[b0dhi]

Quote:

Originally Posted by SmellyGas /img/forum/go_quote.gif This is a common "belief" that is propagated here. I don't necessarily think it's true, even though it would "make sense" that it is true. There is a lot of audio mythology that "makes sense" but doesn't pan out to audible differences. In fact, if you look at loudspeaker amplifier data, when The Audio Critic tests them with stressful inductive, capacitative, and low impedance loads, only rarely does even the most stressful load substantially influence the measurements of interest to audibility.

Are you kidding me? This has nothing to do with head-fi or audiophilia, it's plain old electrical engineering.

Also, I suggest you approach The Audio Critic with some scepticism because in almost every case where I come across something they've said, they exhibit a very narrow perception and often a lack of consistent rational thinking. It's bias behind a facade of scepticism.

Also, I've noticed you like to refer to audibility studies. That's a good thing. However, you should understand that a study with 7 test subjects (regardless of the number of trials) is not a significant enough sample size to be able to say anything about what can potentially be heard by the best ears out there. Granted, many audiophiles don't have the golden ears they claim to, but that doesn't mean none do.

 

[SmellyGas]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif Also, I suggest you approach The Audio Critic with some scepticism because in almost every case where I come across something they've said, they exhibit a very narrow perception and often a lack of consistent rational thinking. It's bias behind a facade of scepticism.

You're certainly entitled to your own opinion.

Quote:

Also, I've noticed you like to refer to audibility studies. That's a good thing. However, you should understand that a study with 7 test subjects (regardless of the number of trials) is not a significant enough sample size to be able to say anything about what can potentially be heard by the best ears out there. Granted, many audiophiles don't have the golden ears they claim to, but that doesn't mean none do.

There's a good paper in the AES Journal that summarizes the data from about 20 different studies that essentially show the same thing as the one I linked. This encompasses far more than just 7 test subjects. Furthermore, if differences in amplifiers were so profound and dramatic as people here claim, I would have expected at least ONE published study to at least be able to show MINIMAL differences in a blinded, controlled comparison. I can provide you a link to the paper if you're interested. I doubt you will be...because people who read JAES are usually electrical engineers and I would surmise that few electrical engineers believe in amplifier mythology (though there are exceptions...just like how there are physicians who smoke cigarettes....)

 

[gilency]

SmellyGas: you have provided very interesting insight to somebody with knowledge of scientific methods. I appreciate your input and skepticism which are something to be valued. Thank you.

 

[SmellyGas]

Quote:

Originally Posted by gilency /img/forum/go_quote.gif SmellyGas: you have provided very interesting insight to somebody with knowledge of scientific methods. I appreciate your input and skepticism which are something to be valued. Thank you.

Thanks, man. I have yet to find a SINGLE person who has given these studies a critical read, has the knowledge/background to appropriately interpret the methodology/statistics/validity, and STILL believes that loudspeaker amplifiers make LARGE, EASILY-AUDIBLE differences in sound.

...now for electrostatic headphone amplifiers...I personally don't think there is sufficient evidence to conclusively believe one way or the other.

 

[00940]

Quote:

Originally Posted by smellygas When it was pointed out that the stat headphones are capacitative loads and that very little current (and hence power) would actually be drawn, people got irritated.

Have you ever tried to push 500V @20khz through 150pf ? Once you start thinking about how to do it properly, it suddenly becomes a lot of current to deal with. I suggest you the reading of the first pages of this article (anyone who isn't familiar with those concepts has no business discussing electrostatic amplifiers):
The Tube CAD Journal, Electrostatic Headphones

Btw, it's worth your time to have a look at the schematics of those electrostatic amplifiers (quite a lot are in the public domain). They don't share much in common with typical audio amplifiers and do use different ways to skin the proverbial cat. It is difficult as such to guarantee that we're doing apple to apple comparatives on technical terms.

The electrostatic amps typically use low or no global negative feedback. As usual with low nfb devices, the THD level and pattern can increase/vary heavily when the output power goes up. And this way before they hit their clipping point. You won't see that with typical solid state amplifiers (as long as the power supply is adequately sized) but it's a sad reality with electrostatic amps. Spice simulators aren't perfectly accurate when it comes to THD predictions, but they'll at least teach you that when comparing two designs.

