THD can finesse most everything else surely - if a design is 0.001% or better across full frequency and power range (ignoring noise contribution) you don't have to worry about any other figure really.
In reality you often only see THD at 1kHz quoted, wherease the 20kHz figure (or intermodulation for 18kHz+19kHz or similar) tells a lot more about the weaknesses of the amp.
And you really want to look at THD into a difficult load too - if its very small with difficult loads at all audible frequencies upto nominal full power, it tells you a lot in one figure. It tells you the ampliier is linear.
For this reason I reckon something like 15kHz/16kHz intermodulation distortion at nominal power is a useful single figure of merit, although perhaps not to be tested with tweeters attached!
In reality you often only see THD at 1kHz quoted, wherease the 20kHz figure (or intermodulation for 18kHz+19kHz or similar) tells a lot more about the weaknesses of the amp.
And you really want to look at THD into a difficult load too - if its very small with difficult loads at all audible frequencies upto nominal full power, it tells you a lot in one figure. It tells you the ampliier is linear.
For this reason I reckon something like 15kHz/16kHz intermodulation distortion at nominal power is a useful single figure of merit, although perhaps not to be tested with tweeters attached!
A quote from Dr. Earl Geddes:
"...we know so little about how humans perceive the sound quality of an audio system, and in particular the loudspeaker, that one should question almost everything that we think we know about measuring it. From what we have found most of what is being done in this regard is naive."
http://www.gedlee.com/Papers/Comments on howard.pdf
He also said:
"This is precisely where the signal-based distortion metrics fail. In our next paper we will show that .01% THD of one type of nonlinear system can be perceived as unacceptable while 10% THD in another example is perceived as inaudible. Even one of these simple examples is sufficient to invalidate THD as a viable metric for discussion of the perception of distortion. Furthermore, 1% THD is not at all the same as 1% IM, but we will show that neither correlates with subjective perception."
http://www.gedlee.com/Papers/Distortion_AES_I.pdf
Then there is Bob Cordell's book on power amplifier design, 2nd ed., has a whole chapter (#16) dedicated to other sources of distortion, including EMI distortion. Some of that stuff doesn't show up very well on a full set of pushbutton AP measurements.
Bottom line for me, don't find THD to be a very useful metric.
"...we know so little about how humans perceive the sound quality of an audio system, and in particular the loudspeaker, that one should question almost everything that we think we know about measuring it. From what we have found most of what is being done in this regard is naive."
http://www.gedlee.com/Papers/Comments on howard.pdf
He also said:
"This is precisely where the signal-based distortion metrics fail. In our next paper we will show that .01% THD of one type of nonlinear system can be perceived as unacceptable while 10% THD in another example is perceived as inaudible. Even one of these simple examples is sufficient to invalidate THD as a viable metric for discussion of the perception of distortion. Furthermore, 1% THD is not at all the same as 1% IM, but we will show that neither correlates with subjective perception."
http://www.gedlee.com/Papers/Distortion_AES_I.pdf
Then there is Bob Cordell's book on power amplifier design, 2nd ed., has a whole chapter (#16) dedicated to other sources of distortion, including EMI distortion. Some of that stuff doesn't show up very well on a full set of pushbutton AP measurements.
Bottom line for me, don't find THD to be a very useful metric.
Last edited:
I agree it’s not the magic number that tells all. That being said, THD+N can probably be used as a warning indicating “more review is needed”. With the understanding there will be false positives.
There's more than one way to calculate THD, too. One way only adds up the amplitude of the individual harmonics, but it's also possible to subtract the fundamental and look at the net effect of amplitude and phase.
What's funny is that it's possible for multiple harmonics to cancel out, so the area under the curve on a graph is close to zero. An example is an impulse, consisting of a broad spectrum of tones, which add together to produce a single tall spike.
Something else to ponder: what happens if a sound wave with a distortion spectrum that resembles tall narrow spikes, reaches our eardrums and sets them into motion, yet the unwanted high frequencies are outside of the range for setting the cochlea into resonance?
AFAIK, the eardrum has lots of sensory nerves on its own, as well as muscles. And the brain is part of the loop, controlling the tension dynamically.
