Tantalum Capacitors in audio applications

[2camjohn]

I accquired a load of cheap tantalum caps.

I wanted to use them in decoupling the DC from an audio signal.



Some websites are very critical of tantalums in this application but I get the feeling the websites may be a little out of date.

Im just after opinions here?
Is it a good idea? If so what should I watch out for when using tantalum caps in audio decoupling?

 

[audioguru]

Decoupling the DC from an audio signal???

DC is decoupled from a power supply. Then any capacitor type can be used.
Audio has coupling capacitors.

Metalized plastic film capacitors have the lowest distortion.
Tantalum and electrolytic coupling capacitors change value with voltage so the low frequencies will have even-harmonics distortion.
If the signal across a tantalum capacitor causes reversed polarity then the capacitor is a rectifier producing severe distortion.

 

[whiz115]

Tantalum and electrolytic coupling capacitors change value with voltage so the low frequencies will have even-harmonics distortion

i don't understand that...
capacitance in an electrolytic varies according to voltage?!

also.. do you know if tantalums can be used for decoupling where
high ripple current exists? i'm confused about that because i have read
that they can more easily destroyed than electrolytics.

 

[audioguru]

Go to Google and look at Distortion From Capacitors.
People at www.diyaudio.com say that distorting ceramic, electrolytic and tantalum capacitors have no business in an audio chain.

Tantalum capacitors are small so i think their ripple current rating is low.
I have had many tantalum capacitors blow up and disappear so I don't use them anymore.

 

[2camjohn]

Decoupling the DC from an audio signal???

DC is decoupled from a power supply. Then any capacitor type can be used.
Audio has coupling capacitors.

Metalized plastic film capacitors have the lowest distortion.
Tantalum and electrolytic coupling capacitors change value with voltage so the low frequencies will have even-harmonics distortion.
If the signal across a tantalum capacitor causes reversed polarity then the capacitor is a rectifier producing severe distortion.

Hi audioguru, thanks alot for the reply.

Sorry for my bad semantics. I did mean coupling.
(to remove the DC from an audio signal)


I did notice the difference between two circuits I made, (changing 3 ceramic 4.7uf for 1 22uF electrolytic made the sound seem very loose.. arg loose is a bad word.. I hope you get what I mean)

Would that make sense or did I just make the whole thing up?


I found metal film caps (polyester?) but they are such low values, Im using around 20uF right now to decouple audio signals (maximum 2v peak to peak)

Does that sound appropriate?
If so, are there metal film caps available in the size I require? I cant find any.


Cheers
John

 

[audioguru]

I have some small 1uF 5% metalized plastic film capacitors. Larger size (really big) and higher value ones are used in good speakers' crossover networks.

If your audio circuit uses fairly high impedances then small metalized plastic film 0.33uf capacitors can be used as coupling capacitors. If the impedance is 120k ohms then a 0.33 coupling capacitor will have a flat response down to 20Hz.

 

[Hero999]

Surely the amount of distortion depends on the current flowing through the capacitor. A JFET amplifier with a 100nF ceramic input capacitor and an input impedance of 1M probably won't suffer from much distortion.

 

[Speakerguy]

Electrolytics as AC coupling capacitors are fine. Just make sure that the DC bias across them is enough. Using a tantalum as a coupling capacitor, that I have never tried - no real reason to when a cheap aluminum electrolytic does the job just fine @ audio frequencies. Keep the f3 well below 10hz - if you don't you'll wind up introducing distortion into the low frequencies. 10uF-100uF is plenty.

Do NOT ever allow the tantalum to see reverse bias in this application. An aluminum electrolytic would handle it OK,but a tantalum might go poof.

Honestly the best place to use these depending on voltage and size is as power supply decoupling capacitors and in voltage regulator circuits.

 

[2camjohn]

Thanks again for all the information..

I will relegate my tantalums to power supply decoupling etc only.



Prompted by what you guys said, I set up a blind test with some decent headphones and I can definately tell the difference between two circuits I have here. (identical design, but one has bipolar electrolytics and one has multilayer ceramics).

In comparison to the circuit with ceramic, the electrolytic one sounds a bit like its been recorded in a seashell.


I think ive already decided metal film caps are too expensive in the values I need (the price for each component would cost pounds rather than pennies)

But Im really curious if going to metalised film would give a similar step up in quality over the ceramics. If thats even possible.

