How to determine the voltage rating of an unknown capacitor?

[(In)Sanity]

Short of reading the marked values on a capacitor does anyone know of a method to identify the rated voltage of a capacitor?   Let me also add short of blowing them up or catching them on fire.   Does some characteristic of the capacitor change close to it's rated voltage,  such as perhaps the leakage current, etc? 

Thanks,

Jeff

[houdini]

You could measure the capacitance and the size and see what caps match up with that and they will also have the same or similar voltage rating.

[smashedProton]

If they are scavenged, probe the circuit.  you can also look at surrounding caps.

else, use a really nice equivilant cap

[(In)Sanity]

I actually have two reasons for wanting to figure out how to do this.   First I have hundreds if not thousands of "scavenged" caps as you mention.  Also I have thousands of brand new SMD caps that I have no idea what their voltage rating is.   So short of blowing them up I'm trying to figure out their safe working voltage.

Thanks,

Jeff

[AndyC_772]

For many ceramic capacitors, the capacitance starts to fall off quite sharply as the dc bias is increased. Perhaps you could devise a circuit which measures capacitance with a gradually increasing dc bias on the cap under test until the capacitance falls by, say, 20%, or you chicken out - whichever happens first!

[(In)Sanity]

For many ceramic capacitors, the capacitance starts to fall off quite sharply as the dc bias is increased. Perhaps you could devise a circuit which measures capacitance with a gradually increasing dc bias on the cap under test until the capacitance falls by, say, 20%, or you chicken out - whichever happens first!

Thanks Andy,  that might be a good idea.  I've already had a couple tantalum's go up like a match from this sort of testing.   Electrolytic's of course normally have their voltage clearly visible.   Most of the tantalum's do as well..but not all.   On another thought I wonder if the dielectric starts to break down which might also be detected via leakage current?   Or perhaps I would have to ask what causes the capacitance to drop at higher voltage  (or the VC as it's known).

Jeff

[AndyC_772]

In a ceramic, the physics is quite simple. As the voltage increases, electrons build up on the surface of the plates, and as the voltage increases, the additional electrons form a thicker and thicker layer because they're repelled by the electrons already there. The effective plate separation therefore increases with voltage, and so the capacitance goes down. In some dielectrics it can have dropped by 80% by the cap's rated voltage.

Y5V dielectric - "just say no".

[(In)Sanity]

In a ceramic, the physics is quite simple. As the voltage increases, electrons build up on the surface of the plates, and as the voltage increases, the additional electrons form a thicker and thicker layer because they're repelled by the electrons already there. The effective plate separation therefore increases with voltage, and so the capacitance goes down. In some dielectrics it can have dropped by 80% by the cap's rated voltage.

Y5V dielectric - "just say no".

Thanks Andy,  happen to know of any easy way to measure capacitance at higher voltages? My B&K LCR meter isn't going to help me on this one.

Jeff

[AndyC_772]

Time to spend a few of those long winter evenings playing with Spice, IMHO. I'd probably start by building an R-C oscillator and an adjustable negative voltage supply; apply the negative voltage to one plate, and measure the resulting oscillator frequency to calculate C.

It could be quite a fun little project, though it could also blow caps to smithereens (watch out for any tantalums!) or damage them in more subtle ways by exceeding their voltage specs.

Personally I'd just buy new caps of known quality and spec, but I'd have to admit, it's a less exciting option.

[PA4TIM]

Connect the cap to a DC source, limmited in current. For instance 100V through a potentiometer with a currnt limmiting resistor from wiper to cap. Connect your LCR meter to the cap via two large HV MKT caps ( like 4.7 uF or so)

The value you measure will be 1 / ( 1/4u7 + 1/4u7 + 1/Cx ) but the actual value is not important. This way you can see the capacitance change. And then you know upto which voltage they are usable. They drop in capacitance before they brek down.

If they do not decrease they will breakdown and You know the voltage. Mount the testfixture in a box as protection for your eyes and kin.

But your oscillator idea is nice too. The frequency tels you the capacitance. You can test voltage and if frequency goes up and you note frequency you lo know the capacitance, with some luck the oscillator stops befor the cap explodes. I ould use a Wunderhorn oscillator, this uses two transistors, tubes, or fets and it oscillates from a MHz upto some GHz. Make a switch and some inductors for a broad range. Add a microprocessor ( and maybe a FV converter and to meaure frequency. The uProcessor can calculate the capacitance and also meaure leakage current and voltage  if you want.
Use a servo to control the potentiometer and you even can sweep voltage.
The output of the potentiometer to the X of a scope, a voltge proportional to C from the microprocessor to the Y and you can ake graphs.

Nice, good idea, i have the sme problem, thausants of unknow caps. Most pF ceramic. This test eill give a clue sbout the type and rathing.

i think I'm gonna do that one day. But a bit different. I am not a uProcessor guy but I can sweep the voltsge analoeg using my fingers to turn the pot and use my HP 135 X,Y writer for the plots. The X input to the wiper, the Y input to a IV converter i once build to meaure fF ( but it can do nF or uF as easy), choosing the right gain and frequency that gives a voltage reading in direct in capacitance. I measured fF that way. But the value is not important, it is about the change invalue.