So... When someone tells an anecdote about a 'bad' amplifier sounding fatiguing despite apparently having great THD ratings, I'm inclined to believe them.
What's funny is that it's possible for multiple harmonics to cancel out, so the area under the curve on a graph is close to zero. An example is an impulse, consisting of a broad spectrum of tones, which add together to produce a single tall spike.
Something else to ponder: what happens if a sound wave with a distortion spectrum that resembles tall narrow spikes, reaches our eardrums and sets them into motion, yet the unwanted high frequencies are outside of the range for setting the cochlea into resonance?
AFAIK, the eardrum has lots of sensory nerves on its own, as well as muscles. And the brain is part of the loop, controlling the tension dynamically.
So... When someone tells an anecdote about a 'bad' amplifier sounding fatiguing despite apparently having great THD ratings, I'm inclined to believe them.
Measurement’s like THD%, THD+N%, IMD, SR, BW, PS Rejection, Damping factor, FFT, Nonlinear loads and others are only tools that help you understand how well the amplifier performance is. Beauty is in the ears of the listener.
One of my jobs was with an MI company (Acoustic Control) and I made a new improved Guitar amplifier and it was Not Well Accepted, NOT MUSICAL. I had to add distortion (2nd, 3rd harmonics), change output Z and others to make the product GREAT.
Some of my other jobs like BGW, Harman/JBL and Audio Precision required greater total performance.
One of my best measurements in the toolkit is to listen to the ERROR OUTPUT signal.
Many of the responders in this form know that there is no one measurement that defines the BEST GREATEST AMPLIFIER #
If you like your amplifier then it is for you.
Duke
One of my jobs was with an MI company (Acoustic Control) and I made a new improved Guitar amplifier and it was Not Well Accepted, NOT MUSICAL. I had to add distortion (2nd, 3rd harmonics), change output Z and others to make the product GREAT.
Some of my other jobs like BGW, Harman/JBL and Audio Precision required greater total performance.
One of my best measurements in the toolkit is to listen to the ERROR OUTPUT signal.
Many of the responders in this form know that there is no one measurement that defines the BEST GREATEST AMPLIFIER #
If you like your amplifier then it is for you.
Duke
Hi Markw4
It is the notched distortion residual signal to a speaker or FFT analyzer.
If the signal sounds BAD/UNPLESENT/NONMUSICAL when steady, slow sweeping, or increasing/decreasing levels or reactive loads.
This has shown up in several class G/H designs @ the switching/transitions points.
Duke
It is the notched distortion residual signal to a speaker or FFT analyzer.
If the signal sounds BAD/UNPLESENT/NONMUSICAL when steady, slow sweeping, or increasing/decreasing levels or reactive loads.
This has shown up in several class G/H designs @ the switching/transitions points.
Duke
Agree. If 4nd and above harmonics distortion below 2nd and 3rd, high THD like 0.xxx% sound great. But I don't care about harmonic profile of the distortion is THD so small, example below 0.00x%. CCIF IMD is important, so THD at 20kHz is important. Usually, I try to make slew rate as high as possible (of course still stable) and then I can trade-off the slew rate to THD.
Edit:
Recently, I saw a glitch of class G when switching in my simulation. The glitch only seen when the signal frequency is 20kHz with low impedance load. I add inductor || resistor in series with low voltage PSU, the glitch was reduce significantly.
Last edited:
Mark, nice papers that you pointed to.
I very much liked:
It’s like the story of the cop who asks a drunk under a street light what he is doing on his hands and knee’s. The drunk replies “I’m looking for my car keys.” The officer asks “Where did you loose them?” and the drunk replies “Over there by my car.” Baffled, the officer asks “Then why are you looking for them here?” to which the drunk replies, “Because the light is better.” Everyone knows that THD is meaningless, but it’s easy to do and “the light is better.
As long as Stereophile finds almost every year a new amp that improves on the previous “Best amplifier ever heard”, while having figures going in all different directions, it should be obvious that we still have an incomplete image to what our ears are telling us.
