People don't get into pedantic arguments here? We can even disagree about how disagreeable the disagreements are. 
On this thread, there are comments that could easily be seen as derogatory and provocative for no good reason except to assert a form of dominance over others. This forum is one of the worst - I quit it many years ago because of this. I think it very likely I will again cos there are just too many swinging dicks here.
Case in point - argument over inductance when not all caps have such inductance - e.g. Mundorf Supreme silver/oil are "inductance free" to use their words - to which almost all caps are inferior imo... and yes, you can get them in small values to use as bypasses. I do And if I had deeper pockets...
Anyway, I think we could perhaps help the OP by not pursuing a discussion that has been made ad infinitum on the internet with no resolution... his question has been asked and answered.
...instead, let's hear about the other aspects of the design, which perhaps might be more critical to the sound quality, such as power and any xo. Does it even use an external xo? 6MHz is hardly common. I came across some cheap low phase noise XO's by Taiten (UX-O series) but they only go down to 12MHz.
On this thread, there are comments that could easily be seen as derogatory and provocative for no good reason except to assert a form of dominance over others. This forum is one of the worst - I quit it many years ago because of this. I think it very likely I will again cos there are just too many swinging dicks here. Case in point - argument over inductance when not all caps have such inductance - e.g. Mundorf Supreme silver/oil are "inductance free" to use their words - to which almost all caps are inferior imo... and yes, you can get them in small values to use as bypasses. I do
And if I had deeper pockets...Anyway, I think we could perhaps help the OP by not pursuing a discussion that has been made ad infinitum on the internet with no resolution... his question has been asked and answered.
...instead, let's hear about the other aspects of the design, which perhaps might be more critical to the sound quality, such as power and any xo. Does it even use an external xo? 6MHz is hardly common. I came across some cheap low phase noise XO's by Taiten (UX-O series) but they only go down to 12MHz.
Last edited:
All caps have inductance. To get a zero inductance cap it would have to be zero size, with no wires. This is true no matter how deep your pockets. In most cases the inductance of a cap is roughly equal to the inductance of a wire of the same size. Oops, have I just been "pedantic"?Hugh Jazz said:Case in point - argument over inductance when not all caps have such inductance - e.g. Mundorf Supreme silver/oil are "inductance free" to use their words - to which almost all caps are inferior imo... and yes, you can get them in small values to use as bypasses. I do
The OP asked a question about output caps. Various answers have been given, giving rise to some discussion. That is how this forum works. He seems happy with this.
Maybe not. Typically electrolytic ESR keeps going up too. It does not simply become inductive.
I have a reasonably good multi-frequency LCR (including ESR) meter. Large electrolytics at 100kHz will often confuse the meter. Who knows exactly what they are doing.
EDIT: However, in the audio band they are still generally capacitive. Putting a parallel cap still increases capacitance at audio frequencies.
Last edited:
Yes, but adding a 0.1uF bypass to a 1000uF electrolytic does not increase its capacitance by a useful amount in the audio region.
In the region below series resonance it does nothing useful.
In the region around and some way above series resonance it makes things worse.
Only in the region a long way above does it do anything useful, but that is of no interest for an audio coupling capacitor. Not many tweeters can do 100kHz!
By all means carefully bypass a decoupling capacitor used in an RF circuit, but not audio coupling caps. I suspect the fad for this arose from someone who learnt a little about electronics, but not enough. It is a classic example of a popular tweak which does the opposite of what is intended, and what is intended is in any case unnecessary.
In the region below series resonance it does nothing useful.
In the region around and some way above series resonance it makes things worse.
Only in the region a long way above does it do anything useful, but that is of no interest for an audio coupling capacitor. Not many tweeters can do 100kHz!
By all means carefully bypass a decoupling capacitor used in an RF circuit, but not audio coupling caps. I suspect the fad for this arose from someone who learnt a little about electronics, but not enough. It is a classic example of a popular tweak which does the opposite of what is intended, and what is intended is in any case unnecessary.
Saying it does the opposite of what is intended is probably overly dramatic. It is usually intended to improve HF sound quality in the audio region. Agreed, it may not work as intended.
Also, if not for ESR, a parallel cap would start helping immediately above parallel resonance. That the effect is spread out over a wide range of frequencies is only because Q is low. After all, parallel resonance is the same phenomenon used to tune crystal radios. They are not highly selective, but there is a definite impedance peak that then falls off with increasing frequency.