The voltage swing thing isn't stupid (coming from JCX, it would be surprising). First of all, we know that some electrostatic amps -hev70- do clip when playing music. If I may say, in logic, "one false" is enough to disprove a universal statement. So we know that all electrostatic amps don't sound the same. What about all electrostatic amps with sufficient voltage swing to achieve reasonable levels ? If the cheap one is engineered to barely reach those levels, it'll probably be operating with significant thd. To make things worse, THD is apparently more audible at higher frequencies ( Experimental Study : Distortion - Axiom Audio ), where the amp will be struggling to deliver the necessary current. The amp with a higher voltage swing will probably be "over-engineered" for common listening levels, something that might keep distortion under audible levels.

But of course, all this is only educated guess.

 

[SmellyGas]

Quote:

Originally Posted by 00940 /img/forum/go_quote.gif Have you ever tried to push 500V @20khz through 150pf ? Once you start thinking about how to do it properly, it suddenly becomes a lot of current to deal with. I suggest you the reading of the first pages of this article (anyone who isn't familiar with those concepts has no business discussing electrostatic amplifiers): The Tube CAD Journal, Electrostatic Headphones

Thank you for the very informative link. It's very interesting. However, I do have a slight concern with your calculation. Above, you state that pushing 500V @ 20kHz through 150pf "becomes a lot of current to deal with." However, the capacitative reactance (impedance at a given freq) comes out to ~50kohms, using the standard Xc = 1/(2*pi*f*C) formula. 50kohms at 500V gives you 10mA. 10mA at 500V is 5W. So once again, I am having a lot of difficulty accepting the explanation that electrostatic headphone amplifiers sound different because of the inability of "small" amplifier to deliver large amounts of power. 10mA and 5W at full voltage is hardly a lot.

Quote:

Btw, it's worth your time to have a look at the schematics of those electrostatic amplifiers (quite a lot are in the public domain). They don't share much in common with typical audio amplifiers and do use different ways to skin the proverbial cat. It is difficult as such to guarantee that we're doing apple to apple comparatives on technical terms.

I completely agree. Which is why I don't think the blind listening tests that fail to show LARGE, EASILY-AUDIBLE differences in loudspeaker amplifiers are generalizable to electrostat headphone amps. At the same time, for the reasons stated above and elsewhere, the explanation that different amps are more "powerful" and that electrostatic headphones (essentially capactitative loads) require a lot of "power" just does not make sense.

Quote:

The electrostatic amps typically use low or no global negative feedback. As usual with low nfb devices, the THD level and pattern can increase/vary heavily when the output power goes up. And this way before they hit their clipping point. You won't see that with typical solid state amplifiers (as long as the power supply is adequately sized) but it's a sad reality with electrostatic amps. Spice simulators aren't perfectly accurate when it comes to THD predictions, but they'll at least teach you that when comparing two designs.

Okay, that's interesting. However, I am also aware of a lot of amplifiers that do not use (or have low amounts of) negative feedback and they claim that this is a feature.

Quote:

The voltage swing thing isn't stupid (coming from JCX, it would be surprising). First of all, we know that some electrostatic amps -hev70- do clip when playing music. If I may say, in logic, "one false" is enough to disprove a universal statement. So we know that all electrostatic amps don't sound the same.

That's kind of a ridiculous line of discussion. That's like saying Bob smoked for 50 years and he's perfectly healthy, therefore it disproves the universal statement that "smoking causes heart attacks and cancer." I think the relevant question is, and I think you know this, is do different commonly available commercial (and semi-commercial) electrostatic amplifiers make large, audible differences in sound quality?

Quote:

What about all electrostatic amps with sufficient voltage swing to achieve reasonable levels ? If the cheap one is engineered to barely reach those levels, it'll probably be operating with significant thd.

Maybe, maybe not. I'm sure you've looked at plenty of THD vs. power/voltage curves, and THD typically stays inaudibly low throughout the entire rated output until JUST before clipping, then it skyrockets rapidly. So, I would expect a properly engineered amplifier operating near (but not too near) its max output to not put out significantly more THD than at lower outputs.

[/QUOTE]

 

[00940]

Quote:

Originally Posted by SmellyGas /img/forum/go_quote.gif So once again, I am having a lot of difficulty accepting the explanation that electrostatic headphone amplifiers sound different because of the inability of "small" amplifier to deliver large amounts of power. 10mA and 5W at full voltage is hardly a lot.

