Amplifiers bench tests results-should we care or not, regardless of audibility, are those thresholds such as -100, -80, -40 dB THD+N, IMD etc.useful?

[AcuDefTechGuy]

Hahaha, well you got the company right! I was looking on their website and all they say is "<0.02%"

I need some of those Class D amps!

Gotta look at the measurements on ASR or AH since we are comparing the measurements, not the company specs.

The Denon X3700 is 0.001% on ASR, but you’ll see something like 0.05% on the Denon website.

 

[highfigh]

Gotta look at the measurements on ASR or AH since we are comparing the measurements, not the company specs.

The Denon X3700 is 0.001% on ASR, but you’ll see something like 0.05% on the Denon website.

The FTC requires a representative sample of a manufacturer's products to meet/beat the stated specs and if they only build to the spec sheet, there's a great chance that if the FTC does a spot inspection or test, something will fail. If they build better than the specs, they're far better off because they won't have to stop production or recall products.

 

[PENG]

The amps I want are only 0.02% THD......what should i do???

You might want to find out why it wasn't 0.002%. Or what kind of trade off did they made? For example, was it because they stuck to the negative feedback is bad belief/principle? Or their marketing/finance department have much more say than their design/engineering department?

 

[highfigh]

You might want to find out why it wasn't 0.002%. Or what kind of trade off did they made? For example, was it because they stuck to the negative feedback is bad belief/principle? Or their marketing/finance department have much more say than their design/engineering department?

OK, let's deal with this sensibly- THD is measured at half to full power and we don't usually listen at that level so, again, it's purely academic!

The marketing dept has more clout because they use words against people who are impressionable, partly because they don't understand what they're reading. Audio and Video magazines are very much to blame for this- they blather on about chocolaty midrange and other BS in the reviews without a care in the world, other than how many copies were sold. They also print the best reviews that money can buy.

 

[PENG]

OK, let's deal with this sensibly- THD is measured at half to full power and we don't usually listen at that level so, again, it's purely academic!

For this thread, we are talking about 20-20,000 Hz, worst conditions. That should therefore include power output not only at full power, or half power, but down to well below 1 W if possible, or at least to 500 mW to 1 W. There are lots arguments (valid to some extent) about the importance of the first watt. I don't buy some of those arguments but like most things, there are some truth to such arguments/suggestions. So yes I do think it is a good idea to consider measurements at low output level such as below 1 W. Measuring THD at half to full power is not my cup of tea but of course ymmv.

Fully agree with you on the marketing dept clout. That is often true in manufacturing in general, not just in audio.

 

[PENG]

If the music has pure tones from specific instruments like pipe organ or synths, IM is a bad thing when two frequencies are played and the amp runs out of headroom- the sum & difference frequencies can be audible and those are far from musical, unless it's a song by Yoko Ono, Bjork, Erik Dolphy, etc.

Pure tones or not, music typically will have multiple tones. Any waveforms can be represented as an infinite series of sine waves anyway. In fact, another reason why I would aim for the lowest possible harmonic distortions is that as soon as you have harmonic distortions, you know you will get intermodulation distortions, it is only of matter of how much.

There are numerous articles on IMD, I read tons of them and did come across a very non-technical one for those (not you) who are interested but don't want to their brains to hurt too much in trying to understand something about it.

Understanding Intermodulation Distortion - The Science of Sound (science-of-sound.net)

 

[cpp]

I tell ya, I just love listening to music with my old ears. Measurements although nice, I can't hear like I used to.

A cool thread though.

 

[highfigh]

For this thread, we are talking about 20-20,000 Hz, worst conditions. That should therefore include power output not only at full power, or half power, but down to well below 1 W if possible, or at least to 500 mW to 1 W. There are lots arguments (valid to some extent) about the importance of the first watt. I don't buy some of those arguments but like most things, there are some truth to such arguments/suggestions. So yes I do think it is a good idea to consider measurements at low output level such as below 1 W. Measuring THD at half to full power is not my cup of tea but of course ymmv.

Fully agree with you on the marketing dept clout. That is often true in manufacturing in general, not just in audio.

