Capacitor measurement

[panson_hk]

Hi everyone,

I want to measurement some film and electrolytic capacitors by an Agilent 4284 LCD meter. Cap to be measured are general part as well as expensive audio grade. The measured data will show us whether those $$$ caps are really better technically.

Firstly, capacitance and dissipation factor will be measured at certain freq, e.g. 1 kHz. We can use this factor to get the ESR. Please see the attached doc.

Is it an interesting project? Please comment.

[Elvee]

I think it is interesting: I had begun some times ago, but I think the files are not accessible anymore.
Here they are.
Conrad had also made some interesting measurements in this field.
Hopefully, this post will become a reference library.
This too might be of interest:
paralleling film caps with electrolytic caps
Good work!

[panson_hk]

Hi Elvee,

Thank you! Your results are very interesting.

I will secondly do THD measurement.

[Conrad Hoffman]

Getting D is a good start but won't tell you much about audio performance. It will let you take a good guess at unknown film materials. I don't know the frequency range of the 4284, but some curves up to 1MHz would be interesting if you can keep the lead lengths short- even half an inch will interfere with the measurements. Distortion is more useful. You should be able to clearly see how bad a polyester is, compared to almost any other film. You should also be able to see that some of the high end expensive audio caps don't appear to offer anything more than moderately priced Digikey caps!

I should go back and do some more measurements- stuff I've measured.

[DF96]

Interesting measurements. I was puzzled by your remark that 3rd harmonic residual is "counter intuitive" - that is exactly what I would expect from an unbiased cap. You can't get 2nd harmonic unless something tells the cap which end is which - DC bias or dielectric poling for example.

[panson_hk]

Originally Posted by Conrad Hoffman

Getting D is a good start but won't tell you much about audio performance. It will let you take a good guess at unknown film materials. I don't know the frequency range of the 4284, but some curves up to 1MHz would be interesting if you can keep the lead lengths short- even half an inch will interfere with the measurements. Distortion is more useful. You should be able to clearly see how bad a polyester is, compared to almost any other film. You should also be able to see that some of the high end expensive audio caps don't appear to offer anything more than moderately priced Digikey caps!

I should go back and do some more measurements- stuff I've measured.

Hi Conrad,

Thank you for your input. The freq range of 4284 is 20 Hz to 1 MHz.

[Conrad Hoffman]

DF96- well, it was counter-intuitive to me at the time but I try to get a little smarter every day. In a few thousand years I'll be a genius!

[DF96]

I can assure you that being a genius is not as much fun as people think! 8-) The main problem is the fear that I may be getting a little dimmer every day.

[panson_hk]

I would like to present the data obtained today, though this is not following the planned sequence.

Most people parallel a film cap to an AC coupling/DC blocking e-cap. This is said to lower the impedance and to offset the adverse effect of e-cap such that sound quality is improved. Some people even replace the e-cap with a very expensive film cap of similar capacitance.

I measured the THD of a low-pass filter formed by a capacitor and a 1k metal film resistor. Driving signal amplitude is 10 Vrms. The test setup is taken from Self's small signal book.

A Nippon Chemi-Con KZE series 47 uF, 63 V is used in the first test. The non-linear effect of e-cap is shown by the increased THD in low freq. Shall we add film cap to in parallel with the e-cap. A WIMA 1 uF film cap is employed. However, it does not offer any improvement.

In second test, a 30 uF film cap (mostly like a polyester one) alone is tested. It is effectively distortion free. However, it has a very big axial case.

Two 470 uF caps are used in the third test. One is an inexpensive KZJ series and the other is top-end audio grade SILMIC II. Both of them are effectively identical and distortion free. The SILMIC is more than 30 times the cost of the KZJ. Is it worth?

Dissipation factor (at 120 Hz):
47uF, KZE: 0.018
30uF film: 0.00087
470uF, KZJ: 0.031
470uF SIMLIC II: 0.029

[DF96]

Adding a parallel film cap will not affect LF distortion, as you have found. If it has any effect at all, it will be to modify HF impedance (for better or worse).

A capacitor will only add distortion if it has some signal voltage across it. This will depend on cap value, circuit impedance and signal frequency. For 47uF the distortion begins to rise out of the analyser floor at about 70Hz. Other things being equal, I would expect 470uF to do this around 7Hz. The analyser has a small rise around here anyway so it might not be able to test larger electrolytics.

What you have demonstrated is that electrolytics are fine, provided they have little or no signal voltage across them. This means that electrolytic caps should not determine the LF point of an amplifier, unless this is significantly below the LF limit of the input signal. Alternatively, use a normal cap early on to set the LF point.