Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Tantalum Capacitors in audio applications

Status
Not open for further replies.
Not exactly

Well I'll skip the whole coupling thing as the output of a filter and the input of the power amp are two different things, but my point was that a filter can be used directly connecting speakers to its output, even though they probably are not intended to be used as a speaker crossover.
As for even harmonics, first there are only a few devices thought to pass the odd harmonics, you can read about that on my web page as I've had articles from experts on it for years. So why say that about tantalum capacitors?
Secondly about changing value - there are charts you can google that show that this is certainly not true at all of surface mount tantalums. If you really are stuck in the 1950's or 1060's and can't affor the ten dollars for a lot of 2000 surface mounts on a reel or want to keep a stiff upper lip and pay 50 cents each for small quantities of what are really almost obsolete axial tantalums, go right ahead and be my guest and keep up the guru work.
Where's your ashram, india?
 
Will two tants back-to-back form a 'bipolar' device (large signal AC safe?) If not, will one tant and one aluminum electrolytic do so? Which will create the most even order distortion of these two configurations?
 
Last edited:
I am a Canadian, not an Indian.
I was called an audioguru by a salesman at a job i had.

Odd harmonics are caused by symmetrical compression or clipping.
Even harmonics are caused when something causes one polarity of an AC waveform to be amplified or attenuated more than the other polarity of the AC waveform, LIKE A TANTALUM COUPLING CAPACITOR.

Tantalum capacitors are horrible. They short and blow up. They change their value with applied voltage. They cause bad distortion in audio circuits.

Metalized plastic film capacitors are perfect in audio circuits. They have no distortion.
All high quality speakers use them in passive crossover networks. Cheap speakers use non-polar electrolytics. I don't think any speaker is made with tantalum capacitors in the crossover network.
 
Design science

Sorry but design science does not make such ckaims based on what is used in expensive overpriced products versus ones desiged to wise engineering constraints and cost tradeoffs. You must be saying such nonsense to voice some marketing priciple or mission. Just show us one graph or oscilliscope trace from any industry standatd ASTM methid. That's why thy have to have ASTM _ sOME PEOPLE JUST DON'T GET IT RIGHT. Yes I will cocede that tanralums are less tolerant of poor designs, especially when used in series. You might consider that pricey products have to anticipate that probem and so are trying to design around it when they use metal film capacitors and can affird 9 times the volume they take up cimpared to tantalums because the pricy black box is a clumsy ugly thing anyway.
 
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.
Um what do you mean here? Coupling the output of an amplifier to an 8Ω speaker with a 4.7µF capacitor will sound horrible and tinny. You seem to be talking about amplifier output coupling capacitors and crossover networks all in one sentence.
If I remember tomorrow, I'll try some Mylar, Electrolytic, and Tantalum caps on the distortion analyzer (At the 3db down point on an RC network) and report back. I remember trying some ceramic ones in the past and they definitely added distortion but not as much as I thought they would.
 
You don't need to measure the distortion of capacitors. It has already beedn done and documented:
1) Douglas Self designed an audio power amplifier that has distortion of only 0.0005% at low and mid frequencies. He wrote a tutorial about reducing distortion. He has a graph of the sharp rise in distortion caused by an electrolytic coupling capacitor.

2)AVX tested an "ordinary" tantalum coupling capacitor which has high distortion and tested their best tantalum capacitor that has the same amount of distortion as an electrolytic one. They do not show the zero distortion of a film capacitor.

My metalized poly film capacitors are not large and are not expensive. The EPCOS brand of a 330nF/5%/63V one costs less than 15 cents US at Newark and is about 6mm wide (5mm leads spacing), 5mm high and 2mm thick.
 