Hans
Below a certain value, THD has no meaning. But the THD should not vary much within the audio band, remember that an instrument has a fundamental and overtones. If the the harmonics has higher distortion than the fundamental, the instrument sounds different. Attached is a simulation of my amplifier with and without global feedback. Even without feedback, the THD is too low to matter. But in reality the amplifier is used with feedback to reduce offset and output impedance.
Attachments
If you regress Stereophile's ratings with "retail cost" of the product, you will find a tight correlation! (with outliers like Schiit).
Its costs more to make a good class AB amp today than it did several years ago, no?
Other that there are usually clues in reviews that are used to advise readers when to be wary. Unfortunately, not everyone knows what to look for in way of such clues.
Other that there are usually clues in reviews that are used to advise readers when to be wary. Unfortunately, not everyone knows what to look for in way of such clues.
Last edited:
Have a closer look at their “Recommended Component “ list and you will see a large spead in prices per category, not exactly what you suggest.
A $2400,- NAD amp, a $3,300.- Benchmark and several others in this price range sitting in the same Class A as a $230,000.- Dartzeel amp, so where’s the correlation ?
Hans
That's right. Also, some of their category levels may not have any recommended components.
And the reviewers often differ significantly in their opinions. These are well-intentioned people.
But losing Fremer is a huge blow to them.
And the reviewers often differ significantly in their opinions. These are well-intentioned people.
But losing Fremer is a huge blow to them.
Last edited:
Rayma,
Yes, they are all humans at Stereophile, each with their own set of qualification factors.
That means that their opninions are not automatically overlapping yours.
But thinking “ the more expensive the better the test” is really beyond truth.
Hans
Yes, they are all humans at Stereophile, each with their own set of qualification factors.
That means that their opninions are not automatically overlapping yours.
But thinking “ the more expensive the better the test” is really beyond truth.
Hans
Even when JGH was running the mag, he would caution that you might not share his preferences.
So he usually described the sound, rather than just say how wonderful it was. But on a few occasions,
he would say he could find no fault with it (amplifiers), or that it was truly exceptional (speakers).
And he said this even for the Dynaco A-25 speaker, which cost well under $100. back then.
So he usually described the sound, rather than just say how wonderful it was. But on a few occasions,
he would say he could find no fault with it (amplifiers), or that it was truly exceptional (speakers).
And he said this even for the Dynaco A-25 speaker, which cost well under $100. back then.
So We have speakers with flapping diaphragms that produce masses of distortion.
The majority of recordings have gross distortion introduced at some point on the process.
We also have a litany of over produced digital recordings which are compressed and full of square wave clipping -
Those recordings often are bought - listened to and then relegated to storage - rarely to be listened to again. Artists wonderful music destroyed so that it sounds impactful for the first few minutes when heard on the radio.
In the light of this why do we continue to perceive distortion in audio equipment?
My own take on this is is that Distortion and Intemodulation Distortion may change the characteristic of any amplifier. But the overall output signal energy is not changed, so if a sideband is increased, then the funamental is decreased. Therefore as the distortion- IMD level increases, the dynamic clarity and pitch of the musical note is reduced.
We don't so much listen to distortion, as try to listen through it. To try and perceive the intonation and structure of music. That's why the little details are important.
The majority of recordings have gross distortion introduced at some point on the process.
We also have a litany of over produced digital recordings which are compressed and full of square wave clipping -
Those recordings often are bought - listened to and then relegated to storage - rarely to be listened to again. Artists wonderful music destroyed so that it sounds impactful for the first few minutes when heard on the radio.
In the light of this why do we continue to perceive distortion in audio equipment?
My own take on this is is that Distortion and Intemodulation Distortion may change the characteristic of any amplifier. But the overall output signal energy is not changed, so if a sideband is increased, then the funamental is decreased. Therefore as the distortion- IMD level increases, the dynamic clarity and pitch of the musical note is reduced.
We don't so much listen to distortion, as try to listen through it. To try and perceive the intonation and structure of music. That's why the little details are important.
Looks like a correlation to me, just assigning 2,5,10 values for B, A and A+ The R^2 is 0.74
