56uF 450V I bought in bulk some time ago for around $40. Better still for your application, get some low ESR types then you know you've got everything covered (http://www.ebay.com/itm/140623657644 - no connection with seller).

 

[jim hardy]

If you series up two lower voltage caps put a few megohm resistor across each one. That forces them to share the voltage equally and assures they'll both discharge when power goes off.

If you try to re-form the cap you have, ...

""so what I am gathering is stick to what doesn't blow it and slowly rasie the voltage over a fwe days?""

that's one approach. Other (and safer for unattended condition) is the resistor to limit fault currrent.

Lessee - if you can find 1 meg 1/2 watt restsitos,

p = e^2/r

1/2 = e^2 / 1,000,000 - so to make 1/2 watt e = √500,000 = 707 volts

but the resistors are probably rated 400 volts across the body (they'll take more but one should never dip into one's safety margin)

three of them in series would limit current to harmless value 900v/3meg = 300μa and would not exceed their power OR voltage rating. four would be better.

http://www.vcomp.co.uk/tech_tips/reform_caps/reform_caps.htm

be REAL careful !
a cap that big charged to high voltage could be lethal if it discharged throough your chest. Using one hand while keeping other behind your back , discharge it through a resistor of maybe 10k ohms before handling it.
DONT LET YOURSELF GET HURT BY INATTENTION to this detail. Any tiny fingers in your household?