If you consider the design choices of the smaller amplifiers (in terms of parts and the like), 5W becomes a lot. Everything must be put in perspective of course and related to what exists, not to some amps "in abstraction". The same goes for amps using step up transformers. A 5W output transformer with decent specs isn't that easy to design.

Quote:

Okay, that's interesting. However, I am also aware of a lot of amplifiers that do not use (or have low amounts of) negative feedback and they claim that this is a feature.

Well, the amount of negative feedback clearly influences the pattern of distortion and this seems to be audible. See this article by Nelson Pass for some simple background on this : http://passlabs.com/pdf/articles/dis...d_feedback.pdf

Quote:

" I think the relevant question is, and I think you know this, is do different commonly available commercial (and semi-commercial) electrostatic amplifiers make large, audible differences in sound quality?

Audible is enough. Why large ?

Quote:

Maybe, maybe not. I'm sure you've looked at plenty of THD vs. power/voltage curves, and THD typically stays inaudibly low throughout the entire rated output until JUST before clipping, then it skyrockets rapidly. So, I would expect a properly engineered amplifier operating near (but not too near) its max output to not put out significantly more THD than at lower outputs.

First you assume boldly that all the amps available are properly engineered. I'm a bit more cautious here (Singlepower anyone ?).

Second, the problem is that we don't have thd figures for ESL amps (they're almost impossible to make). How quickly does THD raises with output power in a marginal design and when does it breaks the audibility threshold ? Difficult to say. I quickly ran a cathode follower in sims, without feedback. This buffer can push more than 1W into 300ohms before hard clipping. At 0.5mW, thd is under 0.06%. At 50mw, we're at 0.6%. At 10mW, we're nearing 1%. At 500mW we're at 2% ! So much for matching the thd curves of a chipamp datasheet.

Third, I've indeed seen plenty of those curves. Actually, you could even say that the THD+N often diminishes when power increases, until clipping kicks in. The problem is that those curves don't tell anything about distortion patterns and apply to high gnfb amps.

 

[SmellyGas]

Quote:

Originally Posted by 00940 /img/forum/go_quote.gif If you consider the design choices of the smaller amplifiers (in terms of parts and the like), 5W becomes a lot. Everything must be put in perspective of course and related to what exists, not to some amps "in abstraction". The same goes for amps using step up transformers. A 5W output transformer with decent specs isn't that easy to design.

Well let's do more math then. Take a Stax 727, driving 1kHz through 160pf. The reactance now is ~1 MEGAohm, and even at the 727's MAX VOLTAGE output of 450V, we're talking 0.45mA and 0.2W. Really now. That is just a miniscule amount of current and power. And that's at max output! I personally have a hard time believing that nowadays, it is truly a design challenge to create an audio amplifier that struggles to produce 0.2W. Based on the explanation so far, I just cannot see how differences in "power" can account for why people believe electrostatic amplifiers sound different.

Quote:

Second, the problem is that we don't have thd figures for ESL amps (they're almost impossible to make). How quickly does THD raises with output power in a marginal design and when does it breaks the audibility threshold ? Difficult to say.

What specific aspect of a commercially available electrostatic headphone amp circuit would cause it to have a gradually rising THD vs. power output curve (rather than near-flat, then rapidly rising just before clipping like pretty much every other audio amplifier)? I.e. why would you even think that would be the case?

Quote:

I quickly ran a cathode follower in sims, without feedback. This buffer can push more than 1W into 300ohms before hard clipping. At 0.5mW, thd is under 0.06%. At 50mw, we're at 0.6%. At 10mW, we're nearing 1%. At 500mW we're at 2% ! So much for matching the thd curves of a chipamp datasheet.

That may be the case for your particular extreme scenario in a non-solid-state design. So are you suggesting that it is difficult to design electrostatic headphone amplifiers with low distortion, even though it is dirt simple to produce loudspeaker amplifiers that drive 100+W power with < 0.08% THD? Are you also suggesting that the printed specifications of all the Stax amps are fictitous because they all seem to specify < 0.01% THD (1kHz / 100V)?

 

[00940]

I'm suggesting you are out of your field of expertise and are doing circles in the sand, asking questions I already answered.

