Capacitors: Surely, I'm the only one baffled by them.

Status
Not open for further replies.

Rich D.

Active Member
...that's supposed to be sarcastic. Theoretical capacitors are simple. Real-world capacitors are just nuts. I'm posting today about Electrolytics.

Recently, I've been worried that my capacitor meter was going out of calibration. Values seemed to read much too high. Smaller values measured as expected, but the larger values read 2X up to 5X or more too high. Two different meters show the same reading issues. I soon discovered the difference turned out to be whether they were electrolytic or not. Ceramic, Mylar, Poly caps all measured as expected, and accurately reflect the values from my capacitor decade box. "Some day I'll get this meter fixed" I thought.

The other day I tested a few big caps before using them in a circuit, only to discover that they measured nearly perfectly. So I started measuring a lot of the various capacitors in my collection. My collection consists of brand new parts for more critical and high-end circuits, and also a rather large collection of older "harvested" parts from old TV's, stereos, microwaves, VCRs, etc. By and large, all the new stock parts measured good, and the old ones bad.

At first I was getting worried that all of my older parts were faulty - either damaged by overheating when removing the solder or damaged in-circuit that caused the device it was harvested from to be discarded in the first place, or just plain too old. I assumed of course that they were very leaky, causing the excessively high readings. Only a few very old ones measure good. 90% bad caps? Can't be. High-ohm measurements from the DMM showed no significant leakage - typically measured in the MegΩ range.

So I took a couple of 33μF caps that were known to be much older and measured maybe 100 or 200μF (the measurement drifts down slowly from say 150...120 depending on how long I observe), and soaked them for a while (10-20 min) at 15 volts DC (they were rated 16V). The power supply didn't show any leakage current (<10mA), and they didn't get warm at all. After that and a discharge via shorting they still measured too high. When I removed the 15-volt power, they seemed to hold the charge for minutes. With no DMM attached, they sat for a good hour and held perhaps 90% of their charge. For all practical purposes, these capacitors are good.

So barring the typical guesses of leaky, damaged, high ESR or high inductance-type excuses, or me mixing up the polarity (I didn't), does anybody know what is going on with these older capacitors? I'm pretty sure they are mostly good caps, or are they? What could cause them all to measure too high? If they indeed are good, what will it take to automatically measure them properly? And if bad, exactly what is wrong with them? Could it be only because I have cheaper hobby-grade meters? It just doesn't add up. Baffled.
 
Look at the tolerance on electrolytics: +150%, -20% of marked value is not uncommon.
 
No, these are way outside the + tolerance, and besides, I've never measured one that was on the low side of the tolerance.
 
Take the large caps and use the RC decay time to determine their approximate capacitance as an alternate check.
Connect a 1.1MΩ resistor across the capacitor.
Charge the capacitor to a known voltage.
Remove the voltage and measure the voltage decay with a digital voltmeter (with a typical 10MΩ input resistance).
Observe how long it takes the voltage to discharge to 0.37 of the initial voltage (one time-constant).
That time will be the approximate capacitance in microfarads.
 

What method does your capacitor meter use to measure the capacitance? I think an ESR meter might be a good investment for you.

Ratch
 
Ratchit...I don't know! I have a fairly recent model similar to this one:
**broken link removed**
but they don't make my specific model anymore. The specs are very similar though.
I have an older B&K DMM, and an even older Fluke that also measures caps, again I don't know what their method is. I would assume that it uses the volt-meter in a bridge configuration but that's only a guess. I dont' even know the test frequency used.

I do believe an ESR meter is in my future. I've been looking at various models but haven't made the plunge yet. I think what's holding me back other then my general frugal-ness is whether I want to get an inductance meter with it. Thinking of spending anywhere between 50 ... 200 US$. Any suggestions for what to look into would be welcome, but I'm not convinced that a pricier meter would read much different for basic uF readings.

There is some physical property causing erroneous readings or perhaps a very recent technological breakthrough in capacitor technology. I have seen no evidence to support either hypothesis.
 
Rich D. said:
.....................
I have an older B&K DMM, and an even older Fluke that also measures caps, again I don't know what their method is. .......................
Click to expand...
Do those all measure about the same value for the caps in question?
 
Rich D. said:
No, these are way outside the + tolerance, and besides, I've never measured one that was on the low side of the tolerance.
Click to expand...
It depends how the meter is measuring, what frequency test signal. If it's too high, the electrolytics may be resonant and give false readings.
 
Most capacitance measurement techniques return similar values for low to medium sized, non-electrolytic capacitors. The only method that gives the best result for all cap types is an impedance bridge. As you are learning, electrolytic caps are weird, and nothing gets past the weirdness to the true capacitance value like a low amplitude sinewave.
Rich D. said:
So barring the typical guesses of leaky, damaged, high ESR or high inductance-type excuses,
Click to expand...
Those aren't excuses, junior, those are the consequences of the material degradation that occurs with time in, well, everything. A thin metal layer deposited on a thin sheet of plastic - pretty stable stuff. The glop that is inside an aluminum electrolytic - not so much. Just because you don't like, or understand, an answer doesn't make it an excuse.

ak
 
Last edited:

The "proper" way to measure capacitance is with sine wave excitation. DMMs that measure capacitance don't do this.

Also, the measured capacitance of electrolytics varies with the measurement frequency. Do you know what frequency your Tenma is using to make the measurement?

If you're in the market for a new meter, this is highly regarded nowadays. It's low cost but fairly high performance: https://www.ebay.com/itm/DER-EE-DE-...502152?hash=item3ac0f6c9c8:g:p~cAAOSwHPlWeWrX

You can also find it available with test leads; search for DE-5000 on eBay.

