A forum search for 'capacitors' found just over 30 thousand threads, with many currently active. You will find lots of recommendations for different types and brands, and some recommendations to not worry about such things as the correct value is what really matters. My view is that yet another thread on this is not really necessary, partly because after a few pages you will not know whose advice to follow anyway.
I'll throw an early twig on the fire 
As DF96 says the most important thing is the value and that is going to depend on the design (for both the input coupling cap and the feedback cap).
The Axon Polypropylene caps I would say (provided you have the space) would be a good option for the input caps. They are very reasonably priced, and NP used then on the output of the B1
For the feedback decoupling cap I would say a good quality suitably sized bipolar electrolitic would be the way to go. I can't remember where I read it (might have been Doug Self) but someone, somewhere had done tests that showed that a bipolar in the feedback loop reduced distortion compared to a similarly sized polar electro.
Due to the lack of any evidence backing up the above take it with the grain of salt it deserves
Tony
As DF96 says the most important thing is the value and that is going to depend on the design (for both the input coupling cap and the feedback cap). The Axon Polypropylene caps I would say (provided you have the space) would be a good option for the input caps. They are very reasonably priced, and NP used then on the output of the B1
For the feedback decoupling cap I would say a good quality suitably sized bipolar electrolitic would be the way to go. I can't remember where I read it (might have been Doug Self) but someone, somewhere had done tests that showed that a bipolar in the feedback loop reduced distortion compared to a similarly sized polar electro.
Due to the lack of any evidence backing up the above take it with the grain of salt it deserves
Tony
Your question is too generic and wide, plus you provide no circuit details, but given the typical values used, and painting with a very wide brush, in general any good "plastic" capacitor will be fine for input, any *good* electrolytic will be the regular choice for the NFB one.
Notice the emphasis on "good" on the Electrolytic, simply because it would be the weaker link but on the other side "plastics" there are too bulky, heavy and expensive.
Might be justified anyway if difference in audible quality merited it, but fact is that designers already consider electrolytics will be used so in general they specify values 10X or larger than what´s strictly needed, so they are subject to very low "stress" (very low voltage drop across terminals even at lowest frequency, very low current through them) so they do not have much chance of "showing uglyness" in any case.
This is a practical point of view, of course you can be dragged into endless "mystical/subjective" discussions, on which there is no clear winner.
You might end up way more confused after reading a few dozen posts than if reading none at all
Notice the emphasis on "good" on the Electrolytic, simply because it would be the weaker link but on the other side "plastics" there are too bulky, heavy and expensive.
Might be justified anyway if difference in audible quality merited it, but fact is that designers already consider electrolytics will be used so in general they specify values 10X or larger than what´s strictly needed, so they are subject to very low "stress" (very low voltage drop across terminals even at lowest frequency, very low current through them) so they do not have much chance of "showing uglyness" in any case.
This is a practical point of view, of course you can be dragged into endless "mystical/subjective" discussions, on which there is no clear winner.
You might end up way more confused after reading a few dozen posts than if reading none at all
By "input capacitance" do you mean the value of the coupling cpacitor at the input, or the capacitive element of the input impedance seen by some source feeding the input? Assuming the latter, this is unrelated to the input coupling capacitor. It is not easy to measure, although it can be done with fairly simple equipment. Basically you need to see how the HF rolloff frequency changes with source resistance.
In many cases it may be easier to estimate it by looking at the circuit diagram and reverse engineering the first stage and any RF filter. What has this to do with the topic of this thread?
In many cases it may be easier to estimate it by looking at the circuit diagram and reverse engineering the first stage and any RF filter. What has this to do with the topic of this thread?
lol, my amp is the only stuff i never opened and tweaked as I know the people whom designed it are valuable... and it's also cost me second hand some money !
Damn, you answered the op too : "you have a bad amp, guy"... lol
https://en.wikipedia.org/wiki/Capacitance : I will try to read it tomorow... I risk to stay bloked on the first sentence....
Damn, you answered the op too : "you have a bad amp, guy"... lol
https://en.wikipedia.org/wiki/Capacitance : I will try to read it tomorow... I risk to stay bloked on the first sentence....
Last edited:
I've been trying to find the evidence to back up the above, and have only found the following from Douglas Self
from here Distortion In Power Amplifiers
The problem is I now am unsure whether I mis-interpreted it.