[plazma]

I have the same problem. I got 3 component drawers for free.

Full of 0805, 0606 and 0402 passives (RLC) and even a lot of semiconductor. Nice and compact drawers. Solves my component archive system

The values are marked but no idea about cap dielectrics, inductor current rating or resistor construction.
I'll probably just check digikey what rating the size usually has.

[(In)Sanity]

Plazma , Nice collection of caps.  AndyC and PA4TIM, love the ideas.  I need to re-read everything a few times so I understand the concepts fully. 

I had an idea as well,   what if using a micro with a simple 8 bit DAC + op-amp you charge a cap to say 1 volt,  discharge it for a known time interval and then measure the voltage.   Then charge to 2 volts,  discharge for the same time interval and measure it's voltage,  3 volts...discharge for the same time interval and measure.  This process would continue until the percentage of change during the pre determined time interval fell to some level.   Now this approach might work better to determine the voltage coefficient then the max working voltage,  I'm not sure on that one. 

The micro could also do a low voltage run to determine the time interval needed for the entire test.   For example it would require a longer test period for larger capacitance.   It would then use this exact same interval for each voltage range.

Hope this makes sense,

Ohh and by the way a tantalum cap will still catch fire regardless of the current limiting resistor.   It makes for lots of excitement.   

Jeff

Edit:  I forgot to mention that the discharge process might need to use a constant current vs constant resistance. 

[PA4TIM]

Leakage ( for electrolytics most) will polute your reading. Possible also dielectric absorption. But for non electrolyts it could work. Why not use a 16 or 24 bit ADC for the micro. They cost allmost nothing ( i heard, i never do digital stuff but I have a 24 bit here for a auto reformer, leakage tester, VC tester, ESR, C meter project, a friend made it for me on a pcb together with guidelines to make the program on an arduino, a long term project but sub sections are finnished allready, the rest is still in my head and parts on paper, the C measurement will be done with a capacitance to voltage converter IC from AD)

[(In)Sanity]

Leakage ( for electrolytics most) will polute your reading. Possible also dielectric absorption. But for non electrolyts it could work. Why not use a 16 or 24 bit ADC for the micro. They cost allmost nothing ( i heard, i never do digital stuff but I have a 24 bit here for a auto reformer, leakage tester, VC tester, ESR, C meter project, a friend made it for me on a pcb together with guidelines to make the program on an arduino, a long term project but sub sections are finnished allready, the rest is still in my head and parts on paper, the C measurement will be done with a capacitance to voltage converter IC from AD)

Luckily electrolytic's tend to be well marked.  No reason not to use a 16 bit or higher DAC,  perhaps just do a rough run and then narrow in the final value.   If the concept seams sound I may pursue it.   Need to finish up my lab power supply project first.   It's coming together well fortunately.

Jeff

[jimmc]

Should mention that not all ceramic capacitors have a significant voltage coefficient.
Those using a Class 1 dielectric (NPO, COG) do not change significantly with applied voltage or temperature .
This dielectric is used for lower value capacitors say less than 1nF give or take a decade.

A quick blast with a hot air gun will show if you have a Class 1 dielectric - if the capacity doesn't change significantly with temperature, it won't change with voltage.

For example see http://www.vishay.com/docs/23140/geninfo.pdf pages 6-8

The only way I can think of testing these is to apply a proving voltage of 3 times the voltage you want to use them at (through a current limiting resistor), if they survive for 60sec then you should be OK.
e.g. If you apply 48v through 47k for 60 sec and no significant current flows then they should be safe at 16v.

Jim

[G7PSK]

If you have a quantity of the same capacitors you could just ramp up the voltage until they blow, then run the remaining ones at half or three quarter voltage of that which blows them up.

[SeanB]

Very simple, you make a test jig to place the capacitor in, with 2 very good capacitors to provide isolation of the cap under test from the test voltage. Generally i have seen a 10 or 15uF 250V unit used here ( a bloody big capacitor and most likely the largest film unit you can buy) in series with the leads to the DUT, and the power supply is connected via a pair of 100k resistors. You measure the capacitance at )v 9 you are measuring 3 caps in series, so it will measure low) then slowly wind up the voltage applied till you either are approaching the isolation capacitor voltages or the voltage across the capacitor drops due to leakage ( measure via a second set of 100k isolating resistors to the voltmeter) so you know how the capacitance changes with voltage and how leaky they are.

The isolating caps unfortunately will have to be good ones, polypropylene foil, ( polystyrene is best but a 10uF one is the size of a large shoe box), polycarbonate or polyester foil. Not electrolytic or ceramic. 100V, 250V or 400v best. higher voltage bigger size. If you cannot get 10uF use 3 4.7uF in parallel instead, will be easier to find.

[saturation]

In theory, the breakdown voltage is characterized :



http://en.wikipedia.org/wiki/Capacitor#Breakdown_voltage

For general dielectric strengths, E_ds you can find them here:

http://en.wikipedia.org/wiki/Dielectric_strength

Then guestimate the plate separation distance, d, in meters. 

