What exactly affects an amplifier's "dynamic performance"?

[DNCAgain]

I recently came across a viewpoint suggesting that Topping amplifiers and early KEF speakers may have overly prioritized low distortion measurements at the expense of "dynamic range" performance. (This is perceived as bass that lacks depth and impact, and treble that sounds constrained or fails to soar.) This is a quite common discussion point among audiophiles.

Here's a comment from one audiophile regarding a specific amplifier:

"The d900 doesn't perform well with complex, high-dynamic-range music featuring dense drum passages and high-frequency instruments." (Unfortunately, he didn't specify the reasons.)

So, what exactly is dynamic range, and what metrics are used to measure it?

They provided the following reference as a basis for this discussion (my English isn't strong enough to fully comprehend it):

 

[DNCAgain]

the document

 

[solderdude]

The problem here is that an amplifier is an electronic device and not an instrument.
What it needs to do is change its output voltage linear in a defined gain (volume control dependent).
This means that whenever you do not turn the volume all the way up the output voltage will have peaks that won't be high.
Only at maximum power the output voltage will be high (test signals at full output voltage/power).

For electronics the dV/dt (V/time unit) is the measure of how fast the output voltage can change its value.
This is called the rise-time and fall-time. When the amp does not have an input filter that rise- and fall-time may even differ in practice.
With an input filter that max time is limited by the filter which prevents TIM distortion (the output stage not following the input signal and 'clipping' internally for that brief moment.

Here's the thing... audio consists of sine-waves, even a square-wave consists of a whole bunch of sine waves.
Then there is another thing to consider which is the 'energy' in the treble in music is MUCH lower (20dB or so on average) meaning for music the signals the rise time may even shorter than what would be needed for 20kHz square waves. The latter is a test signal. It does not exist in any recordings.

This basically means that when an amplifier (under load) can reach say ... 30kHz (-0.5dB) at max output power (called power bandwidth) the amplifier is fast enough for all audible frequencies, certainly at normal listening levels and even at high SPL it will be more than enough.

Now... there is another 'dynamic range' which is the word used in 'audio' which is not about the same issue but most people believe they are related.
This is the difference between the loudest and barely audible sounds.
That differs depending on the listening volume.

How 'fast' an instrument sounds is determined by the ratio between the lower frequencies (fundamentals) and its harmonics and is sometime misinterpreted as 'one having a bigger dynamic sound' than another that has a different ratio (speaker or recording thingy).
Its a perception thing and not an electronics thing. The electronics just follow the input voltage. This is measurable in extreme accuracy, far below and beyond human perception.

Then we have the transducer. These are only 'linear' at low excursions (low to normal listening levels) at higher excursions they become less linear, add harmonics at that point and may compress.
The electronics don't care, they just keep delivering. The speaker does not follow the, by the amp applied, output voltage.
This basically means that if one measures the amplifier under load and it performs well everything that sounds 'wrong' is caused by either the transducer (speaker/headphone) or perception (the brain) that interprets an artifact of the transducers or recording differently. Not caused by the amplifier.

In practice (headphone amps that could be clipping or have a high output resistance) there can be interaction between the headphone and amp that are not directly visible in measurements under load, simply because the load varies.
With speaker amps the back EMF of a speaker could throw some dirt in the mix, so will a higher output resistance (think tube amp).
This usually alters the frequency response and thus the balance between fundamentals and overtones (harmonics).

This can all be measured and explained but requires in situ measurements and characterization of the whole audio chain... NOT going to happen .... ever.

So yes, dynamic signals differ from test signals at different levels. Transducers usually are not linear(enough) at higher SPL (and thus also peaks).
Then, in the mix we throw perception (including bias/convictions) and we have a toxic mix of acoustic and maybe, sometimes, some electronics not performing within parameters/specs.

Not so easy to answer.... electronics behave predictable and pre-determind. Transducers stop behaving linear above a certain level (and are not linear in response anyway) and perform much, much worse than electronics. And the final and most variable factor is perception.

The latter is what the mentioned article is about ... transducers and acoustics as well as perception.

