Power vs THD sweet spot

[SubOjectivist]

Hi golden-eared fellas,

From what I've noticed so far, amps usually have a sweet spot (understand the lowest value of THD at a given power).
Let's assume it's around 9W for an amp, but with my listening conditions (room volume, listening distance, model of speakers, pissed neighbors and so on), I find the volume too loud (and what if, on the other hand, the volume was too quiet for my liking).
Can I for instance lower the source volume and still operate the amp at 9W, with an overall lower volume? And still get the lowest THD for this amp?

Thanks.

 

[VintageFlanker]

Can I for instance lower the source volume and still operate the amp at 9W, with an overall lower volume? And still get the lowest THD for this amp?

Nope.

9W is 9W. It will result in the exact same SPL out of your speakers.

By lowering the source volume, you're adding noise, which leads to about the same THD+N at the end of the day.

Keep in mind this is a continuous tone, 0dBFS, (1Khz?). Music content behaves quite differently and you shouldn't notice these THD+N variations (related to SNR increase) in practical listening.

 

[SubOjectivist]

So, there is absolutely no way to kind of cheat to get the lowest possible THD without using the amp at the corresponding power?
Please don't tell me to use ear plugs .

 

[SubOjectivist]

Maybe by using a resistance on the speakers wires (-- or //) to artificially increase the load on the amp?

BTW Vintage, I knew you had golden ears, so no wonder you're the first to reply .

 

[ebslo]

Maybe by using a resistance on the speakers wires (-- or //) to artificially increase the load on the amp?

BTW Vintage, I knew you had golden ears, so no wonder you're the first to reply .

Series R would increase the source impedance seen by the speakers, which would affect tone due to frequency dependence of speaker impedance.
Parallel R would not affect voltage output of amp for given SPL, but would lower load impedance on the amp which usually increases distortion.

 

[SubOjectivist]

Series R would increase the source impedance seen by the speakers, which would affect tone due to frequency dependence of speaker impedance.
Parallel R would not affect voltage output of amp for given SPL, but would lower load impedance on the amp which usually increases distortion.

Thanks for your answer.
So there is no way to get what I would like to achieve?

 

[Beershaun]

Do you have power meters on your amp? Do you know what power your amp is outputting at any given time? It greatly depends on the music being played.

 

[SubOjectivist]

I have a Wattmeter, so I can get input power at the wall socket (and get a rough idea of output power if I assume the amp has a certain efficiency).
But the polling rate is definitely not high enough to consider it is real time data.

 

[Beershaun]

I'll repost this video as I think it does a really good job of demonstrating that the power demanded from an amplifier is much more variable than I think people realize and that the music has a very large impact on the power required at any given instant. So trying to chase a specific continous power level to hit the sweet spot is not really possible.

 

[Blumlein 88]

Need to clarify. Thd plus noise or thd?

Those downwards sloping curves are usually the influence of noise.

 

[RayDunzl]

From what I've noticed so far, amps usually have a sweet spot (understand the lowest value of THD at a given power).

I get to be wrong, but the downward sloping area of the graph represents the increase of the ratio of signal to noise (noise being "interpreted" as harmonic distortion).

Only at the upper limits do the actual harmonics (onset of clipping or other distress) cause the graph trace to rise.

(sorry, I missed Blumlein's equivalent observation immediately above)

 

[SubOjectivist]

Guys, you don't even let my poor exhausted brain rest on a saturday .

I'll repost this video as I think it does a really good job of demonstrating that the power demanded from an amplifier is much more variable than I think people realize and that the music has a very large impact on the power required at any given instant. So trying to chase a specific continous power level to hit the sweet spot is not really possible.

I have not checked out the video yet, but I guess it mentions dynamic, right?
Of course, I don't usually listen to 1 kHz monotonal sounds at a constant volume, so indeed in a song the acoustic volume usually varies quite a lot even when volume (software or hardware) is not whatsoever adjusted over the song duration. So the sweet spot I evoqued should rather be interpreted as an average (maybe median would be better, dunno) ideal power value.

Need to clarify. Thd plus noise or thd?

Those downwards sloping curves are usually the influence of noise.

Ideally, THD+N.
But I'm not sure whether the graph from Amir I've used in my initial post is just Power vs TDH or Power vs THD+N.


What makes an amp usually have poorer values at very low power levels than at intermediate ones? If well designed and built, it should not generate distorsion, right? So it's mainly only noise?

 

[abdo123]

So, there is absolutely no way to kind of cheat to get the lowest possible THD without using the amp at the corresponding power?
Please don't tell me to use ear plugs .

the question you should ask yourself is whether you would benefit from having the lowest possible THD. I personally cannot hear anything lower than 1% and that's in test tones.

 

[Beershaun]

Guys, you don't even let my poor exhausted brain rest on a saturday .


I have not checked out the video yet, but I guess it mentions dynamic, right?
Of course, I don't usually listen to 1 kHz monotonal sounds at a constant volume, so indeed in a song the acoustic volume usually varies quite a lot even when volume (software or hardware) is not whatsoever adjusted over the song duration. So the sweet spot I evoqued should rather be interpreted as an average (maybe median would be better, dunno) ideal power value.



Ideally, THD+N.
But I'm not sure whether the graph from Amir I've used in my initial post is just Power vs TDH or Power vs THD+N.


