Adjustable high current AC source to torture a capacitor, does it exist ?

[BravoV]

Probably I don't know what exactly I'm asking, so please be gentle.   

Its all started when I read various capacitor's datasheet, its the type designed and made for switching power supply in smoothing out the output.

As we all know one of the important specification is the "Absolute Maximum Allowable Ripple Current" or "Rated Ripple Current", which usually or most of these cap's type are specified at 100 KHz in Ampere RMS value at certain temperature.

This makes me wonder if there is such unique measurement tool, or even better a dedicated and carefully designed circuit that carries out this kind of specific test to torture the capacitor ?

Ok, to be honest, really wish such circuit exist that can be built DIY style easily. ... LOL 

Maybe it has the features like an adjustable frequency, current and bias voltage and so on if needed, and not too overly ambitious (cmiiw here) say like 50Hz up to 500KHz (is this too high ?), and maybe from few hundreds of mA up to 5 ARMS  and of course the adjustable bias voltage a well. 

Again, I'm not exactly sure what I'm talking about here, its all started by a simple idea of torturing a PS smoothing cap in a controlled environment.

Edit :

As for "WHY" ... these are not complete list ... feel free to add if you have any  :

• As a pack rat, I do love salvaging used or broken electronics, sometimes its just a matter of convenient to re-use the salvaged part if its still "good enough".
• WAIT !! Before you're shout me to go buy new one instead using used part, sometimes in real life its not that easy. Especially at the place where I live  . I can really afford to buy a capacitor that even cost $10 per piece (this is already expensive one), its not justified and feel stupid to buy just a single cap from Digikey/Mouser/Element11 etc that will hammer me with the shipping cost of US $70-80 ! 
• Sometimes when at urgent situation, you just have to use what ever available caps even its used, and a quick test using this tool to see if it survived at that short period, this at least will be good enough as emergency replacement while waiting the real fresh part to arrive.
• Sample and sacrifice one or two caps to see if its up to it's specification at "accelerated" simulation  , not perfect as real situation I know. Its just sometimes even branded and new cap can go wrong. Or better, torture the cap sample at "way above" it's specification to see if it survived at long period while monitoring the cap's temp.
• If equipped with a decent LCR meter, at least we can watch if a tortured cap experiences significant drift at it's parameter, hell, I'm even thinking of measuring the weight using high res gold weighing scale to see if there is a significant evaporation at the liquid electrolyte inside at a slaughtered cap. 
• etc ... and a lot more ....

Ideas, comments or even critiques are welcome.

[Rerouter]

if your not too fussed about maximum voltage, then a OPA547, you have a resistor settable current limit (0-150uA control input, so you can use a current mirror), adjust your gain for the voltage and adjust your source for frequency, (only good for 500mA so parallel them)

bias voltage just means you sum in a voltage with the input,

yes there may be better choices than the opa547 for the current range, but its an amp I'm a fan of

[electron_misfire]

Please post the results of your experiment if you build it, this sounds like a very good idea.

[mzzj]

Again, I'm not exactly sure what I'm talking about here, its all started by a simple idea of torturing a PS smoothing cap in a controlled environment.

Planning to run your test for 1000, 3000 or even 30 000 hours as manufacturers specifications? 

I have done some tests along the lines you are talking about, I was mostly interested how much you can exceed the manufacturer specifications on short-term use.
I used quick-and-dirty  H-bridge controlled with SG3525-variant PWM controller to get variable frequency 50% duty cycle square wave. Half-bridge output was coupled to 10:1 transformer to boost up the current. Big-Ass audio amplifier can be also used on the lower frequencies as a current source for the testing...

2200uF 35V caps rated for 2A ripple or so  can do 25Arms ripple just fine if you bolt them to heatsink  or use large fan to  blow cool air for caps

[BravoV]

if your not too fussed about maximum voltage, then a OPA547, you have a resistor settable current limit (0-150uA control input, so you can use a current mirror), adjust your gain for the voltage and adjust your source for frequency, (only good for 500mA so parallel them)

bias voltage just means you sum in a voltage with the input,

yes there may be better choices than the opa547 for the current range, but its an amp I'm a fan of

Rerouter, I think the current is too low, at least 1 or 2 A(RMS). I'm afraid parallel op-amps doesn't sound too appealing for me. 

