Electrolytic Capacitor Measurement

[icbcodc]

Hello John
How's it going?
I have several audio electrolytic capacitors for output ripple control of my AC/DC converter. And I got to know ESR value of electrolytic capacitors is the most important thing for ripple controll in DC filtering circuit.
So I tried to measure the capacitors, and attched the dat file.

and also got the data measured by LCR bridge provided by the factory.

1. Are my measurements correct?
2. How to read the ESR value from REW measurement?

My REW version is 5.20.9

 

[John Mulcahy]

From the open circuit calibration there is a larger than usual gain difference between the channels, about 1 dB, it may help to better match the channel gains if you have control over that. The difference between the channels gets larger at low frequencies, perhaps differences in the coupling capacitors on the channels. The short circuit measurement is a bit noisy, and the capacitor measurements are also noisy above about 200 Hz. As the capacitor impedance is very low the noise produces large variations in apparent phase, so the results are not reliable. I'm not sure why they are so noisy, a similar measurement I made is much cleaner (see below). Perhaps the interface output impedance is higher than ideal? The higher the output impedance, the higher the noise. It can be helpful to use an output with good drive capability, such as a headphone output.

 

[icbcodc]

The difference between the channels gets larger at low frequencies, perhaps differences in the coupling capacitors on the channels. The short circuit measurement is a bit noisy, and the capacitor measurements are also noisy above about 200 Hz. As the capacitor impedance is very low the noise produces large variations in apparent phase, so the results are not reliable. I'm not sure why they are so noisy, a similar measurement I made is much cleaner (see below). Perhaps the interface output impedance is higher than ideal? The higher the output impedance, the higher the noise. It can be helpful to use an output with good drive capability, such as a headphone output.

From the open circuit calibration there is a larger than usual gain difference between the channels, about 1 dB, it may help to better match the channel gains if you have control over that.

I will tried to decrease the difference tomorrow.

The difference between the channels gets larger at low frequencies, perhaps differences in the coupling capacitors on the channels.

This is a bit complex for me,should I open the sound card?

Perhaps the interface output impedance is higher than ideal? The higher the output impedance, the higher the noise. It can be helpful to use an output with good drive capability, such as a headphone output.

My current sound card is Beringer UMC 202HD which was mentioned in this thread: Measuring Frequency Response and Distortion of amplifiers （4x150w RMS and 4x80W RMS）
I can find the output impedance value in datasheet of UMC202HD, I will change the output from Right output to headphone output.

 

[sm52]

My current sound card is Beringer UMC 202HD

if you have the opportunity, try a different sound card.

 

[icbcodc]

if you have the opportunity, try a different sound card.

It's the only one available.

 

[Privatier]

I guess that the output impedance of the UMC 202HD is in the kiloOhm range (I could not find the value in the manual), thus the driving current for your DUT is too small. The signal across the DUT gets too small at higher frequencies, therefore you getting close to the noise floor. A back of the envelope calculation: assuming a 2V source voltage and 1kOhm output resistance, you get 2mA current, producing 2mA*20mOhm=40microVolt signal. I suggest driving such large capacitors through a power amplifier, with output impedance in the one Ohm range. This mimics the application as filter capacitor in a power supply. For these low impedances, Kelvin (4-point) connections are necessary to achieve accurate results.

 

[icbcodc]

From the open circuit calibration there is a larger than usual gain difference between the channels, about 1 dB, it may help to better match the channel gains if you have control over that. The difference between the channels gets larger at low frequencies, perhaps differences in the coupling capacitors on the channels. The short circuit measurement is a bit noisy, and the capacitor measurements are also noisy above about 200 Hz. As the capacitor impedance is very low the noise produces large variations in apparent phase, so the results are not reliable. I'm not sure why they are so noisy, a similar measurement I made is much cleaner (see below). Perhaps the interface output impedance is higher than ideal? The higher the output impedance, the higher the noise. It can be helpful to use an output with good drive capability, such as a headphone output.

View attachment 58374

Hello John,
The difference last time might be caused by unexpected short connection within the TRS terminal of the impedance measurment circuit, and I've fixed it.
Today I tried to measure the capacitor mentioned before,and got the result like this with warning about low signal to noise ratio.

But the measurement of capacitor looks not too bad，can I read the value as: ESR 17.86mΩ Capacitor : 3352μF and ESR 14.7mΩ at 1kHz on the curve?

I attached other settings below:

The other warning message about SN ratio when measuring the capacitor:

 

[icbcodc]

I guess that the output impedance of the UMC 202HD is in the kiloOhm range (I could not find the value in the manual), thus the driving current for your DUT is too small. The signal across the DUT gets too small at higher frequencies, therefore you getting close to the noise floor. A back of the envelope calculation: assuming a 2V source voltage and 1kOhm output resistance, you get 2mA current, producing 2mA*20mOhm=40microVolt signal. I suggest driving such large capacitors through a power amplifier, with output impedance in the one Ohm range. This mimics the application as filter capacitor in a power supply. For these low impedances, Kelvin (4-point) connections are necessary to achieve accurate results.

Thank you, Privatier.

I'm also thinking about optimizing the circuit by adding an amplifier into it,but I don't have many clues about the protection component for the input.

 

[John Mulcahy]

Today I tried to measure the capacitor mentioned before,and got the result like this with warning about low signal to noise ratio.

Measurement looks good. The signal-to-noise ratio warning isn't applicable for impedance measurements, I'll fix that.

 

[icbcodc]

Measurement looks good. The signal-to-noise ratio warning isn't applicable for impedance measurements, I'll fix that.

It's a good news,thank you.