Suggestion for reviews: Extend the frequency range for 'THD+N vs Freq.' test

[nick_l44.1]

As to ultrasonics, I don't have to cater to my cats' hearing range with my system.

This may be true if the ultrasound does not cause clipping in amplifiers and does not return as intermodulation within the audible range, which often happens.

 

[amirm]

Legit infrasonic content exists, more frequently in movies than in music. It's not very common, but it's out there. And some people use these dongles with external amps...

Wasn't talking about content but application of these products in systems that reproduce under 20 Hz. You have any? Who is buying Apple dongle for that???

 

[nick_l44.1]

Wasn't talking about content but application of these products in systems that reproduce under 20 Hz. You have any? Who is buying Apple dongle for that???

In fact, no special systems reproducing frequencies below 20 Hz are required. The artifact is clipping, and it is wideband. This problem affects not only Apple dongles, but virtually all devices using Cirrus chips with active DRE.

I myself do not see the need to expand the THD+N vs. Freq test, since this artifact is not harmonic distortion and easily occurs in other cases. It simply requires a special description, and only once. A large increase in harmonic distortion with an increase in frequency of 20 kHz at low signal levels is also of little interest in itself. However, measuring the intermodulation that occurs in this case depending on the signal level would be very valuable. And not only SMPTE.

 

[Sokel]

20Hz is legit as fully in the audio range and may come with suprises even for DACs.
Example:

1kHz

20Hz

I don't see the point lower than that though.

Edit: Don't be fooled by the spikes at -130dBr or so, this is a 512k FFT which is needed for low freq measurements, they are visible just because it digs deep into the noise floor.

And I should measure with DSD, it eliminates this visible horrible grass at 300Hz-2kHz (range varies with tested signal, it's the famous bad implementation of 9038QM which you won't see unless you measure low) THD+N won't be different though.

 

[kyuu]

20Hz is legit as fully in the audio range and may come with suprises even for DACs.
Example:

View attachment 507589

1kHz


View attachment 507590

20Hz

I'm a bit confused. What's the surprise? A second harmonic at -110dB? The grass at -120dB?

 

[Sokel]

I'm a bit confused. What's the surprise? A second harmonic at -110dB? The grass at -120dB?

Pretty much all of it.
Sometime THD+N is tricky, better hunt for N+D or TD+N if you do the chase thing.

At 2020 I would expect a straight performance from 20Hz without penalties as my poor interface does for 20 years now with its AKMs (in fact it's a little better down low that 1kHz but also better from the one you see at the new one) .

 

[BeerBear]

Wasn't talking about content but application of these products in systems that reproduce under 20 Hz. You have any? Who is buying Apple dongle for that???

As Nick said, you don't need speakers that can play those low frequencies. A 5Hz input tone produces artifacts way up in the audible range (even at 500Hz and more). Actually, speakers that can't reproduce very low frequencies might make the problem worse, since they provide less acoustic masking.

I myself do not see the need to expand the THD+N vs. Freq test, since this artifact is not harmonic distortion and easily occurs in other cases. It simply requires a special description, and only once.

Right, there could be better tests for this specific issue.
But my thinking is that the 20-20kHz range is inadequate anyway. I know it's 'standard' and I've seen it used by many people, but it ignores the reality that there's regular audio content outside of it. If the goal is to thoroughly test a device, might as well sweep the full range. Today it can help point to the issues of these dongles and tomorrow it could help with something else. Even from a purely experimental/scientific POV, I think it's good to have that extra data.
The only downside I can see is that the screenshots would need to be bigger, which is not such a high price to pay.

 

[amirm]

A 5Hz input tone produces artifacts way up in the audible range (even at 500Hz and more).

Let's say it did. You are not hearing 5 Hz. No one who created that content heard 5 Hz. So asking for purity there makes no sense.

Actually, speakers that can't reproduce very low frequencies might make the problem worse, since they provide less acoustic masking.