Why are you writing about frequency response when the thread is about THD+N? Nobody is gonna hear .1% when the output is half a Watt if the speakers are moderately sensitive. Period. If they say they can, they're full of crap. .1% means the THD is -30dB down and that will definitely be masked by whatever is coming from the source.

If an amplifier can't pass 20-20KHz without distortion, it's a pretty crappy amplifier. I posted about a couple of mine that weren't considered 'special', but still showed DC-100KHz from one and 10-100KHz for the other (8 Ohms, .02% THD). Seeing amplifiers that sell for big money and can't do better disappoints me.

 

[PENG]

Why are you writing about frequency response when the thread is about THD+N? Nobody is gonna hear .1% when the output is half a Watt if the speakers are moderately sensitive. Period. If they say they can, they're full of crap. .1% means the THD is -30dB down and that will definitely be masked by whatever is coming from the source.

If an amplifier can't pass 20-20KHz without distortion, it's a pretty crappy amplifier. I posted about a couple of mine that weren't considered 'special', but still showed DC-100KHz from one and 10-100KHz for the other (8 Ohms, .02% THD). Seeing amplifiers that sell for big money and can't do better disappoints me.

You seem ro have misread or misinterpreted my posts.

 

[PENG]

Why are you writing about frequency response when the thread is about THD+N?

Not writing about frequency response itself, but the distortions within the audible frequency range 20-20,000 Hz.

Nobody is gonna hear .1% when the output is half a Watt if the speakers are moderately sensitive. Period. If they say they can, they're full of crap. .1% means the THD is -30dB down and that will definitely be masked by whatever is coming from the source.

That's to you, and me too so no argument from me. By the way, 0.1% THD is not -30 dB down, but -60 dB down below the fundamental!

But try convincing guys like @RichB , Nelson Pass and John Curl..
And as I said, to some extent they may be right to some extent, at least in some cases, I certainly would call it "full of crap" lol..

I can easily think of examples that would indicate it isn't a bad idea to consider the distortions better than 0.1% at output level below 1 watt, even in a relatively quiet room.

You mentioned speakers that are moderately sensitive, but there are many popular speakers that are quite sensitive.

Try 96 dB/2.83v/m, 8 ohm nominal:

with two in stereo mode, mlp 8 ft, you get 88 dB spl, so 0.1% may be fine but 0.5% is -46 dB, so harmonics will be at 42 dB average that could be quite audible to trained ears. 96 dB/2.83v/m may be an extreme example but even with 88 dB/2.83V/m speakers, 0.5% THD could still bother some people in a quiet room. Even under more realistic conditions, such as 0.1% THD, 88 dB/2.83V/m speakers, that would still make distortions audible to those with sensitive and trained ears, again I would use people such as @RichB as example because he said he could hear the difference between amps like the ATI class D amp and the Benchmark AHB2 amp.

To add to the above, as you mentioned, IMDs are "bad", so aiming for the lower THD has the additional benefit that it will also increase the chance of minimizing IM distortions.

 

[highfigh]

Not writing about frequency response itself, but the distortions within the audible frequency range 20-20,000 Hz.



That's to you, and me too so no argument from me. By the way, 0.1% THD is not -30 dB down, but -60 dB down below the fundamental!

But try convincing guys like @RichB , Nelson Pass and John Curl..
And as I said, to some extent they may be right to some extent, at least in some cases, I certainly would call it "full of crap" lol..

I can easily think of examples that would indicate it isn't a bad idea to consider the distortions better than 0.1% at output level below 1 watt, even in a relatively quiet room.

You mentioned speakers that are moderately sensitive, but there are many popular speakers that are quite sensitive.

Try 96 dB/2.83v/m, 8 ohm nominal:

with two in stereo mode, mlp 8 ft, you get 88 dB spl, so 0.1% may be fine but 0.5% is -46 dB, so harmonics will be at 42 dB average that could be quite audible to trained ears. 96 dB/2.83v/m may be an extreme example but even with 88 dB/2.83V/m speakers, 0.5% THD could still bother some people in a quiet room. Even under more realistic conditions, such as 0.1% THD, 88 dB/2.83V/m speakers, that would still make distortions audible to those with sensitive and trained ears, again I would use people such as @RichB as example because he said he could hear the difference between amps like the ATI class D amp and the Benchmark AHB2 amp.