Attachments

  • capacitor distortion1.PNG
    capacitor distortion1.PNG
    46.7 KB · Views: 1,011
  • epcos film capacitor.PNG
    epcos film capacitor.PNG
    16.1 KB · Views: 581
  • capacitor distortion2.PNG
    capacitor distortion2.PNG
    29.7 KB · Views: 970
You don't need to measure the distortion of capacitors. It has already been done and documented
Well, I did it anyway, just for fun:
I used a HP8903B audio analyzer and connected a single pole passive RC highpass filter on the output of the audio generator running at 5Vrms @ 160Hz. R was 600Ω, C was 1µF and the distortion analyzer was connected across the resistor. The results are below:

Direct connection with 600Ω load resistor only (No capacitor) = 0.002%


Distortion with no DC bias on the capacitor under test:

Metal film = 0.002%
Electrolytic = 0.016%
Dip Tantalum = 0.425%
Ancient axial Tantalum = 0.028%
SMD Ceramic = 0.420%


Distortion with 9V DC bias on the capacitor under test:

Metal film = 0.0035%
Electrolytic = 0.029%
Tantalum Dip = 0.190%
Ancient axial Tantalum = 0.022%
SMD Ceramic = 0.609%


The only result that puzzled me was the Ancient axial Tantalum until I googled it and saw that it was still listed on the NASA site. :D
 
Last edited:
NASA use tantalum capacitors because they don't freeze like electrolytic capacitors do.
 
Very interesting

Thanks. That data is definitely helpful. But it only confirms that putting filtered DC through capacitors, espeically tantalums, is not a good idea even though maybe 1 or 2 percent of the amplifiers might do that.
The reason if I am understanding where this is going, that tantalums distort more readily if DC biased is that they have electrons crowding one electrode as part of the design - so that the magnetic field created reinforces the flow of capacitive current instead of working against it as the eddy current of any other capacitor would. Have you considered the improvement in lower distortion due to lower resistance and improved response time that tantalums offer. Remember the point of my last post was all about tantalums in series. I was trying to avoid a discussion here of coupling capacitors as it is off topic, but as you said a filter like I specified would sound very tinny all by itself. But without it you can't design a two pass only filter that covers those frequencies that you are calling tinny. Well enough about that as it gets into different types of filters - lets just assume we mean constant current circuits here and very simple and basic textbook stuff. Thats not now what I'm into as a study to improve my passive filters but you never know, you learn something every day. Anyway, if distortion was such a hugh bid deal, whey are not all amplifiers made with MOSFETs. They only have half the distortion of any other type amp. Then there is the whole tube thing that is still popular in spite of my web page rant and rave about that particular shortcoming as well as high end or headroom losses and the limitiing of innovation that they introduce.
I'll keep this short and just say putting 9 volts DC into a tantalum cap and claiming it has higher distortion does not even begin to convince me that electrolytics - even so called audio or np or bipolar ones, are even close to tantalums for superior design and innovation opportunities in audio.
And to speaker guy - yes you can put an electrolytic after a tantalum, but first the electrolytics must always have its neg electrode toward the negative side - you can get away with ignoring that on tantalums must of the time, but in series, I would also orient it as marked, assuming mark is positive. You can't always tell the polarity by using diode mode on a DV meter just so you know. You can do it by picking up the magnetic field as the inner electrode always gets shielding, but unless you've got old vintage stuff the datasheet or tech support should get you straight. It is espeically important to adjust for the time constant by using a tuning resistor on both the capacitors across the + to - side and they must be different values. Unless your trying to do a bit of wave shaping (I may have to kill you but I do this in my new filter design) it may not be worth the trouble or cost unless you already have sunk costs in the components. But sure, you can block AC that way but since 2 in series have less capacitance than the lowest of the pair you may not want to. Any capacitor blocks AC unless we are talking very small values where filtered DC discussions happen or powered circuits where almost anything is possible; unlikely but possible. As far as passive blocking - AC does not get through any audio frequency capacitor. It just caused electrons collected on the plate to jump off or recollect. No current ever flows directly. The leaking is related to oriented domains in the material, as noted above, an intended condition to improve effiency. So, sure the curve for reverse current is steaper than electrolytics that claim to be non-polarized. I could call it non-engineered if I wanted to get a reputation as nasty.
 