 

[Hero999]

Could you pleas post the test circuit.

Have you tried oversizing the electrolytic capacitor?

The chances are, if it's just a simple push-pull (on-sided not bridged) class A-B output stage then an electrolytic will do since there's no chance of it becomming reversed biased.

 

[2camjohn]

Could you pleas post the test circuit.

Have you tried oversizing the electrolytic capacitor?

The chances are, if it's just a simple push-pull (on-sided not bridged) class A-B output stage then an electrolytic will do since there's no chance of it becomming reversed biased.

From what people have posted above, it seems like Im oversizing my components already. Im using 47uF where 2uF would suffice.


Also, since I used bipolar capacitors for the electrolytics, reverse biased wont be an issue here? Or am I missing something?


Ill draw up the schematic for the test circuit and post it up as soon as I can.

John

 

[2camjohn]

OK heres the test circuit. (showing just 1 channel)


C1 and C2 are the caps im talking about.

The POT is 50K

 

[audioguru]

Many analog switch ICs have a pretty high distortion when they use only 5V for a supply voltage. A CD4016 is not even spec'd with a supply voltage as low as only 5V.
A CD4066 has much lower distortion than a CD4016 but it is also not spec'd when its supply is as low as only 5V.

Your C1 feeds 33.8k if the input impedance of the amplifier is high. Then a 1.2uF capacitor will have a response that is flat down to 20Hz.

 

[Hero999]

What are you using as an audio source?

I think that there will probably be more sources of distortion in the audio source and in the analogue switch than the electrolytic capacitor.

I think your reason for preferring the ceramic to the electrolytic is purely physiological. Somewhere you heard that electrolytic are worst than ceramics and believed it to be true so it's no surprise your experiment backs up this theory. Constructing an experiment to prove what you think is true probably won't prove it otherwise, especially if the testing is purely subjective as it is in this case.

I would suggest a blind test, try encasing both of the capacitors in film cases or something that will make them both look the same. Test them both randomly, decide which is the best, then reveil which is the electrolytic and which is the ceramic. You might be very surprised.

 

[audioguru]

One article measured the distortion from coupling capacitors. Ceramic capacitors that were taped in a row by the manufacturer had a range of distortion that was 10 times.

Rod Elliot figures that distortion from "bad" capacitors is lower than 100dB below the music level. So low that the distortion cannot be heard.
**broken link removed**

 

[2camjohn]

What are you using as an audio source?

I think that there will probably be more sources of distortion in the audio source and in the analogue switch than the electrolytic capacitor.

The audio source is just my PC sound card right now.
It claims to be some kind of high fidelity one.

I think your reason for preferring the ceramic to the electrolytic is purely physiological. Somewhere you heard that electrolytic are worst than ceramics and believed it to be true so it's no surprise your experiment backs up this theory. Constructing an experiment to prove what you think is true probably won't prove it otherwise, especially if the testing is purely subjective as it is in this case.

I would suggest a blind test, try encasing both of the capacitors in film cases or something that will make them both look the same. Test them both randomly, decide which is the best, then reveil which is the electrolytic and which is the ceramic. You might be very surprised.

Thats what I thought originally, Im aware that audio appreciation is very subjective.
But I did a blind test pretty much as you suggested and on the headphones I can pick the 'good' and 'bad' circuits every time..

I will investigate further, maybe something else is different..

 

[2camjohn]

Many analog switch ICs have a pretty high distortion when they use only 5V for a supply voltage. A CD4016 is not even spec'd with a supply voltage as low as only 5V.
A CD4066 has much lower distortion than a CD4016 but it is also not spec'd when its supply is as low as only 5V.

I was told that, but the presence or absence of my little switching network doesnt make any difference to the sound. (not that I can hear)

I tried to do all the right things for audio (ground plane on both sides, nice fat tracks for the signals etc)


The chips I have (name escapes me) are rated to 5V, in fact in the current configuration they dont like any more than 5 or less than 0.
But hooking them up to +12 and GND wouldnt be a problem.

I will see how the specs of my analog switches compare with the ones you suggested, as you clearly know your onions.

Your C1 feeds 33.8k if the input impedance of the amplifier is high. Then a 1.2uF capacitor will have a response that is flat down to 20Hz.