Please post the results of your experiment if you build it, this sounds like a very good idea.

Sure, just watch this thread should this grew into real thing.

Planning to run your test for 1000, 3000 or even 30 000 hours as manufacturers specifications? 

C'mon, you knew any sane "hobbyist" will not do that, updated the 1st post on the "WHY" part. Besides its pointless.

I have done some tests along the lines you are talking about, I was mostly interested how much you can exceed the manufacturer specifications on short-term use.

Thats one of the plan, see the 1st post again for the updated "WHY" part.

I used quick-and-dirty  H-bridge controlled with SG3525-variant PWM controller to get variable frequency 50% duty cycle square wave. Half-bridge output was coupled to 10:1 transformer to boost up the current. Big-Ass audio amplifier can be also used on the lower frequencies as a current source for the testing...

As an EEE  (aka hobbyist), appreciate if you could come up with the draft of the circuit ?

2200uF 35V caps rated for 2A ripple or so  can do 25Arms ripple just fine if you bolt them to heatsink  or use large fan to  blow cool air for caps

Good to know, thanks ! And yes, that is part of the plan too, such tool is good for finding an emergency replacement cap if you don't have the right replacement in hand. 

[David Hess]

I have thought about this before.

A resonate mode switcher could drive the capacitor in an LC tank to yield high and controlled ripple current without drawing a lot of power.  The operating frequency would be roughly fixed.

Ripple voltage at high frequencies would be small so a linear high current output stage does not have to be inefficient.

A current driven H-bridge would be fun.  Efficiency could be 90% or higher.  The H-bridge could switch the buck inductor instead of the capacitor allowing one end of the capacitor to be grounded which would be handy for measurement reasons.  There are similar floating inductor buck converters which produce positive and negative supplies with one inductor.

[BravoV]

I have thought about this before.

A resonate mode switcher could drive the capacitor in an LC tank to yield high and controlled ripple current without drawing a lot of power.  The operating frequency would be roughly fixed.

Ripple voltage at high frequencies would be small so a linear high current output stage does not have to be inefficient.

A current driven H-bridge would be fun.  Efficiency could be 90% or higher.  The H-bridge could switch the buck inductor instead of the capacitor allowing one end of the capacitor to be grounded which would be handy for measurement reasons.  There are similar floating inductor buck converters which produce positive and negative supplies with one inductor.

I can only come up with this idea, and I know idea is cheap. 

Really wish if any of you experienced EEs can come up with a real working circuit.   

Of course, please, do not use exotic or hard to find parts, use jelly bean components as many as possible. I believe many electronics enthusiasts will be interested as well if this really happened.

Regarding this project should it become real working circuit, I think it will not be categorized as a novice circuit anymore, so few assumptions must be made, like for example a scope is mandatory, I'm thinking of the preliminary adjustments needed for the 1st time at the "adjustable" current or voltage part, cmiiw.

Again, not very sure, but its safe to assume that this kind of circuit doesn't need to be ultra precise right ? Say current accuracy within 20% should be fine ?  Maybe not for bias voltage as we don't want to bias it way over the rated voltage, again, I'm not very sure here.

I'm all ears.

[robrenz]

You need a EMI Lambda BOS/S or a Kepko BOP, very handy bit of kit!
I show 2 in action in this thread.  I have a spare 50-2 unit that is for sale if you are interested.

You can drive these with a FG to get whatever DC bias and PP voltage you are after. It can be any waveform your FG can output.

[amyk]

2200uF 35V caps rated for 2A ripple or so  can do 25Arms ripple just fine if you bolt them to heatsink  or use large fan to  blow cool air for caps

You know you're not doing regular electronics anymore when you start talking about capacitor cooling...

[mzzj]

2200uF 35V caps rated for 2A ripple or so  can do 25Arms ripple just fine if you bolt them to heatsink  or use large fan to  blow cool air for caps

You know you're not doing regular electronics anymore when you start talking about capacitor cooling...

Had some issues with ripple current:


Yes, those are 500mm2 copper  bus-bars 

[BravoV]

@robrenz,

Thanks for the offer, just droll when every times I saw that, its just I'm afraid the shipping cost will be too unbearable. Also the bandwidth is quite limited for this purpose, I think minimum for this circuit is at least 100 KHz, and also todays switching is already in MHz region. Again, appreciate the offer. 