 

[Keith_W]

That document is about LINEAR PHASE LOUDSPEAKERS. It says so in the title. It has nothing to do with "dynamic performance" of amplifiers! It's like asking a car dealer why his car is better than the other car you are considering, and he hands you a book on medieval archery and tells you to read that.

Dynamic range (DR) is simply the difference between the minimum and maximum a given system is capable of, measured by a relevant metric. For example, DR in camera sensors measures the brightest and darkest the sensor can capture, measured in stops of light. DR in TV panels is the brightest and darkest a display is capable of, measured in candelas/sq meter (or nits). With DAC's, the DR is the loudest and softest the DAC is able to output. In theory, a 16-bit DAC has 96dB of dynamic range, but some very poor DAC's have high noise floors and are incapable of the full 96dB. With loudspeakers, the DR is rather difficult to measure. All loudspeakers (even broken ones) are capable of outputting no sound, the question is what the maximum SPL a given loudspeaker is capable of and how much distortion you are willing to tolerate.

When it comes to amplifiers, the dynamic range is the maximum output an amp is capable of before onset of distortion, compared to the amplifier noise floor (the minimum output). This is measured on a test bench with a dummy speaker load.

When it comes to listening, the maximum is "how loud can my speakers go before clipping?" and the minimum is the room's noise floor (this is typically about 30-40dB at home, maybe up to 55-60dB for offices). A number of factors are at play here: the speaker's maximum SPL, whether the amp has adequate power or not, listening distance from the speaker, etc. In short, it's an uncontrolled environment and you can not simply blame the amplifier for lack of dynamic range. Maybe the amp is the culprit, maybe not.

 

[Spkrdctr]

Wow! Solderdude and Keith gave great answers. I have nothing to add. Well, except that the audiophiles are wrong as usual. Listening to electronics through speakers is crazy. The speakers dominate what you hear. But I really was impressed with two great answers to the audiophile BS.

 

[DNCAgain]

This is exactly what they mean.But they claimed that some amplifiers will weaken the dynamic range no matter how big the listening volume is,which I think may be nonsense.

This is the difference between the loudest and barely audible sounds.
That differs depending on the listening volume.

 

[Spkrdctr]

This is exactly what they mean.But they claimed that some amplifiers will weaken the dynamic range no matter how big the listening volume is,which I think may be nonsense.

That is nonsense. Unless the amp is designed by a 5 year old. If you set the volume at 25% then any anp will provide relatively distortion free sound with peaks as big as required for the music. Any amp turned up to 11 will or can cause distortion in your speakers as it clips. Up until clipping you have full dynamic range. Unless I am missing something here or not understanding the question.

 

[DNCAgain]

That document is about LINEAR PHASE LOUDSPEAKERS. It says so in the title. It has nothing to do with "dynamic performance" of amplifiers! It's like asking a car dealer why his car is better than the other car you are considering, and he hands you a book on medieval archery and tells you to read that.

Dynamic range (DR) is simply the difference between the minimum and maximum a given system is capable of, measured by a relevant metric. For example, DR in camera sensors measures the brightest and darkest the sensor can capture, measured in stops of light. DR in TV panels is the brightest and darkest a display is capable of, measured in candelas/sq meter (or nits). With DAC's, the DR is the loudest and softest the DAC is able to output. In theory, a 16-bit DAC has 96dB of dynamic range, but some very poor DAC's have high noise floors and are incapable of the full 96dB. With loudspeakers, the DR is rather difficult to measure. All loudspeakers (even broken ones) are capable of outputting no sound, the question is what the maximum SPL a given loudspeaker is capable of and how much distortion you are willing to tolerate.

When it comes to amplifiers, the dynamic range is the maximum output an amp is capable of before onset of distortion, compared to the amplifier noise floor (the minimum output). This is measured on a test bench with a dummy speaker load.

When it comes to listening, the maximum is "how loud can my speakers go before clipping?" and the minimum is the room's noise floor (this is typically about 30-40dB at home, maybe up to 55-60dB for offices). A number of factors are at play here: the speaker's maximum SPL, whether the amp has adequate power or not, listening distance from the speaker, etc. In short, it's an uncontrolled environment and you can not simply blame the amplifier for lack of dynamic range. Maybe the amp is the culprit, maybe not.