1/ Doing the calculations at 1khz obviously gives an easy load with stats. Do it again at 18khz. Multiply that dissipation by 3 (for low efficiency class A designs) and see what you come with. Consider what tubes/transistors are used.
2/ You are still not matching your candid a-priori about amplifier designs ("not a design challenge") to actual designs.
3/ Just because it seems the only way to close this topic, I simulated this Stax design: Electrostatic Headphones Amps (keeping in mind it's all indicative figures, probably not the exact truth).

100V swing @1khz into 160pf: 0.17% thd
300V swing @1khz into 160pf: 0.8% thd (clipping isn't reached, clipping is at around 450V swing)
100V swing @18khz into 160pf: 0.23% thd
200V swing @18khz into 160pf: 1.4% thd
300V swing @18khz into 160pf: 8% thd (clipping is reached at around 250V)

This identifies two problems already exposed:
- rising thd with power output, even before clipping.
- highly different clipping points for varying frequencies.

 

[SmellyGas]

Quote:

Originally Posted by 00940 /img/forum/go_quote.gif I'm suggesting you are out of your field of expertise and are doing circles in the sand, asking questions I already answered.

If I were an expert on electrostatic headphone amplifiers, I obviously wouldn't be asking what specifically makes one amp sound better than the other, now would I. That goes without saying.

Quote:

1/ Doing the calculations at 1khz obviously gives an easy load with stats. Do it again at 18khz. Multiply that dissipation by 3 (for low efficiency class A designs) and see what you come with. Consider what tubes/transistors are used.

I think you're completely missing the point. People here claim that some electrostat headphone amps sound better because they can deliver more power. People also claim that electrostat headphones, themselves, consume a LOT of power. As I have already demonstrated, the amount of power drawn under MAXIMUM output conditions is already miniscule - 5W MAX at 20khz and 0.2W MAX at 1khz, assuming typical 160pf stax cans. I don't know why you bring up 18khz, since I know you know that this frequency only makes up a small fraction of music. Whether or not you're using an inefficient circuit (class A, in your example), is irrelevant. Your amplifier could be 0.1% efficient and require 5000W to drive 5W into your headphone load, for instance, but it still won't change the fact that the headphone only required 5W.

Quote:

2/ You are still not matching your candid a-priori about amplifier designs ("not a design challenge") to actual designs.

You're going to have to clarify this statement because I have no idea what you're talking about.

Quote:

3/ Just because it seems the only way to close this topic, I simulated this Stax design: Electrostatic Headphones Amps (keeping in mind it's all indicative figures, probably not the exact truth). 100V swing @1khz into 160pf: 0.17% thd 300V swing @1khz into 160pf: 0.8% thd (clipping isn't reached, clipping is at around 450V swing) 100V swing @18khz into 160pf: 0.23% thd 200V swing @18khz into 160pf: 1.4% thd 300V swing @18khz into 160pf: 8% thd (clipping is reached at around 250V) This identifies two problems already exposed: - rising thd with power output, even before clipping. - highly different clipping points for varying frequencies.

Thank you for taking the time to doing a simulation. That was really helpful. Unfortunately, it basically supports my point. Using the SR-007 as an example, 100V yields 100dB at 1kHz. I'm sorry, but I don't listen to my headphones anywhere near that level, but let's use it as an example. So based on your simulation, driving the SR-007's would yield 0.17% THD at 1khz and 0.23% THD at 18khz. No unless you claim to be able to hear 0.2% THD, AND you claim that the THD produced by the headphones themselves are anywhere near that low, then the contribution of the amp you simulated to THD is negligible.

 

[00940]

100Vrms on stax website = 300V swing.

 

[Bullseye]

(some data is interesting here; nor surprising, though)

 

[SmellyGas]

Quote:

Originally Posted by 00940 /img/forum/go_quote.gif 100Vrms on stax website = 300V swing.

Looking at some more THD vs. power chart of TUBE preamps and loudspeaker amps, I would have to say that your results are probably typical for TUBE amps - THD appears to gradually rise as a function of power. Solid-state amps, on the other hand, have relatively flat (and low) THD until just before clipping. Even though your data only includes two points for 2 different frequencies, it probably accurately describes distortion behavior of tube amps.

However, this doesn't really change anything, and I'm not sure why you chose to model a valve amp. It is already known that tubes have high levels of distortion, and it also already known that tubes have difficulty driving large amounts of power at low distortion. Are you now suggesting that electrostat headphone amps sound different because some have more distortion than others? In that case, solid-state amps, which have far less THD below clipping, should ALL sound better than tube amps.

