How to test low voltage capacitors?

[Uncle Bulgaria]

This has been exercising my mind lately, as I've been trying to work out how to tell whether the tantalum capacitors I've been taking out of my Uher are the faulty ones or not.

With my DMM I can measure capacitance (how accurately I know not) and resistance. It's obvious if a capacitor is shorted, but should a 'good' one otherwise measure OL? Is this what one uses a 'universal bridge' for?

I don't want to put my 15V AVO resistance range on a 3V tantalum capacitor for obvious reasons.

I have read about ESR meters, but aren't they just for aluminium electrolytics?

I'd like to get better at diagnosing and being efficient with my repairs, so don't want to replace components willy nilly until something works.

The tantalums have confused me as the consensus on here appears to be that they break on sight, while around the Web I read reports that NASA considers them the most reliable of components.

What do you gurus consider when testing a small capacitor with limited test equipment?

[Radio Wrangler]

ESR meters work just fine on tantalum capacitors. THey aren't tied to aluminium.

There are several rather different sorts of tantalum electrolytic capacitors.

'Wet Slug' types (sounds lovely!) in hermetic metal cases are high reliability MIL-spec components and are reliable UNLESS subjected to excess ripple current or heating. Some of these are in the £80 bracket nowadays.

Bead tantalums are going dodgy by now, and SMT ones are related. The capacitor makers responded to accelerating prices of tantalum dust by minimising the amount per capacitor, to the detriment of reliability.

State of play today:

If you can afford them, they'll be the sort you'll regret using.

Electrolytic capacitors show their problems either by DC leakage, or by excessive ESR.

My AVO EA113 will measure to megohms, but only uses +/-3v supplies (4xAA) so low voltage leakage is safe. A Peak ESR meter looks after that aspect.

David

[paulsherwin]

It's actually very difficult to test capacitors of all types definitively as they have so many failure modes. You can test the capacitance easily enough, but even that can be misleading.

[Station X]

I use a variable voltage power supply to check for leakage current at the specified voltage.

As far as tants are concerned, in my experience they either go short circuit or blow themselves apart.

[julie_m]

If they are tantalum, they probably are faulty .....

Historically, tantalum capacitors were chosen for special applications which depended on having a lower leakage, lower ESR and smaller physical size than ordinary aluminium electrolytic capacitors. But modern electrolytic capacitors can comfortably out-perform vintage tantalum capacitors in most applications.

[ms660]

If you disconnect them and test they might test as ok, problem is to put them back in circuit you generally need significant heat, that heat can tip anything marginal the wrong way, then you end up with a component that "tests as ok so it can't possibly be that" meanwhile a merry dance can ensue...for the cost of capacitors just replace, standard electrolytic will do if it will physically fit.

Lawrence.

[Bazz4CQJ]

My own (limited) experience of tants is that they have a nasty habit of shorting, and before you know it, they've caused a few other components to cook and may even have pushed the PCB to its limit. In older test gear, they can damage components which could be very hard or impossible to find replacements for, so I've been replacing failed tants with conventional electrolytics, and have never had a problem in doing that.

B

[dave cox]

In my experience, 'standard' low value (tens of u's) / low voltage electrolytics are no match on ESR to a dipped tant. I only have a few 'low ESR' electrolytics (hundreds of u's) but I didn't compare them to tants.

dc

[nutteronthebus]

I use All-in-1 LCR Component Tester Transistor Diode Capacitance ESR Meter Inductance. from eBay. Only £8.35 and have found it quite accurate on caps, gives value and ESR have to discharge caps or may break meter also tests PNP/NPN transistors .

Dave

[crackle]

I have had so many tants explode I never use them now.
Mike

[Craig Sawyers]

The only failure of bead tants is when they are used in power supply decoupling applications. For reasons best known to the manufacturers back in the day (late 70's to mid 80's) was a 6V part for a 5V line and 16V parts for a 15V line. That is a guarantee for catastrophic failure somewhere down the line. Which means that they go dead short, and if there is no limiting mechanism can burn clean through a board. They should have a voltage rating twice (or more) the power supply voltage.

Now I used to design gear for two companies I worked for back then - and I followed the Union Carbide recommendations as above. And those bits of kit (if they are still around) will be catastrophically failing just about now.

In other applications, where they are not under perpetual voltage stress and absorbing power supply transients, they are just fine.

Craig

[Diabolical Artificer]

I find testing a cap with a voltage source whilst monitoring current and voltage is a good way to test caps. If the cap is good voltage and current with have a linear relationship, EG 10v 1uA, if not then the cap is likely to be bad, EG 10v 69uA or unsettled current meter reading.

After you've tested a few this way you get a feel for when somethings wrong by how the current meter behaves, how long the cap takes to charge etc.

Andy.

[kalee20]

I have designed pieces of test equipment, using tantalum bead capacitors, in use since 1994, some of which have clocked up over 100,000 hours.

I don't know of any failures yet. But my policy was to use 25V rated parts for decoupling the 15V lines (to op-amps) and I used 33Ω series resistors (so 15V-rail - resistor - tant-capacitor-to-ground - op-amp). If a capacitor went short, the resistor would burn out, and the op-amp lose its power.

I generally test low-voltage capacitors with a LCR bridge, but if they are polarised capacitors, I make sure that the DC bias is switched 'on' and I measure it with a DMM, just to be sure.