Another thing about electrolytics is they produce some noise from leakage, although probably not as bad as tantalum. They become less effective at filtering out their own noise as they increasingly become less ideal as capacitors with increasing frequency, including due to ESR.
So, I would say that if there is a clear audible improvement from bypassing an electrolytic in some particular case, there may be something going on that has not been completely described by simplified modeling.
However, it most cases it is doubtful there is any actual audible improvement, only what we can probably agree as a fairly useless small increase in capacitance in the audio band, and some other stuff far above that which may not matter at all.
Also, if not for ESR, a parallel cap would start helping immediately above parallel resonance. That the effect is spread out over a wide range of frequencies is only because Q is low. After all, parallel resonance is the same phenomenon used to tune crystal radios. They are not highly selective, but there is a definite impedance peak that then falls off with increasing frequency.
Another thing about electrolytics is they produce some noise from leakage, although probably not as bad as tantalum. They become less effective at filtering out their own noise as they increasingly become less ideal as capacitors with increasing frequency, including due to ESR.
So, I would say that if there is a clear audible improvement from bypassing an electrolytic in some particular case, there may be something going on that has not been completely described by simplified modeling.
However, it most cases it is doubtful there is any actual audible improvement, only what we can probably agree as a fairly useless small increase in capacitance in the audio band, and some other stuff far above that which may not matter at all.
Last edited:
Indeed. And people do hear a benefit, while others argue this is not possible so they disparage anyone who disagrees with them. And this theory vs experience theme repeats ad nauseum.
because I saw some projects here that are fabulous. This one in particular 16x Digital interpolation filter - drive PCM56, PCM58, AD1865 and so on up to 768 kHz
You re-purposed the language - I was comparing the way people denigrate/dismiss/insult the opinions of others without evidence, not the concept of a delusion. Anyway, this thread is now way off topic (my fault).
If you generally interpret disagreement as disparagement then you may find your time on this forum unpleasant. I am unclear who you are quoting when you say "deluded". You will note that Markw4 and I were able to discuss the technical matter without insulting each other.Hugh Jazz said:
The task of a coupling cap is quite simple, as outlined in an earlier post; it needs sufficiently high impedance at DC, and sufficiently low impedance for audio. That is why some degradation (e.g. by adding a bypass) usually does little harm. However, the 'bypass' meme is now so well established that some newbies want to bypass film caps with smaller films caps; this could do more harm than bypassing an electrolytic as film cap ESR is so low that there is little to damp the resultant parallel resonance, although fortunately the resonant frequency may be too high to matter unless it coincides with a local RF source.
In case people missed it, from 2005 Capacitor sound quality, silver mica in particular
From 2002 https://www.diyaudio.com/forums/solid-state/525-using-tantalium-capacitors-2.html#post5762
Here there is an objective attempt to prove that, in one particular instance of coupling which is 1uF into 100k, AC with no DC, there is no measurable or audible difference between a cheap electro and an expensive teflon coupling (who would do this anyway?). The test and subsequent tests focus on distortion at low-mid frequencies and high volumes (I would test high frequencies and low volumes since this is where i've heard audible differences between caps) but at least they are a start towards some evidence, albeit with a clearly expressed and strong bias towards a given outcome, so hardly scientific. SYclotron Audio | Why are people obsessed with coupling caps? Unfortunately, still nothing on bypasses though. Let the debate rage on. and on.
From 2002 https://www.diyaudio.com/forums/solid-state/525-using-tantalium-capacitors-2.html#post5762
Here there is an objective attempt to prove that, in one particular instance of coupling which is 1uF into 100k, AC with no DC, there is no measurable or audible difference between a cheap electro and an expensive teflon coupling (who would do this anyway?). The test and subsequent tests focus on distortion at low-mid frequencies and high volumes (I would test high frequencies and low volumes since this is where i've heard audible differences between caps) but at least they are a start towards some evidence, albeit with a clearly expressed and strong bias towards a given outcome, so hardly scientific. SYclotron Audio | Why are people obsessed with coupling caps? Unfortunately, still nothing on bypasses though. Let the debate rage on. and on.
Short thread with few facts, plenty of unsupported assertions and some misinformation. Nothing of any value to glean from there.Hugh Jazz said:
That is because any weakness in a coupling cap will first show itself at lower frequencies and higher signal levels.Hugh Jazz said:
I am going to guess that is because SY knows a bit about electronics.