Last edited:
I'll keep this short and just say putting 9 volts DC into a tantalum cap and claiming it has higher distortion does not even begin to convince me that electrolytics - even so called audio or np or bipolar ones, are even close to tantalums for superior design and innovation opportunities in audio.
Umm, that's not what the data I posted suggests at all. Adding a 9V DC bias to the tantalum cap reduced the distortion it generated which is to be expected. But in either case, the tantalum had more distortion than the metal film cap.
SSman said:
Then there is the whole tube thing that is still popular in spite of my web page rant and rave about that particular shortcoming as well as high end or headroom losses and the limitiing of innovation that they introduce.
I think most will agree with you that tube amps have more distortion than modern solidstate amps. It's just that some people like tube amp distortion. I'm not one of them. ;)
SSman said:
Any capacitor blocks AC unless we are talking very small values
Surely you mean DC....
 
Thanks, KChrisie,
You confirmed that tantalum capacitors cause distortion in audio circuits.

Passive crossover networks in speaker systems use a capacitor in series with the tweeter as a highpass filter.
Cheap speakers use non-polar electrolytic capacitors and good speakers use film capacitors.

I have seen hundreds of speakers but have never seen tantalum capacitors in them.
 
Well, I did it anyway, just for fun:
I used a HP8903B audio analyzer and connected a single pole passive RC highpass filter on the output of the audio generator running at 5Vrms @ 160Hz. R was 600Ω, C was 1µF and the distortion analyzer was connected across the resistor.
1µF at 160Hz is 1kΩ which is larger than 600Ω, therefore the capacitor will have a higher voltage across it than the resistor and contribute to more distortion than it would at higher frequencies or if it were larger.

The standard test is the 3dB point which would result in less distortion than you have measured.

Oversizing the capacitor would reduce the distortion even further.

Try increasing the capacitor to 10µF and you can expect to see the distortion to be reduced by a factor of 10.

By the way can you even hear 0.016% distortion?

I doubt it, but you certainly can't hear 0.0016% distortion.

Somehow I think it's more cost effective to use electrolytics in most applications.

Suppose your amplifier has an impedance of 10k, therefore you require 796nF so we use a 1µF capacitor.

If we wanted low distortion then we wouldn't use a ceramic capacitor, we'd ideally go for metal film but I think it would be more economical if we used oversized electrolytic capacitor. We could use a much larger 10µF capacitor which would have negligible distortion and would be smaller and cheaper than the 1µF film capacitor.
 
Last edited:
A larger capacitor value might take too long to charge and you would hear severe distortion until it is charged. You also might hear severe distortion for a while after the power is turned off.

I use 330nF film capacitors for all my audio coupling capacitors with an input impedance that allows them to pass the entire audio bandwidth. They cost only 13 cents each American.

Manufacturers are still playing the "lowest distortion" game.
National Semi has their LM4562 dual opamp that has only 0.00003% distortion. It shows that the circuit was designed properly.
 
1µF at 160Hz is 1kΩ which is larger than 600Ω, therefore the capacitor will have a higher voltage across it than the resistor and contribute to more distortion than it would at higher frequencies or if it were larger.
You are correct. I generated 5Vrms output and then adjusted the frequency until 2.5Vrms was measured across the 600Ω resistor. Should have used 3.5Vrms for the -3db power point. :eek: But the test is still valid, even if the numbers are slightly inflated, as it showed that there is a difference in distortion generated by different dielectrics.
Hero999 said:
Oversizing the capacitor would reduce the distortion even further.
Yes, but SSMan was talking about filters and crossover networks. I think. ;)
By the way can you even hear 0.016% distortion?
I can't. But once you use a bunch of crappy caps in a multipole filter I might. I certainly wouldn't use ceramic or tantalum caps in an equalizer.
We could use a much larger 10µF capacitor which would have negligible distortion and would be smaller and cheaper than the 1µF film capacitor.
I won't argue that point. Once the coupling cap is over sized, to say a 1Hz -3db highpass, then distortion becomes negligible in the 20Hz-20Khz audioband.
 
A larger capacitor value might take too long to charge and you would hear severe distortion until it is charged. You also might hear severe distortion for a while after the power is turned off.
True but if the capacitors are oversized by a factor of no more than about 10 for 20Hz, they should charge quickly enough to prevent that from being a problem.

