0015 Capacitor leakage Trap

Pete explains about capacitor leakage and why the reading from an LCR meter is not enough, on its own, to test capacitors during a repair

[@t1d100]

Good info. As you noted, the leakage needs to be tested at the cap's rated voltage. There are lots of DIY circuits that develop the higher voltages needed to test supply voltage smoothing caps, but much care is needed to use them. Example: 400V SMPS caps. They are dangerous, in uneducated hands, IMO. Interestingly, I have never seen anyone include an explosion cover, to house the DUT cap. This would seem to me to be a basic safety necessity.

[@MakeOrRepair]

Yes, you can find many faulty caps using lower voltages, but how confident are you in a 400v cap tested at 30v? I was thinking if I ever made one that I would include auto discharge at end of cycle, displaying results only when full cycle completed, and a cover with interlock that disconnects and switches to a simple discharge when opened. It is still on my endless list of projects!

[@t1d100]

@@MakeOrRepair I am not sure that I was understood... I am saying that there are tons of DIY high voltage cap test circuits on the web and that they are dangerous. I have seen 600V. Yes, a discharge circuit is a great idea.

[@MakeOrRepair]

@@t1d100 sorry, I didn't mean to confuse. I agree that the DIY high voltage testers I've seen are dangerous. I was just thinking that the "explosion cover" you mentioned could in fact be a cover that would prevent touching during testing and disconnect the dut when opened

[@t1d100]

@@MakeOrRepair Ahh... I see. Yes, that is a much needed double-duty. Thanks.

[@fredflintstone8048]

Another consideration is that caps, especially in equipment where power hasn't been applied in awhile will reform. We can see this happening while testing because the current will continue to drop, and sometimes slowly as the cap reforms. It's good to wait several minutes, or until the current through the cap stops dropping before condemning the cap if the leakage current is above spec.

[@MakeOrRepair]

Good point for electrolytic caps in older equipment.

[@MakeOrRepair]

I hope to put together a basic tester soon. I might need to think about the safety if I decide to allow voltages upto 1KV

[@MakeOrRepair]

Ideally you test the capacitor at its maximum rated voltage but this can be both dangerous and not attainable for many people. There is nothing to stop you testing high voltage capacitors at lower voltages, but some faults might not show up

[@kaszaniarz]

good sign that capactor is leaky, that shows higher capacitance on meter that it should (especially if it is old cap), for example 30uF for 22uF

[@MakeOrRepair]

Absolutely! on a proper LCR meter the lower DC resistance makes the equation calculate the Capacitance as higher. I have seen some fairly extreme examples of that like 2200uF reporting as over 6000uF (in a power supply that blew the fuse). I have mentioned it in a couple of my later videos and have been pondering doing a video showing how the phase shift is used to calculate the capcitance in an LCR meter and how low DC resistance affects it - would that be worthwhile do you think.

[@RCook-iy4xk]

I see other video's adding a resistor in the test circuit. Which is correct?

[@MakeOrRepair]

There does need to be a resistor somewhere particularly as voltage goes up otherwise you get some pretty nasty surge currents (the resistor is in the multimater is high enough for most voltages/capacitance that I work with). Mine is just a quick and dirty method but it works pretty well particularly with tantalum and electrolytic caps, I use it mainly with tantalums in older gear because of the very high replacement costs. But there are a few approaches, this was just the most direct way I could think of to demostrate the problem. I'm using a good digital multimeter, so it already has the resistor built in, the meter is really measuring the voltage across the internal resistor, and the lower the meter range (I think I used mA in the video, but with a 3mA scale that means only the left digit is mA, the other 3/4 are uA, in the uA range the resistance is much higher, quite a few k ohms, and the meter resolves nA. I always recommend electronics enthusiasts to get a meter with the LOWEST current ranges they can. Essentially in my brief introduction to the problem I charge the capacitor through a resistor (in the meter) and the voltage across the resistor should drop to Zero when it is fully charged because DC current can then only flow if it is leaking. The eevblog uCurrent gold current meter adapter would make a pretty good low current detector (it uses series sense resistors and amplifies the result to drive a volt meter and you could always use bigger resistors to add greater sensitivity). I know a couple of diyers have done some higher voltage meters, (some safe some not) and they certainly use resistors because both the voltage and current need controlling, often they also use a sensitive coil meter that needs resistors for limiting. The key thing if looking at these is 1. make sure there is a means to discharge the capacitor (and test it without touching) and make sure the switches are in the low voltage half of the circuit, I've seem people some using 30V switches with 340V DC supplies. 3. I don't think I would use one unless it is isolated and current limited. I seem to recall also seeing an approach where a capacitor is charged to a known voltage, then left connected to a really high impedence op amp to see how fast it self-discharges through internal parallel resistance. I guess it is also possible to create a Wheatstone bridge with 3 resistors and the capacitor (using very big value resistor opposite the capacitor) and that would enable you to measure the internal resistance using potentially any DC voltage, again once the capacitor is charged. Of course in any equipment of some age then some capacitors (many electrolytics, old-can tantalums, paper) should pretty much be replaced as a matter of course (smaller caps cost little). There is always some cost/vs. value to think of but in general with the bigger, very expensive electrolytics in a power supply I often look first to see if they are a relaible make, check phystical condition and look at position relative to heat, then look at power supply output ont he scope, any cap that looks to be buldging, leaking fluids, right next to heatsinks etc and any cheap brands are replaced. I also replace any that are close in DC value, e.g. in a 12V supply I would tend to put 20 or 25V caps in to replace any 15/16V that are present. If there is still 100/120Hz ripple on the DC supply after that then any remaining are changed.