That may very well be the intended sound then as only those harmonics, if they were sufficiently high amplitude, would have been heard during production.

 

[BeerBear]

You are not hearing 5 Hz.

Obviously. I thought I made it clear that I'm not hearing the 5Hz tone, but rather some byproducts of that 5Hz tone. (And they're not harmonics. I linked to the thread with the recordings a few posts ago, if you're curious.)

That may very well be the intended sound then

I mean........ sure, if such faulty devices were used during production, then I guess it would be the intended sound.
But don't you think gear should be tested and judged according to objective standards of quality? Low THD+N and all that? What are we even talking about here?

 

[amirm]

I mean........ sure, if such faulty devices were used during production, then I guess it would be the intended sound.
But don't you think gear should be tested and judged according to objective standards of quality? Low THD+N and all that? What are we even talking about here?

For what? Finding whatever is the issue is with CS DACs? Or in general? You all keeping going back and forth and I don't know what you are discussing.

As a general rule, I run more tests and more stringent tests than anyone. Adding more tests is very onerous when multiplied across every product I test. And further, the more tests, the more people's eyes glaze over them and ignore it all. So just because I care about measurements, it doesn't mean I keep adding to them.

To be clear, there is no case whatsoever for testing products at 5 Hz. I don't care if content has some of it in there. Systems that can reproduce that and people who care is insignificant.

 

[tccalvin]

I am very unhappy with all this resistance to change and improvement.

It first became evident with the refusal to test for Inter-Sample Overshoots, even with the overwhelming evidence that they are indeed a problem with oversampling DACs and after several tests have been proposed that aim to measure the impact of such artifacts.

Now, it continues with the refusal to update tests for the distortion behavior of the CS chips. Again, after evidence of the issue has been gathered and working tests have been developed.

And don't even get me started on the "It's inaudible" argument. The issues described fall well within the audible range (way above the -96dBFS threshold and covering the whole 20Hz-20kHz window and beyond, albeit for short bursts). Such sound defects are at least as audible as sub-par SINAD if not more.

Even if it were the case that most people do not notice the aforementioned artifacts, DACs are not just utilized in our personal little home-theater setups. They are used in so many more applications: professional audio equipment, musical instruments, effects pedals, outboard gear, measurement equipment... you name it.

By giving a DAC with a known audible issue a perfect score, we are allowing millions of digital devices to have a clipping fest in their outputs.

Right now, I'm not even referring to the specific proposal of this thread, although I don't see why it should be dismissed so quickly with so much confidence.

I regard this approach as un-scientific and borderline dishonest.

If I see another review get a perfect score even though meaningful measurements have been purposefully skipped, I'm going to call it out.

 

[amirm]

I regard this approach as un-scientific and borderline dishonest.

And I regard your post as insulting and out of line. I am not here to respond to personal remarks, stomping your feet, etc. As I already explained, there is an extremely high bar for adding tests to comprehensive suite I run now. If you can't make a strong enough argument, don't look to me for a problem.

By giving a DAC with a known audible issue a perfect score, we are allowing millions of digital devices to have a clipping fest in their outputs.

If there is a problem with a specific DAC, then it does NOT call for adding a test to every DAC. Half the time you all are arguing this, and half the time you seem to claim this is a universal need. Make your mind and be clear with your justifications and arguments.

 

[amirm]

It first became evident with the refusal to test for Inter-Sample Overshoots, even with the overwhelming evidence that they are indeed a problem with oversampling DACs and after several tests have been proposed that aim to measure the impact of such artifacts.

Oh, we are back to this argument now. Once more, forcing the industry to reduce input level by a few dBs to deal with this, which in turn reduces performance for ALL content is not something I am going to advocate. You want it anyway? Just reduce the volume ahead of the DAC in the EQ system that you need to utilize anyway. You would be doing the exact same thing you expect the DAC to do. There is no magic in a DAC that has such ability.