To add to the above, as you mentioned, IMDs are "bad", so aiming for the lower THD has the additional benefit that it will also increase the chance of minimizing IM distortions.

DOH!

WRT THD vs IM, THD uses only one frequency, IM uses two and since music isn't pure sine waves, I think IM distortion is a good indicator of an amplifier's performance, although I think a newer test could include more frequencies, to better simulate wider bandwidth amplification.

My point about .01% THD is that, even with more sensitive speakers, the distortion is still far below the level of the sound, so it's very difficult to hear- it's not the area where the amp is clipping.

WRT Curl, etc- isn't part of achieving 'best' specs for bragging rights, AKA 'reputation among those who would use their equipment', AKA 'marketability'?

 

[PENG]

DOH!

WRT THD vs IM, THD uses only one frequency, IM uses two and since music isn't pure sine waves, I think IM distortion is a good indicator of an amplifier's performance, although I think a newer test could include more frequencies, to better simulate wider bandwidth amplification.

My point about .01% THD is that, even with more sensitive speakers, the distortion is still far below the level of the sound, so it's very difficult to hear- it's not the area where the amp is clipping.

Again, there is no disagreement there, and my additional point is, since we know for a fact that if you have HD, you will have IMD, because IMD arises due to a similar mechanism as HD. The two are sort of intertwined, so amps that have lower HD will likely have lower IMD too relatively speaking, and on all else being equal basis.

WRT Curl, etc- isn't part of achieving 'best' specs for bragging rights, AKA 'reputation among those who would use their equipment', AKA 'marketability'?

I would agree those are likely important considerations, but there is merit for always aim for higher bar even if just to be on the safe side and/to cover all possible worst scenarios. For me, I am not too greedy, but I would still want to see <= 0.03%/-70 dB, 20-20,000 Hz (50 kHz minimum bandwidth) and output level between 50 (ok, will accept 100) mW and rated output for power amps. If that bar is clear than I would look at other metrics. OMMV.

 

[ski2xblack]

Peng, by that reasoning, amps we both have in our quivers, the ACA specifically, with it's inherently high harmonic distortion, should result in a spray of intermodulation products, making them virtually unlistenable. Except they're not. (Caveat: I could be stone deaf.)

 

[PENG]

Peng, by that reasoning, amps we both have in our quivers, the ACA specifically, with it's inherently high harmonic distortion, should result in a spray of intermodulation products, making them virtually unlistenable. Except they're not. (Caveat: I could be stone deaf.)

I agree, but I think most people can tolerate high % of distortions. I recently watched a pod cast the feature an interview with Earl Geddes, he cited a number that seems shockingly high. I know remember what it was but pretty sure it was higher than 10%. The ACA amp's THD+N was almost 3% at rated output as measured on ASR but THD may still be at around 1% or slightly below. Also, you can see in the FFT that the harmonics structure is dominated by 2nd and 3rd, with the higher order ones dropping rapidly. So for sure it would have a fair bit of IMD as well but likely not as bad as the typical class AB amp that also has THD around 1% or higher.

Gene wrote something years ago that is still relevant to our discussion:
THD and IMD Distortion—Sidebar | Audioholics

He noted the following:

Today we have much better (ie. wider bandwidth, faster) output transistors which allow audio amplifiers to have extremely wide bandwidth (>100kHz) octaves above the limits of human hearing. Thus, THD+N testing can typically give us all of the relevant data we need if done properly. This involves testing the amplifier full bandwidth (20Hz to 20kHz) at low and high power to determine consistency of performance. If bandwidth is preserved and the distortion remains low (< 1% at rated power) than this is usually a good indication that the amplifier will also perform similarly well under IMD testing.

 

[dlaloum]

And we should also remember that we are feeding speakers that will if they are excellent, have THD of 0.5%... and if they are average (for THD) - 3% or more - especially in the bass

 

[PENG]

And we should also remember that we are feeding speakers that will if they are excellent, have THD of 0.5%... and if they are average (for THD) - 3% or more - especially in the bass

Yes, but also remember that the quoted average THD will always include the THD produced by the signal source contents that could be poorly recorded and by amplified musical instruments, dac, dsp, preamps and power amps and everything else that can produce distortions. Speakers end up getting everything, good or bad, it is everything!