[@SylwerDragon]

Good advise.. i didn't think about this before. But you are right. will check it that as soon as i can :)

[@MakeOrRepair]

Glad to help. I only test the capacitors I think are likely candidates - old electrolytics, canned tantalums etc that otherwise test fine but are going to be expensive to replace. For paper and waxed caps I just replace them.

[@SylwerDragon]

@@MakeOrRepair tested them yesterday..they were on the limit .but they are pretty expensive so i will use them..but to be honest tantalum caps..usually those big one hermetically sealed last for ages before they go bananas ..so they are usually fine.but you made good point..after all ebay ius full genuine..almost new ..never used..just stocked...20 years old tantalum caps... :) haha

[@MakeOrRepair]

@@SylwerDragon Good capacitors seem very expensive at the moment. Some of the better makes seem to last forever, then I get a device that seems to be full of hopeless cases (often the smaller values) In one device I repaired it was full of them with 50% really bad, but looking at the schematic, except for a few places there seemed to be no justification (perhaps size) and I replaced most with higher voltage low esr electrolytics. TOo be honest, many parts of circuits are not bady affected but when I was repairing an HP power meter the leakage of the ones in the range switch was the critical factor so I hand selected replacement tantalums. I wonder if some of the bad examples relate to how close they cut thigs for the rated votage vs. the rails they are connected to. In some cases I've even made little carriers for modern alternatives but where the "authenticitiy" needed to be retained I've purchased a few ITT make from ebay (being a bit selective about suppliers) and they tested ok, At a push even recycled some from other boards. It all depends on the need for authenticity and the quality/value of what I'm repairing.

[@SylwerDragon]

@@MakeOrRepair yeah you are right..but there is reason why they used in some cases tantalums. Usually it is better parameters ..after all. these caps are also used in space industry and are on probes that lasted decades and will last more if power will permit.. If designed correctly there isn't usually any need for them to go bad...for example i was fixing 1kv voltage standard 7.5 digit..and i wasn't even able find good replacement for those super low leakage caps they used..i was able find few in usa..same brand but new. And rest i went for tantalum. Where i was thinking they needed to be for stability anyway..hermetically sealed tantalums will last for decades if they are not overheated but better if too much voltage isn't used..leakage current is always problem but not so big..they are awesome. Awesome expensive ;..haha 100uF /60V could cost 100 Euro/usd like nothing..so yeah not cheap ones but in all voltage standard where there is some oven you will find only tantalums..normal capacitos would last year with luck...3 with super good quality.no more ..tantalums will last 30 years so yeah big difference..but ofc nothing last forever :)

[@mikebond6328]

Cool. Thanks!

[@MakeOrRepair]

No problem, happy you liked it.

[@siavasharash5448]

I hav make my own capacitor leakage tester many years ago and it works with 6 volts and max 2000 uA and it working nice to me . Recently i order a bonch of 47 uf 16 volt tantal capacitors (chinese) to replace instade of some electrolethic's in my analog tektronix oscilloscope. Before i use them in my stauff i test them for leakage and the result shocked me. All of them have leakage more than 100 uA in 6volt I dont know all of them is faulty or this leakage is for aplaying just 6 volt in order to test them. What do you think???

[@MakeOrRepair]

I think that the leakage should be in the ballpark of 10nA/µFV for tantalum , so no more than about 10x47x16 = less than 7.5uA at 16Vor about 2.5uA at 6V. So they sound pretty bad to me. Of course don't know how your tester is doing the test and if there is something that could be catching you out - when I get a whole lot of bad test results, I tend you just stick a known good component in the tester as a sanity check!

[@siavasharash5448]

@@MakeOrRepair last night i found some strange fact that line marked on tantal caps is positive😮 i was connect my caps on reverse polarity so its normal that poor cap shows leakage. Thank you anyway

[@MakeOrRepair]

@@siavasharash5448 well at least that explains it! The line on the end of a tant is normally positive in my experience, but I was recently puzzled by some tants that had a line as normal but with what looked like a "-" sign in it, I had to double check and discovered it was a badly printer "+"

[@rectify2003]

Caps are the bane of my life

[@MakeOrRepair]

Always suspect a capacitor - unless its a sprague

[@rectify2003]

@@MakeOrRepair Whats a Sprauge? The capital of the Netherlands? 😀

[@sstorholm]

At this point, I’ve gone with the approach that I replace all electrolytics I can bother with starting with the biggest ones, then if there’s any problems left, I go hunting for tantalums, if I find a faulty one, I replace anything that remotely resembles one. And then I might start looking for less obvious faults, like a blown fuse. 😂

[@rectify2003]

@@sstorholm I’ve never found a faulty Tantalum