Yeah the headphone amp is class A and has a high input impedance.
Thanks! Looks like I can go down a bit in the value of my coupling caps, thats really helpful.

One article measured the distortion from coupling capacitors. Ceramic capacitors that were taped in a row by the manufacturer had a range of distortion that was 10 times.

Rod Elliot figures that distortion from "bad" capacitors is lower than 100dB below the music level. So low that the distortion cannot be heard.
**broken link removed**

Cheers Ill read that this evening. 100dB thats alot

 

[Hero999]

The audio source is just my PC sound card right now.
It claims to be some kind of high fidelity one.

Have you looked at the circuit board?

My sound card has electrolytic capacitors on it and I wouldn't be surprised if yours does. I'm reasonably sure that there will be some electrolytic capacitors somewhere in the audio chain, even if it was before the sound was converted to a digital format.

 

[SSman]

Im on it

I'm all over this issue 2camjohn. I have been selling tantalum capacitor passive filters for high end audiophile use (at a discount).
First, don't believe all that crap. Its mostly bad judgement propted up by big business that can't change. Jargon doesn't help. You can use tantalums where you use bipolar electrolytic. Actually a non-polar electrolytic can have higher reverse voltage than forward voltage which is why they started the jargon -nonpolar or bipolar. It is misleading. It only means this condition has been designed out by making the cathode larger instead of tiny so that if the liquid electrolyte does break over (normally it binds its ions tightly) there is no damage. First lets back up. All capacitors are polar. All have reverse currents called leakage. This reverse current must be present to reset the device to begin its next cycle of operation. To decouple the DC - a GREAT IDEA I may ad, you just use a large capacitor because it cannot pass the class A DC bias than is common on power amps and some op amps operate in class A mode. To couple with a capacitor in audio usually means just using a low value like 4.7uf because at somewhere between 3000 and 5000 hertz it will have an 8 ohm capacitive reactance and since that matches the speaker it will optimize power transfer. It also helps cover the high end of the spectrum. and NO YOU ARE NOT PICKING HIGH VALUES for your filter in the 30 to 90 uf range. The very small value use a very large inductor, (hi filter) and are trying to duplicate a Bessel filter. Really bessels are powered devices but the Butterworth have been called 'slow' due to the larger value capacitors.
Its easier to get a fast rise in the capacitive reactance and inductance combined called impedance when you use that method. But keep working on it and be VERY careful about talking to anyone who is not active full time in circuit design of filters and component selection. I have had no problems with tantalums at all. There may be a justified fear of DC, but you seem to already realize how to solve that problem. Block it. One side is enough. And to put to bed some other nonsense that is online all over the place - Tantalums store just as long as electrolytics. No, you can't put to tantalums back to back and make the bipolar. It is just jargon for the way the electrodes sizes are constructed. Yes they have much lower leakage currents. All capacitors act like rectifiers, it just means its a better capacitor, about 2-1/2 time more capacitance per surface area than electrolytic, thats why. Also the low ESR makes the coil Q higher, but that is not good if the Q causes reverberation past the time of the next cycle of the capacitor. Thats where the absurd idea that tantalum capacitors cause distortion when used in audio circuts came from originally I think, and now everybody with a theory is jumping on the bandwagon to claim a new piece of escoterica. Since my stereo splitter boxes get kinda ugly to use huge coils like a speaker crossover, I don't sweat this problem at all but sure, if you need to get more volume out because your pre-amp is a weener and your power amp cant budge out more than say 35 or 40 watts per channel, by all means, use 150 micro henries or more and pay some stupid price. I'll settle for turning up the amplifiers to max and not having it really loud, but just real solid backed sound. Some people should just listen to themselves. They'ld be emailing Kennedy Audio and asking more questions - like that ever happens..

 

[audioguru]

Modern speakers rely on the extremely low output impedance of a modern solid-state amplifier to damp their resonances. Modern amplifiers certainly don't have an 8 ohm output impedance. Many have an output inpedance of only 0.02 ohms for a damping factor of 400.

Tantalum capacitors change their value when their DC voltage changes. So they "modulate" the signal with even harmonics when used as audio coupling capacitors.
Electrolytic capacitors modulate a small amount and plastic film capacitors are perfect.