For example the RC constant of 10:mu:F and 10k is just 0.1s so it will charge to 95% in just 0.3s which you'll probably notice but is not long enough to be a big deal.

I use 330nF film capacitors for all my audio coupling capacitors with an input impedance that allows them to pass the entire audio bandwidth. They cost only 13 cents each American.
I don't know about in the US but here in the UK, a 3.3:mu:F electrolytic costs half the price, is well under a quarter of the size and you probably won't be able to hear the difference.

Manufacturers are still playing the "lowest distortion" game.
National Semi has their LM4562 dual opamp that has only 0.00003% distortion. It shows that the circuit was designed properly.
Anything to make more money from silly audiophiles who'll pay twice as much for half the distortion that they can't hear anyway.
 
If you want to hear 20Hz at the same level that it was recorded at (if your speakers go that low) then you want a flat response at 20Hz and a -3dB frequency of 4Hz.
If there are two coupling capacitors then the -3dB frequency becomes -6dB so the -3dB frequency of each section should be 2Hz.

If you increase the value of a coupling capacitor because it is the distorting kind then its time constant becomes 0.4 second which is a noticeable charge-up time.
Two sections each would have a 0.8 second charge-up time
 
If you want to hear 20Hz at the same level that it was recorded at (if your speakers go that low) then you want a flat response at 20Hz and a -3dB frequency of 4Hz.
If there are two coupling capacitors then the -3dB frequency becomes -6dB so the -3dB frequency of each section should be 2Hz.
So using your logic you need 10:mu:F capacitor if you have two 10k stages, those 10:mu:F metal film capacitors are going to be pretty bulky and expensive.:D

If you increase the value of a coupling capacitor because it is the distorting kind then its time constant becomes 0.4 second which is a noticeable charge-up time.
Two sections each would have a 0.8 second charge-up time
Why would you worry about distortion for frequencies below 20Hz?

The capacitors will all charge at the same time, not one after the other.

Seriously, a 10:mu:F capacitor is fine for a 10k input impedance.

The only time metal film cacitors are useful are in crossovers and filters where you don't want to chance the cut-off frequency.
 
I don't use an input impedance as low as 10k ohms. I use 120k so that I can use a very good, small and inexpensive 330nF film input capacitor.

Many products use cheap electrolytic coupling capacitors. They deform and become useless if they don't have a DC voltage across them.
 
I don't use an input impedance as low as 10k ohms. I use 120k so that I can use a very good, small and inexpensive 330nF film input capacitor.
I agree, make the input impedance as high as possible but sometimes you don't have a choice.
 
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.

Yep. For anyone finding this in a Google search, ceramics typically have much worse distortion than electrolytics (as well as being pizeoelectric/microphonic), so this test result seems mistaken.

Vishay Capacitance Change with Applied DC Voltage:

Tantalum capacitors in general ... demonstrate very stable performance over the DC voltage (bias) applied in an application. At the same time, the majority of capacitors utilizing ceramic or polymer dielectrics (monolithic ceramic, disc ceramic, MLCC, polyester, film, etc.) demonstrate significant shift in both directions - sometimes 40 % to 50 % or higher.

Wolfson A.C. Coupling Capacitor Selection:

Capacitance change with d.c. voltage:
  • Ceramic C0G or U2J: Not Significant
  • Ceramic X7R or X5R: Significant – see individual part graph
  • Ceramic Y5V: Very bad – do not use for a.c. coupling
  • Tantalum Electrolytic: Not Significant
Standard aluminium [electrolytic] parts are much lower cost than tantalum, but are usually physically bigger. They offer good stability of capacitance with voltage and temperature and have similar ESR, so are recommended where there is enough space.

Voltage Coefficient of Capacitors:

  • Ceramic: Weakness: Large Voltage Coefficient (X7R, X5R, Y5V)
  • Aluminum Electrolytic: Advantage: Stability: Voltage
  • Tantalum Electrolytic: Advantage: Stability: Voltage & Temp
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top