Low THD figures from amplifiers are masked by high THD in speakers and IEMs???

[proplayer69]

Hello I’ve been seeing a lot of hype over amps with ultra low THD numbers like <0.001% THD+N as if that’s the holy grail of audio performance. But honestly, even amps with 0.05% or 0.1% THD are already excellent and practically transparent IMHO. I would rather pick an amplifier (my preference) with thousand watts of power continuous maintaining at atleast 0.1% thd+n to play with great dynamics rather than amps with 0.001% thd+n that can only handle about hundreds of watts and will get about 10% THD+N at <1k watt. Most speakers and even the most highest end IEMs already have >0.1% distortion.., so the amp with so called transparent very hifi signal will always get masked by the transducer's own distortion. We are chasing numbers that, in real-world listening, are well below what we can perceive.

To test this myself, I’ve actually done AB comparisons using a cheap $0.50 (20 php) Chinese USB dongle dac and a genuine Apple once which measures -101 dB THD+N or around 0.0009% with high-end like performance IEM <1% THD and lossless Apple Music (while switching with different genres). After hours of critical listening, I swear I couldn’t reliably tell them apart. No night and day difference, no magic fairy dust.

What’s the main reason for the obsession with super-low THD figures? Is it just the spec sheet chase or are people actually hearing something that I’m missing?

Till now im still contented with my cheap modified class d amp fosi audio v3, the switching freq. was too high heating the whole amp as if it was like a class AB amp, from 600kHz down to 250kHz, stone cold, no inductor ripple noise because i changed the LC filters to make cut-off frequency of 20kHz from 50kHz factory setting. And also, I bumped up the gain from 23dB (my actual measurement differed from spec'd 26dB) to ~38dB, still sounds clean but, after that gain change, a very subtle hiss became present on 8 ohms silk dome tweeter if you listen close enough..

 

[kyuu]

All else being equal, lower noise and distortion is a good thing. Of course, all else is never equal, but the prevailing opinion here seems to be that once you're into good-enough territory on SINAD that other factors are more important (power, features, price, build quality, etc.). I don't know what "hype" you're referring to, I don't see anyone going around saying that impressive SINAD numbers trump any other consideration.

So no, you won't hear an improvement going from audibly transparent to even more audibly transparent. But most people won't benefit from a 1kW amp in a domestic setup either, unless they have a very large listening room with incredibly low sensitivity speakers and the desire to listen at ear-splitting levels. And maybe not even then. I mean sure, excessive headroom doesn't hurt, but neither does excessively low noise/distortion.

 

[DVDdoug]

I don't even bother looking at distortion numbers for amplifiers... or any other electronics!

You rarely see a 10% distortion spec, but if you do it's usually to push the power spec as high as possible, rather than distortion under normal conditions when it's not clipping.

I've NEVER heard distortion from ANYTHING unless it was broken or overdriven.

But when you combine for THT+N or SINAD, then noise can be an issue. The problem with noise specs is that there are many ways to measure it. Of course the measurements Amir does are standardized and comparable. Even then, the audibility of the noise also depends on the sensitivity of your speakers/headphones, how close you are to the speakers, and other ambient noise in the room, etc.

I usually assume frequency response is flat too. With (analog) electronics, noise is the main thing I'm concerned with.

 

[Coltr@ne]

I'm inexperienced.
But I believe that if your speakers distort 1% with a clean signal, the more you increase the input distortion, the worse they will get. Wait for an expert to do the calculations.

 

[oleg87]

While Amir might rank amps according to such (dubiously useful) metrics in his reviews, I don't think anyone is really chasing those specs. Among the objectively inclined it's usually understood that 0.01% is approximately the limit of where distortion of any kind matters for any practical context, even for the most extraordinarily sensitive listeners.

That being said - transducers have gobs of primarily low-order distortion, but higher-order distortion is what's most audible.

 

[wwenze]

Pretty much the general concensus. However this is only true for SINAD components that are comparable to each other. Noise from amplifier shows up easily on speakers, which by definition don't produce noise.

Similarly if you have an amplifier that rings or do mic feedback then that sound is not the speaker's fault either.

As for distortions from linearity issues, speakers are the main component.

 

[TurtlePaul]

Chasing SINAD is useless. But the corollary argument is that a $80 DAC, $100 headphone amp or $250 power amp has 100 dB SINAD these days, so why accept anything else.

 

[valerianf]

Let me give a real life example.
I just made the transition from a Yamaha RX-A700 to a Onkyo TX-RZ30.
DAC Yamaha --> TI PCM 1681 S/N: 105 dB dynamic range: 105 dB THD: -94 dB
DAC Onkyo --> TI PCM 1690 S/N: 113 dB dynamic range: 113 dB THD: -93 dB

Well, with the same speakers there is an obvious difference between the two AVRs.
The Onkyo has far more dynamic.
But the DAC is only one piece of the puzzle: the amp and DSP sections are also contributing.
Onkyo example: if the tone corrections (bass, treble) are not at zero, then you can hear at several meters a permanent noise coming from FL and FR.
The good S/N is gone because of an engineering weakness!

 

[proplayer69]

I'm inexperienced.
But I believe that if your speakers distort 1% with a clean signal, the more you increase the input distortion, the worse they will get. Wait for an expert to do the calculations.