I'm unsure whether he is saying that the use of bipolars is impractical because of the physical size (being roughly twice as large for the same value as a non-polar) or whether he is saying you need a bigger value bipolar to do the same job as a non-polar without increasing distortion.
I originally interpreted it as being that the non-polars shifted the frequency down to 1/10th the frequency of a bipolar so if a non-polar was showing increased distortion at 20Hz and lower, a bipolar of the same value would shift that down to 2Hz and lower.
Tony.
Unfortunately the term "conventional capacitor" used in the highlighted sentence is ambiguous. I think he means that a bipolar electrolytic will give at x Hz the same distortion as a polar electrolytic of the same value will give at 10x Hz.
He seems to be saying that bipolars are bigger, for a given value, and so may be impractical. Yet their performance is sufficiently better that a smaller value could be used and so end up the same size, so it is not clear to me why he says that.
Capacitor physical size matters here because large size means high stray capacitance and we do not want to inject a noise/hum/interference current into the feedback aplication point. Hence the apparently simple solution of using a film cap may not be possible, as a film cap of 10% of the electrolytic value may still be much bigger than a bipolar.
He seems to be saying that bipolars are bigger, for a given value, and so may be impractical. Yet their performance is sufficiently better that a smaller value could be used and so end up the same size, so it is not clear to me why he says that.
Capacitor physical size matters here because large size means high stray capacitance and we do not want to inject a noise/hum/interference current into the feedback aplication point. Hence the apparently simple solution of using a film cap may not be possible, as a film cap of 10% of the electrolytic value may still be much bigger than a bipolar.
^ Yes That pretty much nails why I got confused. If the nonpolar is better by a factor of 10, but only twice as big physically for a similar value in uF then surely the best performance would be to get the biggest value non-polar that will fit without causing problems, as probably it should be around 1/2 the value of the biggest polar electro that can be used, but around 5 times better performance...
Assuming of course my rule of thumb that a bipolar is roughly twice the size of a similarly valued polar cap holds true.
I guess without doing actual distortion tests comparing though, there is not too much point in discussing.
Tony.
Assuming of course my rule of thumb that a bipolar is roughly twice the size of a similarly valued polar cap holds true.
I guess without doing actual distortion tests comparing though, there is not too much point in discussing.
Tony.
I'll throw an early twig on the fire
The Axon Polypropylene caps I would say (provided you have the space) would be a good option for the input caps. They are very reasonably priced, and NP used then on the output of the B1
For the feedback decoupling cap I would say a good quality suitably sized bipolar electrolitic would be the way to go. I can't remember where I read it (might have been Doug Self) but someone, somewhere had done tests that showed that a bipolar in the feedback loop reduced distortion compared to a similarly sized polar electro.
Due to the lack of any evidence backing up the above take it with the grain of salt it deserves
Tony
Thanks Tony:
I seem to recall tests I did where the feedback cap did seem to audibly affect sound quality. Polar caps are among the worst in the signal chain. Overspeccing the voltage seems to alleviate the issue to some extent as does over speccing the capacitance, but still the sound can sound congested. Its harder to spot differences in the nfb loop, much easier to see the difference in speaker networks. I remember removing bipolar lytics from a KEF crossover and replacing them with polyprops and the speaker resolution increased many fold. The caps were rated for 63v and my tests were less than 29vAC RMS. Still the difference was very obvious even for my tin ears...
Polypropylene seems to sound good at a reasonable cost. Tantalums also sound colored and have high distortion. Interestingly they were recommended in many mods/ kits in the 80's.
I hear Paper in oil sound very nice, but they degrade with age... so they are out.
This leaves us with polypropylene and Polyethylene types as good options.
Coming to reasonably priced good caps... WIMA seems to be the flavor of the day for most folks... but with a dizzying array of models... I wanted to know what you guys preferred.
MKS, FKP, FKP2 etc....
Last edited:
That is because the cap at the bottom of the NFB loop is essentially a coupling cap so provided it is is big enough it has little AC voltage so can generate little distortion - and signal levels are low there anyway. Passive crossover networks are audio frequency filters run at significant AC voltage so cap properties show up, although they are still swamped by driver properties.K-amps said:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.