While those two brands of caps are the most difficult to sort.
Anyway, the guy who wrote that caps are transparent in an audio system is, or dishonest, or endowed with an impaired hearing, or own a hifi system that I will not envy.
The last hypothesis is that he absolutely don't know how to listen and what listen to.
On a (my) subjective point of view, electrolytics can bring a grainy sound, and the other flaws play on "micro contrast" or "Sharpness" of the instruments. Instant micro dynamic.
Staring at their 'continuous signal' generators and distortiometers, these are things they do not know about. It would not even happen to them ... to listen ?
What I do not understand is why, whenever someone talks about feelings, listening impressions, witch is, at the end, the matter of our work in audio, it puts them in a state of caricature anger.
(As you can see, If I want, I can be as disagreeable as some around here ;-)
This works and extends distortion measurement abilities of common soundcards. I use this method as well.
Yes, but that was related to your usage of Cyril´s test results. Whenever you apply something the way you´ve mentioned in your last post ,you´re establishing a model theory from which you conclude about what´s possible in reality.
Scott was highlighting the double standard being applied: any sort of DC on the input significant enough to worry about leakage across a coupling capacitor will completely blow a DC servo circuit up (and the amp with it).
Tournesol: if we take pathological cases of both capacitors and servos, we can make them audible. On the other side, we can make both have errors that are miniscule. It all comes down to implementation and parts selection. And you're the one that rants about some people being locked into "the one and only truth?!". C'mon man, that's a bit ridiculous.
Music-like or modified real music test signals can be generated still nothing shows up. You are not very disagreeable.
Pointing out someone else's bad design decision does not invalidate the underlying design. People have used 1/4W resistors in the feedback network of 100W amps, Hi-K ceramic coupling caps, and at least one popular mic pre-amp kit uses horribly micro-phonic input coupling caps.
All of these things can be rectified via old fashion good engineering.
Just a simple question or two:
How many caps in the signal path does it take to become audible? Is one enough?
In that case, we are completely screwed, all of us, because all microphones that require phantom power, send the signal through an coupling cap with a serious DC bias over them.
Then there is the issue of loudspeaker crossovers. They are loaded with caps that are doing some serious work. Active xovers idem. DSP solutions? All DAC's I know have caps somewhere in the signal path, as per manufacturers recommendation.
Etc. etc.
So what makes caps in amps especially evil and audible?
How many caps in the signal path does it take to become audible? Is one enough?
In that case, we are completely screwed, all of us, because all microphones that require phantom power, send the signal through an coupling cap with a serious DC bias over them.
Then there is the issue of loudspeaker crossovers. They are loaded with caps that are doing some serious work. Active xovers idem. DSP solutions? All DAC's I know have caps somewhere in the signal path, as per manufacturers recommendation.
Etc. etc.
So what makes caps in amps especially evil and audible?
Last edited:
It's been quiet on this issue but since I'm back home from holiday with all my toys, so here is another input. This is the spectrum of a 200Hz tone FM modulated at 2Hz with a peak deviation of 20Hz.
Anyone listening to this would say the 2Hz modulation is obvious but of course there is no 2Hz component to the sound.
The scale is relative dB with arbitrary reference and the green plot is just 200Hz without modulation as a reference. The frequency scale is 0.1Hz per bin so divide by 10.
This signal will not have any low frequency components unless is passes through a non-linear process, the DA model with a ladder of ideal R's and C's does not qualify. Therefore filtering this signal at 20Hz has no effect on the envelope whatsoever.
Anyone listening to this would say the 2Hz modulation is obvious but of course there is no 2Hz component to the sound.
The scale is relative dB with arbitrary reference and the green plot is just 200Hz without modulation as a reference. The frequency scale is 0.1Hz per bin so divide by 10.
This signal will not have any low frequency components unless is passes through a non-linear process, the DA model with a ladder of ideal R's and C's does not qualify. Therefore filtering this signal at 20Hz has no effect on the envelope whatsoever.
Attachments
Scott, would you mind posting the wav? Or, I can generate the FM tone myself based on your definition. I was playing with FM test tones about 15 years ago (generated by CoolEdit). For a reasonably linear amplifier, they had shown nothing. I would go even farther and declare that the real music is the easiest test signal for the amp that we can imagine. Just opposite to popular beliefs.
Only if your source material involves a microphone. And, some popular mics are far from clean....
It was just CoolEdit.
Dropbox - vibe.wav
I used a 10 sec .wav and 441000 FFT to avoid any ambiguity. I can't go back to tools that only do power of 2 FFT's too much fiddling with details.
Last edited:
Yes, 20Hz peak at a 2Hz rate. The level is flat i.e. there is no envelope but the 2Hz is an audible impression almost as much as with AM. One point might be that it is very easy to fool yourself into thinking something is there when it isn't. See for instance Diana Deutsch's AES presentations.
Last edited:
I am sure you know that a question asked this way is a total nonsense.
Yes, and no, of course. The point I wanted to make is: given that it is inconceivable that your audio signal did not pass through at least a couple of caps by the time it reaches you ear, what is it that would make caps in amps especially problematic. Is it the straw that breaks the camels back? If so why? Without further explanation, I have to conclude that a well proportioned coupling cap is acoustically benign.
Last edited:
- Status
- Not open for further replies.