[Maarten]

Originally Posted by Craig Sawyers

The only failure of bead tants is when they are used in power supply decoupling applications. For reasons best known to the manufacturers back in the day (late 70's to mid 80's) was a 6V part for a 5V line and 16V parts for a 15V line. That is a guarantee for catastrophic failure somewhere down the line. Which means that they go dead short, and if there is no limiting mechanism can burn clean through a board. They should have a voltage rating twice (or more) the power supply voltage.

Now I used to design gear for two companies I worked for back then - and I followed the Union Carbide recommendations as above. And those bits of kit (if they are still around) will be catastrophically failing just about now.

In other applications, where they are not under perpetual voltage stress and absorbing power supply transients, they are just fine.

Craig

Except they sometimes aren't. You say 16V on a 15V line, but troublesome examples can be found on 12V lines too. Dry tantalums probably come in more than one quality grade, so you may have specified better parts than used in consumer electronics and/or used even larger margins. They can and will fail even with the larger margin. Used as coupling capacitors they can go rumbly or noisy and they can still go short when used on power rails. Possibly not more often than a normal electrolytic of the same age goes open, but to me they are parts that can show more than average wear just like those electrolytics. Next time when I get in a Revox or a Grundig, I'll check the margins they used, though.

[Uncle Bulgaria]

Thank you for all the interesting comments. So it looks like an ESR meter is a useful adjunct for all capacitors in the µF range, whether it's one of the nice EVB ones or the sub-tenner eBay option mentioned above.

As I'm living in a building site with the same 2 feet square of desk being used for work and all hobbies, not having an extra piece of bench equipment to put under the bed would be handy!

Attached is part of my tantalum problem. These are presumably the 'bead' type rather than the metal cans mentioned earlier, so are not the leakage-free, never-dry-out, infinitely reliable component I had happily imagined. It all appears to work at the moment, but what to do if it doesn't?

Do I buy higher-voltage-rated modern tantalums (the big suppliers seem to offer them at several pounds a go, which makes this option alarming), or are film or aluminium types worth going for (if they fit)? The particular 1979 circuit diagram for this unit that prompted the question specifies six different capacitor types with different symbols, so whether this is still valid today is a question I am not qualified to answer.

[Craig Sawyers]

Originally Posted by Maarten

Except they sometimes aren't. You say 16V on a 15V line, but troublesome examples can be found on 12V lines too. Dry tantalums probably come in more than one quality grade, so you may have specified better parts than used in consumer electronics and/or used even larger margins.

Hey - Tektronix did precisely the same thing in the same era. That is why Union Carbide and other premium manufacturers supplied voltage ratings of 6V and 16V. Tek even used 50V tant beads on 50V lines with expected consequences. They in all likelihood made 13V rating caps for 12V lines too.

So it was not me capriciously designing (prototype defence) equipment using those voltages - that was what the datasheets at the time said, and why manufacturers provided those voltage ratings. It was "best practice" back in the day.

Craig

[Craig Sawyers]

Originally Posted by Uncle Bulgaria

Do I buy higher-voltage-rated modern tantalums (the big suppliers seem to offer them at several pounds a go, which makes this option alarming), or are film or aluminium types worth going for (if they fit)? The particular 1979 circuit diagram for this unit that prompted the question specifies six different capacitor types with different symbols, so whether this is still valid today is a question I am not qualified to answer.

In the last post I mentioned Tektronix. In one of the longest continuously manufactured products (the 7A26 dual amplifier plug in) they started with tant beads for many years. Then after repeated failures, in later plug ins they substituted aluminium electrolytics.

Craig

[G6Tanuki]

"Wet" Tantalums - just because they don't seem that leaky - aren't always good.

An example being those metal-cased milspec parts in the synth of the Clansman PRC320: when they age they become 'a bit' leaky - enough to inject a few tens/hundreds of microvolts of noise into the loop - which stays in-lock but the frequency randomly wibbles by a few hundred Hertz - enough to make SSB operation impossible.

They're 130V-rated parts in a circuit where voltages up-to-125V occur. New NATO-stock-Numbered ones cost silly money - I replace them with 160V-rated modern electrolytics of reputable manufacture/source, and spend the money saved on wine.

[Uncle Bulgaria]

That's very interesting, Craig. From my reading I had had the impression that a tantalum was a superior product to an aluminium electrolytic. What's the opinion of the aluminium polymer electrolytics that are offered now?

G6Tanuki - I can imagine it's a combination of all these leakages that's caused my signal to go funny in this board! Replacing that single tantalum (which I'd disregarded in my igorance as it 'tested' all right and was supposedly bomb-proof) brought the signal up as it was presumably leaking current.

I've made an order to replace the tantalums in the circuit with various Al-els. There's one interesting part in the board where the capacitor specified is +-10% - not an option available at Farnell for any capacitor. There were some large film motor capacitors at +10%-5%, but nothing else below 20%.

Most are higher voltage than spec'd, if the sizes fit, as there's no noticeable difference in price at the quantities I need.

In my valve amps the coupling capacitors are usually recommended to be various film types, but the ones in this circuit are tantalum and quite high values, so they're to be electrolytic too.

[Craig Sawyers]

The usual argument for using tant beads was low leakage and low ESR. But modern electrolytics (despite what the spec sheets say) are usually as good or better leakage in practice than tant beads, and have better ESR.

There is a point of view that electrolytics should have three to five time the capacitance of the tant bead they replace, but I've yet to see a substantive argument to support that.

Craig