Why do caps have leakage and ESR??

Can anyone explain why capacitors have any resistance serially and in parallel?

i.e. --> Rp and ESR

Maybe the electrolyte conducts through the insulation???

It seems nowadays there should be no resistance through a capacitor..

Any cap experts out there?

Thanks

Aram

 

ESR comes from the fact that impedance is a complex quantity. There is a samll positive real part that is independent of frequency. You can think of them as losses rather than as a DC current path.

 

Forgive me... Let me clarify ..It's not with my DMM (plus I use a cap analyzer)..

The question is why does it even exist on the data sheet? Why IS there leakage (resistance) in the first place. How or why does this phenomenon exist?

Aram

 

Because a real capacitor is not an ideal component. What separates the plates of a capacitor is not an ideal insulator. Small currents will exist because of this.

 

Just to expand upon this...

There are NO ideal components, except in simulations.

Resistors are mainly resistive, but they do have parasitic inductance and capacitance.

Capacitors are mainly capacitive, but they do have parasitic resistance and inductance.

Inductors are mainly inductive, but they do have parasitic capacitance and resistance.

Even semiconductors have parasitic resistance, capacitance and inductance.

At very low frequencies, the parasitics have very little effect. As frequencies increase, the parasitics can become quite pronounced. For example, multilayer ceramic capacitors might measure 47pF at 10MHz, but 170pF at 500MHz. This can (and does) wreak havoc with RF filter designs.

 

Thank you Gentleman:

The article you recommended Bertus triggered the question in the first place. I had a discussion a few weeks back with a fellow engineer regarding these very expensive "audio grade" capacitors being sold. I failed to see anything different about them except they were high quality, but nothing novel or unique.

I was more interested in the DC leakage. My mind slipped a bit and I forgot that ESR was relative to A/C only (i think).

It's difficult to measure these parameters under stand alone conditions, so a cap has to be in a circuit. I just find it hard to believe that a brand new capacitor could have any Rp at all (my main interest). Of course all insulators have a resistance, but for practical purposes, it should not be existent in when used in any real circuit, if fabricated correctly.

Okay, so a cap gets old after 25 years, and the insulation breaks down. It starts to get "leaky" and violla! DC resistance starts to appear. That I can accept.

But on day one there should be zero resistance... Isn't my thinking correct here?

Repsonses Welcome..

Aram

 

If the dielectric breaks and the plates touch there will be a short or partially shorted (not contacting very well as a short). The plates may don't touch and the capacitance changes. Or because of the crack the capacitor may arc internally and blow.

 

In post #9, you say:

"Of course all insulators have a resistance, but for practical purposes, it should not be existent in when used in any real circuit, if fabricated correctly."

You understand that "...for practical purposes..." does not mean (sic) "no resistance at all", don't you?

Since it's Rp you're talking about, you should be saying no "leakage", or "infinite" resistance.

Then you say:

"But on day one there should be zero resistance... Isn't my thinking correct here? ". You really mean zero leakage, or infinite resistance (Rp). The word "zero" is an absolute; it does not include "...for practical purposes..."; zero means zero.

Consider how a capacitor is constructed. It consists of, in effect, two conducting metal plates of substantial area, separated by a thin dielectric film. This construction tends to exaggerate any leakage that the film may exhibit, so, yes, even a brand new capacitor will have measurable leakage (less than infinite Rp).

 

Your are correct sir. I used the wrong terminology. It is leakage (which i still think is a form of resistance isn't it?)

"Consider how a capacitor is constructed. It consists of, in effect, two conducting metal plates of substantial area, separated by a thin dielectric film. This construction tends to exaggerate any leakage that the film may exhibit, so, yes, even a brand new capacitor will have measurable leakage (less than infinite Rp)"

I guess the above statement deemed correct is the part that bugs me. How is leakage measured or determined? For instance I take a typical 50v .1uf polyester cap, and put DMM across it to measure the resistance. Eventually the meter goes off scale and you really don't know anymore. Okay so I now apply a voltage and measure DC current???

Leakage seems non-existent...For all practical purposes (not academic) isn't it?

Aram

 

You can always place a charge on the capacitor and measure the voltage across it with a meter having a very well-known, very high input resistance. It will form a discharge path, so the voltage should decrease at a predictable rate. Any leakage in the capacitor will increase the rate of discharge, and thus be indirectly measured.

Leakage is very noticeable in electrolytic capacitors

 

Yes, of course, but since it's treated as a parallel resistor, the desired value is infinite, not zero.

To measure the leakage of a good quality cap, you need a picoammeter and a voltage source.

Of course, "for all practical purposes" it's negligible, but it's not zero and it can be measured. There are applications where the leakage needs to be especially low. Perhaps it's an integrator cap in a very long time constant integrator. Then you have to use a cap with teflon dielectric, or something similar.

The Fluke DVMs often have a "conductance" scale. My Fluke 189 can measure the Rp of a cap up to 100 gigohms. I measured a new Panasonic film cap that was in my inventory, and after several minutes the conductance reading went to zero, meaning the cap Rp is more than 100 gigohms.

 

Thank you gentleman:

I learned something new today. After 20 years of this I never realized there was a piece of equipment called a pico ammeter. I should try to find one.

Just for fun I went out into the lab and grabbed a 0.1 63v mono. I used my Agilent 34410A DMM. The best resolution it could do was 0.001 in the uA scale; essentially one nA, but that too was not enough to get any current flow using 36VDC from the PS.

I switched over to a 470 50v elec. the leakage finally is settling down to about 0.5 uA as expected... I guess the electrolyte absorbs through or gets around the dielectric material...

Okay let me make a bold statement:

Probably 95% of all circuits don't care about this level of leakage, thereby for the most part irrelevant. (ouch.. did I go over the edge)

Aram

 

Minimizing ESR is very critical in the design of switch-mode power supplies. The ESR has a degrading effect on the efficiency of the power supply and it also can hasten the destruction of the capacitor due to the power that is wasted in the form of heat internal to the capacitor. To address this concern, most switchmode power supplies place many capacitors in parallel. By doing this, the ESR of all of the capacitors are in parallel and thus the effective ESR is reduced accordingly.

hgmjr

 

Here are a couple of picoammeters that can measure down to 20 femtoamps:

http://www.emelectronics.co.uk/spec/e26.html

http://news.thomasnet.com/fullstory/5636

 

The ESR is meaningless unless a particular frequency is considered. For DC there is leakage to consider not all capacitors are biuld the same. A rate multiplier amplifier intregator with a cap across that leaks is worthless. Some caps can cost $$$ and big in size to quarantee leakage to fanto-amperes. Each cap that you buy has a specific purpose.