I note you did not comment on the second link I posted. Well worth reading all of those posts I think - nothing new under the sun?
14VAC ? and no DC ? This is silliness - what gear uses signal coupling in such a situation - no cap is necessary. Low-mid frequency THD does not measure detail retrieval/smothering/slurring etc. I didn't see any details about the capacitance in the source - or any mention of that possibly affecting subjective results. The lack of a control test for both subjective and "objective" testing (e.g. box with a wire and another with a ceramic coupling) shows he set out to prove his own bias. People rarely fail at that and his experiment would be rejected by peer review because of the many flaws. It's barely less unreliable than those claiming they hear a change from a bypass, which might fall under some interpretivist study. Maybe someone will do some real experimentation rather than bias confirmation. In the meantime, there is overwhelming data that caps sound different and there are distinct differences even among common types in different circuits. Google is your friend on that if that is not a given. Anyway, I await real evidence re bypassing. And in the meantime, I will continue to enjoy the superior sound quality of DC coupling
Last edited:
We know that some types of caps behave more like RC ladder networks than like simple capacitor models. Such ladder networks are linear and do not create harmonic distortion. However, they may still produces audible effects that some people find objectionable. It is just a type of linear distortion instead of nonlinear distortion.
DC on a coupling cap merely turns 3rd order distortion into 2nd order distortion. It doesn't fundamentally change the nonlinearity of the capacitor.
No test will convince the doubters, because they will always claim that the wrong test was carried out. Unfortunately they never come up with the right test themselves; maybe they prefer an air of mystery?
There is overwhelming anecdote that coupling caps sound different; very little evidence. Where evidence of genuine audible difference is found then there is usually a simple electrical explanation, and the actual difference is quite small. If not, we would have to discard circuit theory as a valid means of designing audio and replace it with some form of alchemy. Some have tried this, but usually they end up with circuits with glaring fundamental errors and clearly audible problems.
Who mentioned THD? He measured individual distortion components, and showed that a 'good' cap and a 'bad' cap were the same (apart from hum pickup, where the 'good' cap was worse). There is no other source of problems with "detail retrieval/smothering/slurring" but distortion; low distortion means low problems.
Are you saying he should have measured a wire too? Perhaps he should, just to show doubters that his test equipment is good enough to do the capacitor test. Interestingly, a 'really bad' ceramic cap might have done reasonably well in this test (at least at 1kHz) - thus demonstrating that being a coupling cap is a fairly easy occupation.
No test will convince the doubters, because they will always claim that the wrong test was carried out. Unfortunately they never come up with the right test themselves; maybe they prefer an air of mystery?
There is overwhelming anecdote that coupling caps sound different; very little evidence. Where evidence of genuine audible difference is found then there is usually a simple electrical explanation, and the actual difference is quite small. If not, we would have to discard circuit theory as a valid means of designing audio and replace it with some form of alchemy. Some have tried this, but usually they end up with circuits with glaring fundamental errors and clearly audible problems.
He started with subjective listening - for this to be credible, the listeners have to be tested. If they can't hear the difference between a cap and a wire, then either the listeners are unsuitable or his test is - equipment, environment, etc.
I've assumed that this is what I'm hearing - a reduction in the "objectionable" effects inflicted on high frequencies of low amplitude passing through a "big" electro. I've tried different bypasses and found, for example, Wima MKS to have no effect on this. The Vishay MKP18xx series do but it seems to me this is not a dependable result - I've assumed there are complexities in real use that are beyond circuit simplification theories. But I've never heard one of these Vishays have a detrimental effect. Given the low cost of an MKP1837 I can see no reason not to try and see if it improves fine detail on soft passages. To me, this is like tight, fast, deep, clear, tuneful bass. It's very satisfying.
audible difference.
Why pre-suppose an investigation should target weaknesses? What if both have none in that test? What does that prove?
This is where this discussion should be heading imo because a debate is futile. I think we need a voltage out DAC like AKM with no signal filtering capacitance at all, and/or perhaps "boutique" caps - make it as "good" as a subjective listener would expect. For subjective testing, it is coupled into a DC headphone amp - it's far easier to hear this kind of change in decent cans. The coupling ... hmmm... it will sound ghastly with no filtering ...
...objective testing is easier? - white noise, pink noise, HF and ultrasonics in various combos, with some LF for IMD etc. Measure amplitude, slew, distortion, etc. But then, wow, so many caps in so many combinations to test...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.