 

[highfigh]

Yes, but also remember that the quoted average THD will always include the THD produced by the signal source contents that could be poorly recorded and by amplified musical instruments, dac, dsp, preamps and power amps and everything else that can produce distortions. Speakers end up getting everything, good or bad, it is everything!

I think that s bit of the reason for designers striving for low distortion in their electronics has to do with "If the system hits 3% THD, it didn't come from MY equipment!", AKA "Don't blame me!".

 

[PENG]

ASR finally measured one of the popular power amp, a Rotel:
Rotel RB-1070 Amplifier Review | Audio Science Review (ASR) Forum

First, the usual ranking chart by SINAD that seems to be a target by other reviewers (such as some Youtubers) who often cited ASR's obsession of SINAD blablabla. In this review, SINAD was well below many AVRs:

Yet the Rotel did get on his recommended list. If you read the whole thing, you may understand why he would recommend this amp, yet not the very popular AVR-X3800H or the Onkyo RZ50, also reviewed not long ago.

Or forget reading the whole thing, just jump to his conclusions:

Conclusions

As one of my favorite (looking) brands, I was very anxious to measure a Rotel amp, hoping it would not let me down. Fortunately it did not. No, the SINAD doesn't break new ground but the rest of the measurements are very good with plenty of power available. It is a great choice for people wanting good amount of power in an attractive package with competent performance.

I am going to recommend (used) Rotel RB-1070.

So the man can be (or has been, if people read the fine print every time) reasonable!

 

[AcuDefTechGuy]

ASR finally measured one of the popular power amp, a Rotel:
Rotel RB-1070 Amplifier Review | Audio Science Review (ASR) Forum

First, the usual ranking chart by SINAD that seems to be a target by other reviewers (such as some Youtubers) who often cited ASR's obsession of SINAD blablabla. In this review, SINAD was well below many AVRs:


View attachment 59721

Yet the Rotel did get on his recommended list. If you read the whole thing, you may understand why he would recommend this amp, yet not the very popular AVR-X3800H or the Onkyo RZ50, also reviewed not long ago.

Or forget reading the whole thing, just jump to his conclusions:



So the man can be (or has been, if people read the fine print every time) reasonable!

What’s your takeaway on what is salient to him from reading his reviews?

Does he seem to value the DAC’s performance the most?

Is that the reason he didn’t recommend the x3800 (less DAC performance than the x3700), but recommend the Rotel amp that has worse measurements than the x3800 and doesn’t even have any DAC?

The Rotel amp has one channel with THD+N of 0.01% (SINAD 79dB), which is bad in terms of ASR standards compared to other amps with THD+N of 0.001%.

As for being reasonable, I don’t think recommending a product that has worse measurements than a non-recommended product is “reasonable”. I might be the only one who thinks this, but that’s my thoughts.

 

[PENG]

What’s your takeaway on what is salient to him from reading his reviews?

Does he seem to value the DAC’s performance the most?

Factually speaking, no! Based on hearsay, many took his commentary out of context, yes!! Some forum posts might have made you believe that all ASR cares about is SINAD, the rest is history. Such kind of talks got repeated enough, so many people believe it is a fact that ASR/Amir cares only about SINAD of the preamp/dac, but that is not true.

Is that the reason he didn’t recommend the x3800 (less DAC performance than the x3700), but recommend the Rotel amp that has worse measurements than the x3800 and doesn’t even have any DAC?

Let's quote exactly his own words in the conclusion of his review on the Denon first, and then compare some of the measurements:
Denon AVR-X3800H Review | Audio Science Review (ASR) Forum

Conclusions
I had high hopes going into this review thinking the company had seen the advantage its superior objective measurements in the past and would try to capitalize on them. Sadly, the reverse seems to be true with the DAC section taking a large step backward. Considering that the 3800 costs $500 more than 3700H, this is very surprising to me. Yes, inflation has a lot to do with that but surely the eye needed to be focused on making sure they at least met the same level of performance as last generation.

