Altaic said:
My opinon about low leakage is that this parameter is most of time totally irrelavant in audio circuits as long as the leakage current is "normal".
One trick though to reduce it: Apply DC voltage at the rated voltage or a little bit higher (you must measure the leakage current so you don't distroy the cap) for a day or a week or longer if you have the time. You get much better results and you get even better if you use the capacitor at half of it's voltage rating. When you apply voltage you build up the oxide at the aluminium foil. The oxide is the isolation. Aluminium (= positive electrode), oxide at the aluminium (= isolation), and "geggamojja" as we say in swedish (=negative electrode along with the aluminium can).
myth of the bad electrolytic
Okay:
Really is there somthing wrong in using a large modern electrolytic as a DC blocking cap? With a low -3dB point the cap's impedence is miniscule in the audio frequency range, compared to the resistive element. I would worry about the cap quality say in a crossover design when the impedence is high right smack in the middle of the audio frequency range, but when your -3dB point is somewhere down at 1 Hz, then it should not matter. Should it?
As for high frequency effects, the inductive effects don't start to become signficant until well past 20kHz (as mentioned by Jens). In my experience not until past 100kHz and even then the impedence change is still very small.
Of course there is the issue of listening tests. But on technical merit, are there really problems with *modern* electrolytics for DC blocking?
That said, in my last DIY amp, I used polypro DC blocking caps.
-willryu
Okay:
Really is there somthing wrong in using a large modern electrolytic as a DC blocking cap? With a low -3dB point the cap's impedence is miniscule in the audio frequency range, compared to the resistive element. I would worry about the cap quality say in a crossover design when the impedence is high right smack in the middle of the audio frequency range, but when your -3dB point is somewhere down at 1 Hz, then it should not matter. Should it?
As for high frequency effects, the inductive effects don't start to become signficant until well past 20kHz (as mentioned by Jens). In my experience not until past 100kHz and even then the impedence change is still very small.
Of course there is the issue of listening tests. But on technical merit, are there really problems with *modern* electrolytics for DC blocking?
That said, in my last DIY amp, I used polypro DC blocking caps.
-willryu
not a myth
The typical resistive part is < 1 k (e.g. 18K over 1 k = gain of 19). Typical ESR of small electrolytics is 3-10 R and has substantial variation with voltage etc. Assume 1 R variation, this will give you approximately 0.1% nonlinearity in the transfer function.
Besides, voltages will be around zero with both signs occuring. A polar cap will be reverse polarized, with nasty effects on impedance. A bipolar cap will have some crossover behavior.
Last not least, there is the issue of dielectric absorption which is suspected to smear impulses.
Eric
The typical resistive part is < 1 k (e.g. 18K over 1 k = gain of 19). Typical ESR of small electrolytics is 3-10 R and has substantial variation with voltage etc. Assume 1 R variation, this will give you approximately 0.1% nonlinearity in the transfer function.
Besides, voltages will be around zero with both signs occuring. A polar cap will be reverse polarized, with nasty effects on impedance. A bipolar cap will have some crossover behavior.
Last not least, there is the issue of dielectric absorption which is suspected to smear impulses.
Eric
?
If we are talking "typical" amplifiers, the first stage usually doesn't have voltage gain, such as the ESP 3a (first stage is differential pair) that started off this thread.
So, the resistive part can be much higher than 1k. A simple DC blocking RC can be say, 22 uF cap with a 10kOhm resistor. Assuming a high input impedence input stage, that gives an distortion figure of <0.01% (using your ESR values).
-willryu
willryu said:
Sorry but I must correct you, the amp has indeed gain in the first stage (approx. 16). 99.9% of all amps has gain of at least 10 in the first stage. The cut frequency is given by pull-down resistor and the input impedance of the stage in particular case. If the stage has feedback the impedance is "high", no worries.
Personally I thrive to avoid electrolythics as much as possible, feels better with plastic and they also last longer.
As Eric alluded, performance about the zero crossing is the real killer for polar (or "bipolar") caps. Small absolute nonlinearities are pretty awful distortion-wise with low-level signals.
Even .01% distortion at the input stage is bad when the amp as a whole has less distortion than that. Whether .1% or .0025% /really/ matters is eminently debatable, but it seems a bad idea to throw performance away by using an electrolytic.
Even .01% distortion at the input stage is bad when the amp as a whole has less distortion than that. Whether .1% or .0025% /really/ matters is eminently debatable, but it seems a bad idea to throw performance away by using an electrolytic.
Peranders, I'll reply to your email in reverse order :
. Many 3 stage solidstate amps have all the voltage gain in the second stage. First stage is typically a differential pair with no voltage gain.
-willryu
:I also avoid electrolytics when I can, but I have always wondered the *technical* reasons why we should avoid electrolytics for *blocking DC*when the impedence of the Cap is way out of the audio band.
So I think you are agreeing with me here. Distortion in the audio band (20 - 20kHz) due to having blocking electrolytic will be very low. Less than 0.01% for figures quoted above.
No problem with correcting me! I mispoke (and misremembered). ESP3a does have gain in the input stage. But I will argue that your 99.9% figure is slightly off
. Many 3 stage solidstate amps have all the voltage gain in the second stage. First stage is typically a differential pair with no voltage gain.-willryu
willryu said:
I have no idea what the chemistry really does to the sound. I have never sat down and tested with and without electrolythics. As Nelson says (I agree with him) the biggiest contribution to good sound quality are the active elements and how they are connected. Tubes and semiconductors serves for 99.99% of the quality (no proof for that, just my opion)
willryu said:
No problem with correcting me! I mispoke (and misremembered). ESP3a does have gain in the input stage. But I will argue that your 99.9% figure is slightly off
I must add that it's very unsusual with minimal gain in the first stage. Nothing wrong really but you will get more noise probely. It's smart to have enough gain so the input transistor creates all of the noise. The second stage probely have more noise due to higher currents than the first.
gain of input stage ESP3a
I get the same figure for gain as peranders.
Gain of the differential stage is (R-collector / 2*r-emitter).
Total bias current is about 3ma (35V/12kOhms) so r-emitter is about 17 Ohms, using the approximation that r-emitter = 25/ I-collector (mA). This gives a gain of 560 Ohms / 2*16.6 = 16.5 or 24 dB.
This is quite high in my opinion!
Warning I am not an EE, so always double check my calcs.
-willryu
I get the same figure for gain as peranders.
Gain of the differential stage is (R-collector / 2*r-emitter).
Total bias current is about 3ma (35V/12kOhms) so r-emitter is about 17 Ohms, using the approximation that r-emitter = 25/ I-collector (mA). This gives a gain of 560 Ohms / 2*16.6 = 16.5 or 24 dB.
This is quite high in my opinion!
Warning I am not an EE, so always double check my calcs.
-willryu
Re: gain of input stage ESP3a
The gain is not very high, rather normal actually. The gain should be 3 to 10 (at least) in order to get low noise (if you think it is important).
willryu said:
This gives a gain of 24 dB.
This is quite high in my opinion!
Warning I am not an EE, so always double check my calcs.
The gain is not very high, rather normal actually. The gain should be 3 to 10 (at least) in order to get low noise (if you think it is important).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.