@mzzj,

Thanks for sharing that, I counted 80 caps, meaning 25 Arms x 80 = 2000 Arms. 

Btw, the heat conduction path are only through the cap's pins ? What about the cap case, does stripping off the cap's thin plastic label jacket will help too ?

[David Hess]

Btw, the heat conduction path are only through the cap's pins ? What about the cap case, does stripping off the cap's thin plastic label jacket will help too ?

Temperature and cooling have a first order effect on operating life.  Physically larger capacitors have more surface area for heat dissipation which is one factor in their ripple current rating but removing the plastic jacket is not going to make significant difference.

[David Hess]

Really wish if any of you experienced EEs can come up with a real working circuit.   

I would still consider capacitor ripple current rating pretty esoteric and the tests take extended amounts of time so a piece of test gear specifically for this is going to be pretty specialized.  It is difficult to see a lot of demand for such a thing.

[David Hess]

You know you're not doing regular electronics anymore when you start talking about capacitor cooling...

I tend to think of it as "overlooked" electronics.

[SeanB]

I get kind of worried looking at that H bridge, and seeing that the cabling for it probably could be better implemented with a hydraulic bending unit rather than just a few cable ties and a clamp. Just what was that for, a 20kW inverter?

[BravoV]

Btw, the heat conduction path are only through the cap's pins ? What about the cap case, does stripping off the cap's thin plastic label jacket will help too ?

Temperature and cooling have a first order effect on operating life.  Physically larger capacitors have more surface area for heat dissipation which is one factor in their ripple current rating but removing the plastic jacket is not going to make significant difference.

I was commenting at the unique case at mzzj's stuff there. 

I guess any means necessary in getting those poor overdriven caps to get cooler really count.

[T3sl4co1l]

I have thought about this before.

A resonate mode switcher could drive the capacitor in an LC tank to yield high and controlled ripple current without drawing a lot of power.  The operating frequency would be roughly fixed.

Ripple voltage at high frequencies would be small so a linear high current output stage does not have to be inefficient.

A current driven H-bridge would be fun.  Efficiency could be 90% or higher.  The H-bridge could switch the buck inductor instead of the capacitor allowing one end of the capacitor to be grounded which would be handy for measurement reasons.  There are similar floating inductor buck converters which produce positive and negative supplies with one inductor.

Meh, shouldn't need the resonance at all: there's not much besides ESR up there (say, 1-100kHz), just drive it from a beastly transformer and inverter.

I did a test of this sort the other day; I was testing an inductor with reasonable Q, so I did do it resonant.  I used an 8:1 transformer (just a core and some hookup wire -- nothing fancy) to drive a series resonant tank with the function generator, having calculated that it would deliver just about enough power, and match reasonably well with an easy ratio.  I ended up doing about 0.25W into the inductor, and measured it as 35uH, with a Q from 50 to 30 (dropping at higher signal levels).  And matching up these measurements with the manufacturer's data (since I know the core type on this particular item).

Tim

[T3sl4co1l]

2200uF 35V caps rated for 2A ripple or so  can do 25Arms ripple just fine if you bolt them to heatsink  or use large fan to  blow cool air for caps

You know you're not doing regular electronics anymore when you start talking about capacitor cooling...

Had some issues with ripple current:
https://dl.dropboxusercontent.com/u/93888484/IMG_3856.jpg

Yes, those are 500mm2 copper  bus-bars 

I don't know what that is, and I want it.

[BravoV]

I don't know what that is, and I want it.

Let me guess, for the chunky metals ? 

[T3sl4co1l]

That, after I blow it up, which is after using it to blow up an awful lot of other things.

[mzzj]

Just what was that for, a 20kW inverter?

~40kW pwm-controller for a pair of 24vdc motors. Not even a inverter, just a upscaled buck-mode smps.
(for a electric go-cart)

T3sl4co1l: and yes, that thing was capable to melt or blow up pretty much anything from 20" wrenches to large dc motors.   
Roughly 3kArms maximum output @ 10 volts or so with 50% duty cycle.

[AleXis6]

guys do you know any news for this topc?
how to evaluate caps current ripple at home?