Let me just summarize:
It is commonly believed here that different electrostatic headphone amplifiers sound different because some are more powerful and electrostatic headphone requires lots of power. Unfortunately, as we have calculated, the amount of power consumed by headphones at MAXIMUM output is miniscule - 5W at 20khz (where there is little musical input) and 0.2W at 1khz. So that doesn't fly. Then there's the notion that different electrostat headphone amps differ in their ability to provide power, which implies that it is somehow difficult to design an amp that delivers 0.2-5W. Perhaps this is the case for a tube amp, as your simulation of an old Stax tube amp shows, but it certainly cannot be the case for solid state designs. Unless you happen to know of solid state designs that generate high levels of THD (or other audible aberrancies) at normal listening volumes because of power limitations, I still don't see how this is a plausible explanation for why electrostatic headphone amps should sound appreciably different, assuming that they even do. That's basically where I'm coming from.

 

[00940]

Quote:

Originally Posted by SmellyGas /img/forum/go_quote.gif However, this doesn't really change anything, and I'm not sure why you chose to model a valve amp. It is already known that tubes have high levels of distortion, and it also already known that tubes have difficulty driving large amounts of power at low distortion.

I dunno... maybe because it's a top of the line Stax design I simulated ? Maybe because many of those available stat amplifiers are still tubes based ? Maybe because there's a reason designers continue designing tubes based amplifiers (hint: high voltage and transistors don't mix well) ?

Quote:

Are you now suggesting that electrostat headphone amps sound different because some have more distortion than others? In that case, solid-state amps, which have far less THD below clipping, should ALL sound better than tube amps.

I thought you were approaching this as a scientist... define "sound better". However, I would indeed suggest that most electrostat headphones amps have an unique distortion pattern (level and harmonic distribution) which can reach audibility levels. How else would I justify that different amps sound different ?

Quote:

Let me just summarize: It is commonly believed here that different electrostatic headphone amplifiers sound different because some are more powerful and electrostatic headphone requires lots of power.

You are over simplifying the argument. "Lots of power" must be judged wrt the available amplifiers. And obviously, some of those lack the power necessary to reach properly high sound level at high frequencies and thus probably distort already at lower levels. It isn't ridiculous for buyers to be attracted to more powerful amplifiers, which shouldn't exhibit those problems.

Quote:

Then there's the notion that different electrostat headphone amps differ in their ability to provide power, which implies that it is somehow difficult to design an amp that delivers 0.2-5W. Perhaps this is the case for a tube amp, as your simulation of an old Stax tube amp shows, but it certainly cannot be the case for solid state designs.

Oh really ? Please remove that "certainly cannot". I assure you it isn't so easy to design a solid state amplifier, with a correct frequency response, which can run on +/- 300V supplies and swing that much voltage/current properly. Ask the guys who just designed a diy stat amplifier. Sorry but you are making wild assumptions here.

 

[Donald North]

Quote:

Originally Posted by SmellyGas /img/forum/go_quote.gif Looking at some more THD vs. power chart of TUBE preamps and loudspeaker amps, I would have to say that your results are probably typical for TUBE amps - THD appears to gradually rise as a function of power. Solid-state amps, on the other hand, have relatively flat (and low) THD until just before clipping. Even though your data only includes two points for 2 different frequencies, it probably accurately describes distortion behavior of tube amps.

Relatively low and flat THD+N curves are due to feedback. Reduce or eliminate the feedback and THD will rise with increasing signal level. Here's a Pass Labs Aleph 1.2 for example:

Here's a link to all the measurements:
Stereophile: Pass Aleph 1.2 monoblock power amplifier

Quote:

Originally Posted by SmellyGas /img/forum/go_quote.gif It is already known that tubes have high levels of distortion

A triode when loaded by a high impedance is a lower distortion voltage amplifier than a transistor when both are compared on their own without feedback. Transistors need to be used in multiples and/or with feedback to attain low measured distortion.

http://www.its.caltech.edu/~musiclab...er-acrobat.pdf

Here's a tube preamp with markedly low distortion:
SoundStage! Measurements - Convergent Audio Technology SL1 Ultimate Mk 2 Preamplifier (6/2007)