That document is about LINEAR PHASE LOUDSPEAKERS. It says so in the title. It has nothing to do with "dynamic performance" of amplifiers! It's like asking a car dealer why his car is better than the other car you are considering, and he hands you a book on medieval archery and tells you to read that.

Dynamic range (DR) is simply the difference between the minimum and maximum a given system is capable of, measured by a relevant metric. For example, DR in camera sensors measures the brightest and darkest the sensor can capture, measured in stops of light. DR in TV panels is the brightest and darkest a display is capable of, measured in candelas/sq meter (or nits). With DAC's, the DR is the loudest and softest the DAC is able to output. In theory, a 16-bit DAC has 96dB of dynamic range, but some very poor DAC's have high noise floors and are incapable of the full 96dB. With loudspeakers, the DR is rather difficult to measure. All loudspeakers (even broken ones) are capable of outputting no sound, the question is what the maximum SPL a given loudspeaker is capable of and how much distortion you are willing to tolerate.

When it comes to amplifiers, the dynamic range is the maximum output an amp is capable of before onset of distortion, compared to the amplifier noise floor (the minimum output). This is measured on a test bench with a dummy speaker load.

When it comes to listening, the maximum is "how loud can my speakers go before clipping?" and the minimum is the room's noise floor (this is typically about 30-40dB at home, maybe up to 55-60dB for offices). A number of factors are at play here: the speaker's maximum SPL, whether the amp has adequate power or not, listening distance from the speaker, etc. In short, it's an uncontrolled environment and you can not simply blame the amplifier for lack of dynamic range. Maybe the amp is the culprit, maybe not.

What’s interesting is that these audiophiles seem to define “dynamics” as the difference between the loudest and softest parts in a piece of music, regardless of the actual volume level. And they believe that some headphone amplifiers can compress this range (again, regardless of volume).

 

[DNCAgain]

That is nonsense. Unless the amp is designed by a 5 year old. If you set the volume at 25% then any anp will provide relatively distortion free sound with peaks as big as required for the music. Any amp turned up to 11 will or can cause distortion in your speakers as it clips. Up until clipping you have full dynamic range. Unless I am missing something here or not understanding the question.

I got it.Before reaching the maximum capacity of its circuit design, an amplifier does not affect dynamic range.

 

[solderdude]

What’s interesting is that these audiophiles seem to define “dynamics” as the difference between the loudest and softest parts in a piece of music, regardless of the actual volume level. And they believe that some headphone amplifiers can compress this range (again, regardless of volume).

Well you used the correct word .... believe ...
They should prove that this is so.
There is nothing seen in measurements that such is the case.
The onus is on them to show it with properly controlled and statistically valid blind tests... not going to happen anyway.
As usual we 'have to take their word for it'.

 

[MaxwellsEq]

What’s interesting is that these audiophiles seem to define “dynamics” as the difference between the loudest and softest parts in a piece of music, regardless of the actual volume level. And they believe that some headphone amplifiers can compress this range (again, regardless of volume).

Unless a comparison between two devices is level-matched to 0.01V, then descriptions are meaningless. It's been fully understood for decades that one device can sound "more exciting" or "more dynamic" than another if the levels are not exactly matched, even if the devices sound as if the levels are matched. Most often, the slightly higher voltage device sounds more dynamic and the slightly lower voltage device is described as flat or dull or unexciting.

 

[DNCAgain]

Unless a comparison between two devices is level-matched to 0.01V, then descriptions are meaningless. It's been fully understood for decades that one device can sound "more exciting" or "more dynamic" than another if the levels are not exactly matched, even if the devices sound as if the levels are matched. Most often, the slightly higher voltage device sounds more dynamic and the slightly lower voltage device is described as flat or dull or unexciting.

That's clear enough.Thanks!

 

[Spkrdctr]

Solderdude, audiopiles are the ultimate "Just trust me bro" kind of guys. That is all the science they put into it. I am always amazed.