This is taken from Paschen's law but actual construction and conditions to the capacitor matters, so in the end destroying a few is your best and most simple test.  Using 1/4 or 1/3 of the V_bd as a derating is also just overall good practice for high reliability systems.

http://en.wikipedia.org/wiki/Paschen%27s_law#Basics

Bolded below are 2 techniques I use.   One can get 'grab bag' bulk capacitors in the 100s from someone, even if marked, but could have been QC rejects.  Its good to test those caps are to some spec, including V_bd. 

 Also, watching them blow is also more fun than measuring capacitance    so you have incentive to characterize you're salvaged caps 

Should mention that not all ceramic capacitors have a significant voltage coefficient.
Those using a Class 1 dielectric (NPO, COG) do not change significantly with applied voltage or temperature .
This dielectric is used for lower value capacitors say less than 1nF give or take a decade.

A quick blast with a hot air gun will show if you have a Class 1 dielectric - if the capacity doesn't change significantly with temperature, it won't change with voltage.

For example see http://www.vishay.com/docs/23140/geninfo.pdf pages 6-8

The only way I can think of testing these is to apply a proving voltage of 3 times the voltage you want to use them at (through a current limiting resistor), if they survive for 60sec then you should be OK.
e.g. If you apply 48v through 47k for 60 sec and no significant current flows then they should be safe at 16v.

Jim

If you have a quantity of the same capacitors you could just ramp up the voltage until they blow, then run the remaining ones at half or three quarter voltage of that which blows them up.

[westfw]

Then guestimate the plate separation distance, d, in meters.

Yeah, right.  How do you propose to do that, for a modern ceramic or electrolytic?

[olsenn]

The simple answer to this one is, you don't! Throw away your unknown caps and buy a new batch that provide you with a rating.

However, electrolytics usually imprint the voltage rating on the cap sleeve itself, and other caps are usually pretty standardized.

[smashedProton]

Blow the ass out of it!   

[(In)Sanity]

Blow the ass out of it!   

I would much rather do that then throw them away.

[Alexei.Polkhanov]

Hmm, when I am not happy about ESR of an electrolytic capacitor I usually try to replace it with one of same capacity but higher voltage rating. Usually it works so that 6.3V cap will have twice ESR of 16V one and so on. It is true at least for manufacturers that I usually order/use (Panasonic). Higher voltage rated ones are usually (not always) bigger.
Reverting this dependency should give a clue about voltage rating knowing the ESR of similar cap of known voltage rating. Also longer taller caps tend to have lower ESR - I am not sure if there is any such dependency for chip capacitors, but those with higher profile (same size i.e. 3.2x1.6mm) tend to have lower ESR as well.

[saturation]

Multilayer ceramics range from 30um to 1 um between plates.  With air as a cheapest dielectric, at 3MV/m then estimated breakdown voltage such caps would be ~ 3V to 90V.  Since dielectrics in reality are rarely air, then most generally purpose caps can be used in 5V applications.  Always best to test it empirically as its is possible to get a ~ < 3V capacitor.

http://my.execpc.com/~endlr/ceramic.html

From the equation, as V_bd ~ d, then physically bigger capacitors, are more likely to have greater V_bd too.  Thus, through hole types are thus likely suitable for 5V applications compared to SMT. 

Older style disc capacitors or electrolytics have larger d, you may be able to see the plate distances if you decapitate one with a dremel then use a micrometer to measure the plate distances. Luckily, larger caps like these still have readable V_bd values printed on them.

Then guestimate the plate separation distance, d, in meters.

Yeah, right.  How do you propose to do that, for a modern ceramic or electrolytic?

[(In)Sanity]

Very simple, you make a test jig to place the capacitor in, with 2 very good capacitors to provide isolation of the cap under test from the test voltage. Generally i have seen a 10 or 15uF 250V unit used here ( a bloody big capacitor and most likely the largest film unit you can buy) in series with the leads to the DUT, and the power supply is connected via a pair of 100k resistors. You measure the capacitance at )v 9 you are measuring 3 caps in series, so it will measure low) then slowly wind up the voltage applied till you either are approaching the isolation capacitor voltages or the voltage across the capacitor drops due to leakage ( measure via a second set of 100k isolating resistors to the voltmeter) so you know how the capacitance changes with voltage and how leaky they are.

The isolating caps unfortunately will have to be good ones, polypropylene foil, ( polystyrene is best but a 10uF one is the size of a large shoe box), polycarbonate or polyester foil. Not electrolytic or ceramic. 100V, 250V or 400v best. higher voltage bigger size. If you cannot get 10uF use 3 4.7uF in parallel instead, will be easier to find.

Doesn't this just equate to indirectly increasing the voltage until the leakage current starts to increase ?   How does the test isolation capacitors keep for example a tantalum from catching fire ?    Also are you saying the isolation capacitors need to be very high quality to reduce leakage current?    Sorry if I'm being a bit dense on this one.

Thanks,

Jeff