Capacitor distortion

[BR52]

…of which you still did not answer the main question. All you have shown is that bad electrolytics do exist, otherwise your sweeping statement “I would never use electrolytics…” is nothing but FUD. And thats before going into things like threshold of audibility.

Hey, what's going on here ? See here a some guy's fighting for what? Leadership? Who is the Best.....
Please let us communicate in a more civil way.

 

[BR52]

For me, it would be really interesting to see how Polyester, Polypropylene and so on work for audio in the signal chain.
Unfortunately, I haven't a qualified measurement equipment. For this reason, my question in #17. Hopefully Pavel is interested and we can motivate him to do some measurements as well.
Bernd

 

[solderdude]

That's why I suggested maybe @pma (who has the tools including the nice pre-amp) could test for the series and parallel electrolytic caps which is a common way to use high capacitance electrolytics as coupling caps when there is not much physical room.
It may just be a matter of replacing the caps in his test thingy and run the test.
John's suggestion to use similar values makes a more level playing field.
Maybe include certain high value ceramics and maybe even film ?

 

[pma]

That's why I suggested maybe @pma (who has the tools including the nice pre-amp) could test for the series and parallel electrolytic caps which is a common way to use high capacitance electrolytics as coupling caps when there is not much physical room.
It may just be a matter of replacing the caps in his test thingy and run the test.
John's suggestion to use similar values makes a more level playing field.
Maybe include certain high value ceramics and maybe even film ?

I did that, years ago, but I had worse DAC/ADC.

Bipolar electrolytes were of course much better than a single one, with heavier load.
There was no difference between MKT and MKP.
High K (correction) ceramics was worst of all, with high distortion even if lightly loaded.

I will try to repeat some of the measurements.

---

 

[fpitas]

I did that, years ago, but I had worse DAC/ADC.

Bipolar electrolytes were of course much better than a single one, with heavier load.
There was no difference between MKT and MKP.
High Q ceramics was worst of all, with high distortion even if lightly loaded.

I will try to repeat some of the measurements.

We avoid high-K ceramic for anything that needs to be stable, like PLL reference integration, for similar reasons. Fine as bypass. NPO or COG is fine, but the values are small.

 

[solderdude]

I did that, years ago, but I had worse DAC/ADC.

Bipolar electrolytes were of course much better than a single one, with heavier load.
There was no difference between MKT and MKP.
High Q ceramics was worst of all, with high distortion even if lightly loaded.

I will try to repeat some of the measurements.

Interesting could be MLCC in the 1-10uF range X7R quality as those are often used as these values do not exist in C0G/NP0 and in MKP/T/S they usually take up more board space and cost more. You may have something in that range floating around somewhere.

 

[BR52]

With high q ceramics, I could see a microphone effect as well.

 

[pma]

I made a mistake, not high Q, but high K. Like X7R and similar.

 

[fpitas]

With high q ceramics, I could see a microphone effect as well.

Oh yes. Sometimes very bad.

 

[pma]

OK, probably the last measurement.

100n X7R capacitor loaded with 100 kohm impedance

1. Frequency response

At 50Hz, the roll-off with the capacitor is only -0.6dB.

2. THD vs. frequency at 1.7V ac input voltage

I hope this is more than clear! Improperly chosen capacitor has devastating effect to system distortion. (Though at 1kHz it may still look good)

 

[solderdude]

Thanks. Caps like this are often found in (cheap) SMD audio stuff and not just used for decoupling.

They work fine for decoupling though, what they are intended for.

matches this well:

With high q ceramics, I could see a microphone effect as well.

Cass 2 ceramic caps can also emit sound (higher frequency) as they are piezo electric.

Physically larger sized and higher voltage rated X7R are performing somewhat better.

 

[pma]

Thanks. Caps like this are often found in (cheap) SMD audio stuff and not just used for decoupling.

Yes, unfortunately. I have one "semi-pro" soundcard that is very reliable, well built, but the distortion starts to rise below 300 Hz in a similar way I have shown here above.

 

[solderdude]

Here's an example that is (most likely) caused by ceramic input caps being used. Has great performance at 1kHz SINAD was 120.
Shows the added value of multitone measurements..

Using film caps or even 2 anti-series electrolytics would have been much better...

 

[pma]

Here's an example that is (most likely) caused by ceramic input caps being used. Has great performance at 1kHz SINAD was 120.
Shows the added value of multitone measurements..

Yes and with full scale sine sweep it would be even worse, because individual components of multitone signal have much lower amplitude than a sine sweep at a given frequency. That's why I like multitone as a 1st indicator, but then run specific measurements according to indication of nonlinearity.

 

[pma]

In anti-parallel 2x 47uF = almost 100uF and in anti-series 2 x 220uF is slightly higher than 100uF.

@pma might be in for these type of experiments. I am too lazy for that.

Bipolar 2 x 47uF/35V in series vs. 22uF MKT film capacitor

I have finally decided to make this test. Same DAC/ADC as in #1 and other my tests in this thread, Audio Buffer in the loop. Audio Buffer has output impedance of 50 ohm, then the capacitor under test, then load resistor 50 ohm.

Test fixture

Again the THD vs. frequency test was done. AC voltage behind the Audio Buffer 50 ohm output resistor is 0.9Vrms approx. when loaded with 50 ohm.

We can see that even bipolar electrolytic capacitor inserts LF distortion (red), when loaded. Although the ac rms input voltage is low compared to DC voltage rating of the electrolytic capacitor. One can see that film capacitors like MKP or MKT are better choice for filters. For coupling capacitors, bipolar electrolytic cap may be fine if it works to high load impedance. Not fine when working into low impedance like 1 - 2 kohm we may now often see in "SOTA" ADCs and preamps. This is no "FUD", just technical facts, all supported by measurements.

50Hz distortion

I thing I will create a web page summarizing these latest measurements.

I was also curious if I would see the LF distortion effect of the bipolar cap in multitone test. The answer is NO, I could not see it. The reason is much lower level of individual sine components in the multitone than is the level of sine frequency sweep used in my test.

@pma might be in for these type of experiments. I am too lazy for that.

As of myself, I am too lazy to participate in so many threads as you can do .

 

[solderdude]

Thanks for the measurement. Really appreciated.

I would assume that the same measurement but with the 100k input would be much better as the -3dB point shifts to the left.
0,003% at 30Hz is something I would not worry about myself though.

 

[pma]

Thanks for the measurement. Really appreciated.

I would assume that the same measurement but with the 100k input would be much better as the -3dB point shifts to the left.
0,003% at 30Hz is something I would not worry about myself though.

I agree, I do not worry as well, just showing the technical difference. In this forum you can see debates on DACs with better distortion than -120dB, so why not to show -90dB effect of a mere capacitor, degrading the parameters considerably?

 

[BR52]

I agree, I do not worry as well, just showing the technical difference. In this forum you can see debates on DACs with better distortion than -120dB, so why not to show -90dB effect of a mere capacitor, degrading the parameters considerably?

Thank you so much for the time you spend in this discussion! It's fun to see the engagement of some participants ( @solderdude,@restorer-john,@syn08,..........)
here.

 

[theREALdotnet]

Not specifically about audio applications, but I found this video by Alan Wolke from some years ago very good. It may be of interest to anyone but the most hardcore gurus

 

[Scytales]

Pma, I suggest you clearly state which color corresponds to which device under test on each and every of your graphs.

Douglas Self has also reported some interesting distortion measurements on coupling electrolytic capacitors in Small Signal Audio Design and Audio Power Amplifier Design Handbook. The conclusions he has drawn from his tests are : use the highest capacity you can, but beware of time constants in your circuit.