The good news is that the amplifier seems to be same design as last generation and has only taken a small hit.

Denon had been my "goto" recommendation for AVRs and even AVPs. When anyone asked me about either, I would just say "get a Denon AVR." While subjectively the performance of this new generation may be similar, I can't accept the regression in objective measured performance.

It is with much sadness that I cannot recommend the Denon AVR-X3800H.

My interpretation is that as you said, it at least has a lot to do with the DAC, but don't you think in this case it is understandable, considering the SINAD spec of the DAC chip dropped from 107 dB (best case) to 93 dB (best case)? Okay we agreed many times in the pass that 93 dB SINAD is not an audible issue, but even if this is true under any conditions, would you be disappointed if you found out you next CX-A5300 got the DAC chips downgraded to the one D+M are now using, so the SINAD spec, or THD+N spec dropped from your CX-A5100's -110 dB to -93 dB, a whopping 17 dB increase in THD+N, and the price had gone up by 50% to offset inflation?

The Rotel amp has one channel with THD+N of 0.01% (SINAD 79dB), which is bad in terms of ASR standards compared to other amps with THD+N of 0.001%.

As for being reasonable, I don’t think recommending a product that has worse measurements than a non-recommended product is “reasonable”. I might be the only one who thinks this, but that’s my thoughts.

I have agreed with you in the past that THD+N of 0.01% is fine , even 0.05% is fine if it is the worst case scenario but you can't say the same by looking at a single measurement point such as the typical 1 kHz test, and without specifying the frequency bandwidth and output level.

With that in mind, please compare the following, that I have also posted using ASR's measurements as example. You can used Gene's too, but it is easier to find Amir's simply because he has done so many.

I remember you suggested that we should look at SNR in conjunction with THD+N so now let's compare their SNR:

See that the Rotel is at least 10 dB better, and by the way, the example RX-V377 you cited showed very good SNR but S&V did not specify the output level, so it could be (most likely) measured at rated output, and their figures were almost always "A weighting", ASR's are for 22.5 kHz bandwidth, unweighted.

Cautionary note: Here we are comparing SNR measured on an AVR versus that measured on a power amp, I don't what the SNR would be if the AVR's power amp itself was measured.

So to me, the takeaway of his reviews, it you always read the entirety, would be the following:

- ASR typically provide many charts, the first one is the SINAD ranking, may be because that is usually the first of a battery of tests, and also because he does think the preamp/dac section is important as many ASR members use external power amps that could have gains from as low as under 20 dB to as high as over 30 dB.

- ASR does not look at SINAD as a single number, just because their is a ranking chart, making it easy to compare, and index the large number of reviews for easy find, does not mean that is the only metric to look at. Members who frequent the site, usually know well enough to look through the rest of the reviews before making their jumping to decision.

- When looking at SINAD, one should look carefully on the graphs if provided, again, the single point one measured with a 1 kHz signal is only for a quick compare, and if it is >>100 dB that may be good enough, but if it is lower, such as <90 dB, go look for the details, as that may tell you a different story.

- It is not true that Amir's recommendations are based on a single SINAD number. The Rotel example shows that he does consider other metrics, even the price point is often considered, remember his comments on those $20, or was it $12 dongle dacs? I picked the Rotel as example because it's the latest review, readily available. You can find many more examples that shows what some of those youtubers said about the site's is not always true, though in some cases may be true to a point. Remember that Youthman vido that got posted and reposted, sometimes that's how hearsay became hearsay, as those guys have the megaphones that you and I don't have.

Lastly, back to your first question, I can think of a few AVRs/AVPs that have excellent DAC chips that did not get on Amir's recommended list.

Arcam AVR/AVPs - Arcam typically used better DAC chips than the mass producers such as D+M's, and Amir has reviewed at least 3 or 4 of their AVRs and AVPs, none got his recommendations.

Anthem AVRs also typically used above average DAC chips (for AVRs) and out of the 4 AVRs/AVPs, only 2 made it to his recommended list.

Marantz AVR/AVPs - He reviewed quite a few, recommended none, even the $5,000 AV8805 that has AKM's ex flagship DAC did not make it, but almost, if you read the fine print that he sort of said ..buy it for the feature..