What makes an amp usually have poorer values at very low power levels than at intermediate ones? If well designed and built, it should not generate distorsion, right? So it's mainly only noise?

Watch the video and see the fluctuations in the power meter for both the very powerball hungry music track and the less power hungry audio track. You will see large swings in power both in terms of absolute power and multiples of power.

To answer your question about the distortion vs frequency graphs:
You are seeing the relationship between constant noise floor and increasing power creating that slope. The noise floor of the amp.is constant regardless of the power output. So as you increase the power output the power to noise ratio increases and the thd% decreases. The engineering challenge is to build an amp with both high power and low noise. If you look at the thd for many of Amirs headphone amps you will see many have extremely low noise. It's easier to accomplish this when you only need to produce less than 1 watt of power. It's a lot harder when you need to build something with components that produce 200watts and have it be dead silent when it's only delivering 1-10 watts of power.

 

[SubOjectivist]

the question you should ask yourself is whether you would benefit from having the lowest possible THD. I personally cannot hear anything lower than 1% and that's in test tones.

You can't physically hear it or you don't like the sound with very low THD (maybe you were used to tube amps )? But I suppose it's the former.
To be honest, perhaps I won't be able to hear THD at 1% maximum, but as long as it doesn't involve additional costs, why not aim for the best possible value?

 

[DonH56]

The noise floor of an amplifier is usually constant for all practical purposes. That means that, as power goes down, the relative noise contribution increases. Noise stays the same, signal goes down, so the signal-to-noise ratio (signal/noise) gets worse. It is not that there is more noise, just less signal, so the proportion of noise is larger when the signal gets smaller.

Distortion increases with increasing signal, but at the same time increasing the signal means more and more signal than noise, so you have a similar situation. At very low levels distortion is very small, but so is the signal, so the effect of the noise is to raise the curve since any noise falling on harmonics of the signal will be treated as harmonic distortion. As the signal rises, distortion increases, eventually dominating the curve.

Not to scale, illustrative only:

At low level the THD+N (SINAD -- signal to noise and distortion) curve is dominated by noise. As the signal level increases distortion rises and the curve begins to flatten. Near clipping the output distortion dominates. The actual shapes of the curves and where the turn from noise to distortion dominates is determined by the component's noise and distortion performance.

In general you won't hear distortion until the amplifier is near clipping and the signal is very loud. When the level is very low you will hear only noise and essentially no distortion. (I am ignoring crossover and some other types of distortion that flatten the distortion curve at low levels, and noise types that change with power level -- this is not a rigorous analysis, but illustrates the general trend close enough for most of us.)

Bottom line is set the volume where it sounds good and forget about the curves. As a very (very!) rough guide, if the sound is too noisy, you need an amp with lower noise floor. If it is distorted, you need an amplifier with more power.

HTH - Don

 

[SubOjectivist]

The noise floor of an amplifier is usually constant for all practical purposes. That means that, as power goes down, the relative noise contribution increases. Noise stays the same, signal goes down, so the signal-to-noise ration (signal/noise) gets worse. It is not that there is more noise, just less signal, so the proportion of noise is larger when the signal gets smaller.

Distortion increases with increasing signal, but at the same time increasing the signal means more and more signal than noise, so you have a similar situation. At very low levels distortion is very small, but so is the signal, so the effect of the noise is to raise the curve since any noise falling on harmonics of the signal will be treated as harmonic distortion.

Not to scale, illustrative only:

View attachment 126001

At low level the THD+N (SINAD -- signal to noise and distortion) curve is dominated by noise. As the signal level increases distortion rises and the curve begins to flatten. Near clipping the output distortion dominates. The actual shapes of the curves and where the turn from noise to distortion dominates is determined by the component's noise and distortion performance.

In general you won't hear distortion until the amplifier is near clipping and the signal is very loud. When the level is very low you will hear only noise and essentially no distortion. (I am ignoring crossover and some other types of distortion that flatten the distortion curve at low levels, and noise types that change with power level -- this is not a rigorous analysis, but illustrates the general trend close enough for most of us.)

Bottom line is set the volume where it sounds good and forget about the curves. As a very (very!) rough guide, if the sound is too noisy, you need an amp with lower noise floor. If it is distorted, you need an amplifier with more power.

HTH - Don

Very well explained and easy to understand (but so far, that's not very technical), thanks.

So at low volumes, it might be preferable to listen with good headphones and an headphone amp like the Topping L30 to get a better overall SINAD?

 

[lizhuoyin]

Very well explained and easy to understand (but so far, that's not very technical), thanks.

So at low volumes, it might be preferable to listen with good headphones and an headphone amp like the Topping L30 to get a better overall SINAD?

For me, low volume listening is for background. Headphone is totally different experience. For critical music listening, I prefer speakers with adequate volume. I don't think my ears or mind are sharp enough to tell difference. Just make sure no background noise (ears next to speakers) and no distortion at my volume. Just enjoy the music.

 

[Vini darko]

Want to use more power without increased volume? Buy less efficient speakers. Sorted

 

[SubOjectivist]

Want to use more power without increased volume? Buy less efficient speakers. Sorted

I hope you were being sarcastic, because I assumed people would understand I wanted a solution that didn't involve additional costs (or just minimal costs), but perhaps that wasn't obvious for everyone, my bad.
After all, I could also get a Benchmark AHB2 that produces very low THD+N, even at very low power levels. Sorted too, right?