 

[egellings]

I recently came across a viewpoint suggesting that Topping amplifiers and early KEF speakers may have overly prioritized low distortion measurements at the expense of "dynamic range" performance. (This is perceived as bass that lacks depth and impact, and treble that sounds constrained or fails to soar.) This is a quite common discussion point among audiophiles.

Here's a comment from one audiophile regarding a specific amplifier:

"The d900 doesn't perform well with complex, high-dynamic-range music featuring dense drum passages and high-frequency instruments." (Unfortunately, he didn't specify the reasons.)

So, what exactly is dynamic range, and what metrics are used to measure it?

They provided the following reference as a basis for this discussion (my English isn't strong enough to fully comprehend it):

Dynamic range is simply the difference between the softest and the loudest sound an amplifier can reproduce. The softest sound is limited by the amplifier's noise floor, and and the loudest sound is limited by the amplifier's maximum power output just before the onset of clipping or reaching the maximum permissible specified distortion level.

 

[nanook]

There's one thing in speakers (besides others already mentioned), that indeed has an effect on the dynamic behavior of a speaker at very high volumes.
During loud passages, the voice coil heats up and its resistance increases. Since the usual amplifier is a voltage source the current through the voice coil fades when it heats up. The magnetic field of the voice coil - and thus force - is proportional to the current.

This "thermal compression" may also be noticeable in the tweeter during a burst of heavy treble content since the time constant is short due to the small thermal mass.

We once used an amplifier that drives a proportional current through the tweeter of an active speaker.
This is only possible when the chassis has a flat impedance far beyond its operating frequency range otherwise you would screw up everything because Qes (the electrical damping) gets lost.
The tweeter in that system was an Air-Motion-Transformer with a completely flat impedance way beyond the audio band.

https://en.wikipedia.org/wiki/Power_compression?wprov=sfla1

Modern active speakers keep track of the voice coil temperature in order to protect the chassis. Here the DSP could in principle compensate for this effect - no idea if that's implemented in a commercial speaker.

 

[MatrixS2000]

What’s interesting is that these audiophiles seem to define “dynamics” as the difference between the loudest and softest parts in a piece of music, regardless of the actual volume level. And they believe that some headphone amplifiers can compress this range (again, regardless of volume).

DR is file locked into a music file. If your volume is set to 50%, then the loudness reduction impacts the entire file equally thus maintaining the DR.

 

[Chanur]

The problem here is that an amplifier is an electronic device and not an instrument.
What it needs to do is change its output voltage linear in a defined gain (volume control dependent).
This means that whenever you do not turn the volume all the way up the output voltage will have peaks that won't be high.
Only at maximum power the output voltage will be high (test signals at full output voltage/power).

For electronics the dV/dt (V/time unit) is the measure of how fast the output voltage can change its value.
This is called the rise-time and fall-time. When the amp does not have an input filter that rise- and fall-time may even differ in practice.
With an input filter that max time is limited by the filter which prevents TIM distortion (the output stage not following the input signal and 'clipping' internally for that brief moment.

Here's the thing... audio consists of sine-waves, even a square-wave consists of a whole bunch of sine waves.
Then there is another thing to consider which is the 'energy' in the treble in music is MUCH lower (20dB or so on average) meaning for music the signals the rise time may even shorter than what would be needed for 20kHz square waves. The latter is a test signal. It does not exist in any recordings.

This basically means that when an amplifier (under load) can reach say ... 30kHz (-0.5dB) at max output power (called power bandwidth) the amplifier is fast enough for all audible frequencies, certainly at normal listening levels and even at high SPL it will be more than enough.

Now... there is another 'dynamic range' which is the word used in 'audio' which is not about the same issue but most people believe they are related.
This is the difference between the loudest and barely audible sounds.
That differs depending on the listening volume.

How 'fast' an instrument sounds is determined by the ratio between the lower frequencies (fundamentals) and its harmonics and is sometime misinterpreted as 'one having a bigger dynamic sound' than another that has a different ratio (speaker or recording thingy).
Its a perception thing and not an electronics thing. The electronics just follow the input voltage. This is measurable in extreme accuracy, far below and beyond human perception.