The DE-5000 can measure capacitors and inductors at 5 frequencies: 100, 120, 1000, 10000 and 100000 Hz. You will find that electrolytic capacitors will measure substantially different capacitances at different frequencies. They can even measure negative capacitance if the measurement frequency is above the self resonant frequency of the capacitor.
 
Thanks Electrician, looks good. I hope it ships with a printed English manual. I did find one on line if it doesn't. Decent price, and I like the tweezer thingie for SMT-sized measurements.

crutschow said:
Do those all measure about the same value for the caps in question?
Click to expand...
Yes they measure the same independent of the meter making the measurement. (Allowing for the over-value reading drifting down over time, they are in the same ball park.)
 
Under low constant current type C measurements in older parts, errors can arise from the a couple effects that resemble the memory effects in NiCad batteries. WHich is like a very high ESR cap with a much larger value, which can slowly restore the voltage of an abruptly shorted cap..

Imagine a very high ESR C1 with a long time constant ( the hidden capacitance from long term absorption...) when a charged at V1 then connected to a low ESR cap C2 at V2=0 ( the shorted real cap) . We know from total conservation of charge Q=V(C1+C2) , that if it restores to 50% V1 that the lossy absorption capacitance is equal. ( C1=C2) and if it restores to 90%V1 it must be very oxidized with a large dielectric absorption factor. This absorption capacitance is also responsible for a low voltage breakdown and thus old caps must be rejuvenated with a low current , large value resistor to prevent puncture damage inside.

I believe this is what may cause high erroneous results in old oxidized wet e-caps.


Rubycon explains as this:

It is said that generation of Absorption Current is related
to the change in polarization of dielectric with the
passage of time and response time of space charge
polarization would affect it.

It is also said Voltage Recurrence Phenomena, such
that voltage arises between terminals of capacitor even
after discharge, is related with the delay in response
time of above space charge polarization
 
Thanks Tony, a lot to think about. This suggests to me an experiment:

I charge a suspected "old" 10uF cap to say...10Vdc. Next I connect it to a new, Lo-Z cap of 10uF for a second and allow the charge to transfer. Assuming that the new Lo-Z cap has no significant errors, the voltage I read on the new cap after removing the older cap should indicate apx 5 volt for theoretically perfect capacitors. 5 volt assumes that 50% of the charge transferred from the older to the newer cap if they are indeed the same value. Anything less would indicate that amount of available charge from the old cap that didn't get out, so in effect the loss of voltage represents the fraction of the capacitance that is "old" or that which is the "Hi ESR" part of the older cap model. Of course the value would actually be less because of the losses in the "new" capacitor being less than perfect. (I can quantify that by using two "perfect" caps to see what's lost).

If my math is right (again ignoring for the moment imperfections in caps): If I were to read 2.5 volts on the new cap, that would indicate that entirely 1/2 of the charge was 'stuck' in the old cap's dielectric. I can then assume that the old cap effectively is now a 5uF cap in performance, because all though it may appear to take the equivalent charge of a 10uF cap, it will only return the equivalent of a 5uF cap. I'm thinking this experiment would allow me to see tonight (before I get an ESR meter) the effect of dielectric losses & ESR. Even better I could see which ones have this effect and correlate it to my wacky readings.

I believe you are saying the design of my cheapo hobby-grade cap meters causes older caps with a dielectric-absorption-caused high ESR to read too high. Have you or anybody actually seen this effect and can confirm it?
 

That's correct.

It is the main reason capacitance measurements for e-caps changes with test frequency or impedance ratio and why RC tests don't always match LC resonant tests.
Your experiment should work, keeping in mind Q=CV the stored initial charge with C & V changing.

The energy in the stored charge is still greater than the discharge loss.
 
Last edited:

Rich, do yo have an oscilloscope and signal generator with square wave output?
 
The Electrician said:
The model you've linked to (Figure 9.16) does not show this: "C + (1/2 + 1/4 +1/8 +1/16... )* C = 2C"

It shows this: (1/2 + 1/4 +1/8 +1/16... )* C =C
Click to expand...
Yes I noticed that too late, and it's an approximation and not quite a binary successive parallel lump adding to C and can have different ratios depending on the construction , aspect ratio etc, of the cap and multi-layer material properties....

The ESR in each is not the same for each value either, where older parts may have higher ESR on the inner layers that were hotter.

All the author said was "it's complicated"

If you imagine the 1C having a large ESR then when driving high currents, it does not contribute much current, but when floating, can restore voltage to the cap after a temporary short. ALso when tested at low currents with a pulse LCR tester or low frequency, may show higher values of total C when it does contribute current.

This non-linear behavior also can be adapted to this model to the wide range of Ah capacity in batteries depending on 1/20C to 1C to 20C load rates but the factors are not exact for every device as shown in the formula.
 
Last edited:
Status
Not open for further replies.

Similar threads

P
  • Question
One 12v 30w motor controller by 2 x 5k pots
Replies
17
Views
2K
Pelock
P
G
  • Question
Need for these capacitors and resistors
Replies
12
Views
3K
ChrisP58
J
  • Question
Remote Switch Devices Turn On By Themselves & Energy Savings
2
Replies
28
Views
4K
Nigel Goodwin
  • Question
What frequency does Victor vc97 multimeter test capacitors?
Replies
5
Views
722
danadak
F
  • Question
Train chargers only for laptops?
Replies
2
Views
1K
ZipZapOuch
Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…