You’ve got a point. However, I think distortion generation in mechanical parts (like speakers) is not a linear process. I believe it works similarly to how distortion progresses in THD measurements in amplifiers starting with little to no distortion, then suddenly spiking into heavy distortion once a threshold (electrical limitations in amplifiers [max voltage], mechanical in speakers [diaphragm breakup]) is crossed. Is there a shared principle or law that both systems follow? Correct me if I’m wrong, these are just assumptions.;/}

 

[Holmz]

Hello I’ve been seeing a lot of hype over amps with ultra low THD numbers like <0.001% THD+N as if that’s the holy grail of audio performance. But honestly, even amps with 0.05% or 0.1% THD are already excellent and practically transparent IMHO. I would rather pick an amplifier (my preference) with thousand watts of power continuous maintaining at atleast 0.1% thd+n to play with great dynamics rather than amps with 0.001% thd+n that can only handle about hundreds of watts and will get about 10% THD+N at <1k watt. Most speakers and even the most highest end IEMs already have >0.1% distortion.., so the amp with so called transparent very hifi signal will always get masked by the transducer's own distortion. We are chasing numbers that, in real-world listening, are well below what we can perceive.

To test this myself, I’ve actually done AB comparisons using a cheap $0.50 (20 php) Chinese USB dongle dac and a genuine Apple once which measures -101 dB THD+N or around 0.0009% with high-end like performance IEM <1% THD and lossless Apple Music (while switching with different genres). After hours of critical listening, I swear I couldn’t reliably tell them apart. No night and day difference, no magic fairy dust.

What’s the main reason for the obsession with super-low THD figures? Is it just the spec sheet chase or are people actually hearing something that I’m missing?

Till now im still contented with my cheap modified class d amp fosi audio v3, the switching freq. was too high heating the whole amp as if it was like a class AB amp, from 600kHz down to 250kHz, stone cold, no inductor ripple noise because i changed the LC filters to make cut-off frequency of 20kHz from 50kHz factory setting. And also, I bumped up the gain from 23dB (my actual measurement differed from spec'd 26dB) to ~38dB, still sounds clean but, after that gain change, a very subtle hiss became present on 8 ohms silk dome tweeter if you listen close enough..

Other than being likely totally off base with the need for kWs of power, you sort of hit the nail on the head.

IEMs and Speakers can be had in low distortion, which generally cost more.
On the other hand, most people either cannot hear the distortion (Harmonic Distortion), or they tend to like it.

Some experts will tell you to spend some percentages on this and that from speakers to cables.
Having the speakers be the majority of the expense, usually results in not replacing them very quickly.
But there is a bit more to a speaker than solely the distortion.

 

[Mnyb]

I've read somewhere but i have no quotable reference to this . That Genelec for example consider THD figure 1/10 of the drivers THD good enough for thier active speakers .
Noise of course cant be to low .

The devil is in the details , distorsion can have different characteristics etc , so better keep it very low so that the nature of the distorsion becomes totally irrelevant .
There are lot of discussion how different distorsions sounds like for better or worse ,why not just keep it magnitudes lower than human perception so it does not matter

Luckily for us it's not rocket science and don't cost that much . I consider electronics without vanishingly low thd or sinad figures incompetently designed .
So some of these figures can bee seen as proxy for a well executed design .
And they usually go together , well designed feedback for example flattens the fr response lower output impedance and lowers THD at the same time .
I'm certainly not an amp expert , there probably more detail . But i bet the basic nonlinear mechanism inside amps ( whatever they may bee ) makes it hard to just improve one aspect. And its also a matter of magnitude a very "good" small amp will be bested by a rather "crappy" big amp if it cant drive the speakers or overheat or something.

 

[Mnyb]

And its important with sites like ASR that holds manufacturers accountable and measure this stuff even if its should be a no problem these days !

High end Audio is filled with self taught gurus and "know it alls" that internalized all the folklore and builds quite terrible and expensive products .

Instead of doing the actual work of going to university ( or by yourself ) and study the textbooks that exists and slowly build up actual knowledge and verify your work by for example measurements .
Even if it's not rocket science its actuall hard work . It's known quantities , the usual standing on the shoulders of giants thing. It was rocket science when someone originally explored the unknown and figured it out for us to learn.

Total DAC or Audio GD can be interesting to search on ASR

 

[mtcn77]

I have always heard distortion at track ends when my IEMs are set very loud and I hated it. I think it causes long term tinnitus that you can feel.
I recently changed the earplugs to a smaller included size and it worked out great. I no longer need to set volume so high since they seal better. My ear canal was indeed narrow hence they plugged nicely. They even stopped falling out from my ear when I'm busy.
While not as important as IEMs, speakers might also sound clearer at the high end with a lower THD, you never know until having tried it.

 

[kemmler3D]

Is there a shared principle or law that both systems follow?

There are multiple sources of distortion in a speaker, some are basically linear (I think Doppler distortion qualifies) and some are very non-linear (breakup modes). There are a lot more than just these two!

 

[Hiten]

Hello I’ve been seeing a lot of hype over amps with ultra low THD numbers like <0.001% THD+N as if that’s the holy grail of audio performance.

These mentioned distortion probably is at 1khz. At full power (required at transients)and at both end of frequency they rise. Noise at low power affects signal to noise ratio. i.e. if not controlled it may get amplified.
I am newbie so i hope i am correct.