Then we have the transducer. These are only 'linear' at low excursions (low to normal listening levels) at higher excursions they become less linear, add harmonics at that point and may compress.
The electronics don't care, they just keep delivering. The speaker does not follow the, by the amp applied, output voltage.
This basically means that if one measures the amplifier under load and it performs well everything that sounds 'wrong' is caused by either the transducer (speaker/headphone) or perception (the brain) that interprets an artifact of the transducers or recording differently. Not caused by the amplifier.

In practice (headphone amps that could be clipping or have a high output resistance) there can be interaction between the headphone and amp that are not directly visible in measurements under load, simply because the load varies.
With speaker amps the back EMF of a speaker could throw some dirt in the mix, so will a higher output resistance (think tube amp).
This usually alters the frequency response and thus the balance between fundamentals and overtones (harmonics).

This can all be measured and explained but requires in situ measurements and characterization of the whole audio chain... NOT going to happen .... ever.

So yes, dynamic signals differ from test signals at different levels. Transducers usually are not linear(enough) at higher SPL (and thus also peaks).
Then, in the mix we throw perception (including bias/convictions) and we have a toxic mix of acoustic and maybe, sometimes, some electronics not performing within parameters/specs.

Not so easy to answer.... electronics behave predictable and pre-determind. Transducers stop behaving linear above a certain level (and are not linear in response anyway) and perform much, much worse than electronics. And the final and most variable factor is perception.

The latter is what the mentioned article is about ... transducers and acoustics as well as perception.

Thank you for making this understandable for me.

I have a follow up maybe related questions. Are transients a real thing in audio , or is it one of those made up audiophile things ? Are they measurable in testing? That was Robb Watts claim to fame for the M Scaler, which as we know showed no frequency change.

 

[solderdude]

In electronics there is dV/dt (or rise-time and fall-time).
It basically states how fast a signal can go from 0 to max (well technically from 10%... to 90%) in either direction when a 'step' signal is applied.

Transients in music are something very different. How 'fast' and instrument or sound is, is mostly determined by upper-mid frequencies. (2-4kHz or so).
Increase that part of the frequency range and the music sounds 'faster'.
These 'transients' are incredibly 'slow' compared to the dV/dt of any amp.

Rob Watts makes the boldest claims which I would like to see him demonstrate under controlled conditions.
He has a company to run and making claims as he does helps sell equipment.
That's my standpoint.

 

[nanook]

Yes, I agree - the term "transients" is used differently in technical contexts compared to talking about music and perception.

In some pieces, there are sudden changes in volume that can really startle you—even if you know the piece—especially when they carry a lot of energy in the upper midrange. When it comes to music, that’s what I consider transients.

As @solderdude writes, these are “boring” for an amplifier as long as the amplifier doesn’t clip—they usually don’t even require a high slew rate from it.

“Beauty is in the eye of the beholder”—which brings us back to subjective perception and any preconceived expectations.

 

[levimax]

Thank you for making this understandable for me.

I have a follow up maybe related questions. Are transients a real thing in audio , or is it one of those made up audiophile things ? Are they measurable in testing? That was Robb Watts claim to fame for the M Scaler, which as we know showed no frequency change.

Dynamic contrast or Impact or Slam (there are a lot of words for the same thing) for electronics is dV/dt as mentioned and for speakers it is dP/dt dP= change in pressure. While an audio amplifier is not going to have an effect on this there is something that will which is "filters", either electronic DSP filters or DSP crossover filters or passive crossover filters or even the interaction between the speakers and the room which acts as a kind of filter . Every filter has a "cost" which is it changes the "time domain" behavior of the signal which always results is a larger "dt" (change in time) which does reduce the Dynamic Contrast. These filter effects only become audible, if at all, on sharp transient signals like a kick drum and not on a steady signal like a pipe organ. The audibility of these effects is argued even among scientists with most of the testing saying they are at best subtle and in many cases inaudible, however, the science is not really settled and there is some research like you linked to that says time domain effects of filters can be audible. This is not a "black and white" issue and filters in the signal chain should always be used carefully to avoid as much as possible any negative effects.