 

[tccalvin]

I would like to make it very clear that there is nothing personal about what I wrote and I don't plan on attacking anyone on this forum or anywhere. I am simply expressing my feelings and my opinions on the extremely conservative approach adopted even in the face of strong evidence (I don't know what would clear the "high bar" if not that).

 

[Biagiod]

I am very unhappy with all this resistance to change and improvement.

It first became evident with the refusal to test for Inter-Sample Overshoots, even with the overwhelming evidence that they are indeed a problem with oversampling DACs and after several tests have been proposed that aim to measure the impact of such artifacts.

Now, it continues with the refusal to update tests for the distortion behavior of the CS chips. Again, after evidence of the issue has been gathered and working tests have been developed.

And don't even get me started on the "It's inaudible" argument. The issues described fall well within the audible range (way above the -96dBFS threshold and covering the whole 20Hz-20kHz window and beyond, albeit for short bursts). Such sound defects are at least as audible as sub-par SINAD if not more.

Even if it were the case that most people do not notice the aforementioned artifacts, DACs are not just utilized in our personal little home-theater setups. They are used in so many more applications: professional audio equipment, musical instruments, effects pedals, outboard gear, measurement equipment... you name it.

By giving a DAC with a known audible issue a perfect score, we are allowing millions of digital devices to have a clipping fest in their outputs.

Right now, I'm not even referring to the specific proposal of this thread, although I don't see why it should be dismissed so quickly with so much confidence.

I regard this approach as un-scientific and borderline dishonest.

If I see another review get a perfect score even though meaningful measurements have been purposefully skipped, I'm going to call it out.

"I am very unhappy" by the rudness of your comments towards Amirm and ASR. Do not really understand how people can think of having the right to actually INSULT (dishonest ?) The founder of this rigorously maintened scientific audio website.

We can all indeed disagree but hopefully always in a civil manner. Your attitude instead is disgraceful and as a dire consequence also your comments whether valid or not become secondaries.

Amirm already commented that yours is not as a strong argument in his view

If you are so adamant, how about doing yr own tests at sub frequencies and posting them here at ASR? You may even consider keeping drama out and sticking to science

As we say, it is a difficult world out there...

 

[Sokel]

Industry always search for ways to "game" results, depending the trend of the era.

Remember 9038QM before the fix?
Data was already there, at the IMD vs level chart, right on the money when nicely recorded music bangs.

Yet it got away with it for sometime but not long.
It got recommendations alright, both objective and subjective until some people here decided to search deeper.
And bingo!
Same with faulty DSP, people searched and at least one company listened.

Who's the fault here? Data was present but most relied on the THD+N and nothing else.

Lesson? Just wait for more data, early adopters are always at risk.
We have lots of people here measuring, it always comes around.

 

[tccalvin]

We can all indeed disagree but hopefully always in a civil manner.

Last clarification.

I am disagreeing in a civil manner. Not liking what I said doesn't make me "rude" or "disgraceful".

I'm not interested in ad-hominem insults. I don't know Amir personally and I am grateful for what he's done with ASR and I continue to be a member because I like the forum and its mission. I shouldn't even have to say this, since it's irrelevant to my point.

My point being: if I know that a product has issues that show up with certain measurements and I keep only showing measurements that make the product look good (regardless of my intentions or biases), then my approach (with regards to this specific issue, not me, as a whole) is un-scientific and borderline dishonest.

Measurements should speak for themselves. People (preferably informed people) will decide what value the results of those measurements hold for their specific use-case.

That's my take. If you keep reading personal insults in what I wrote, then that's on you.

I would like to believe that user feedback is still welcome here.

 

[MAB]

Last clarification.

I am disagreeing in a civil manner. Not liking what I said doesn't make me "rude" or "disgraceful".

I'm not interested in ad-hominem insults. I don't know Amir personally and I am grateful for what he's done with ASR and I continue to be a member because I like the forum and its mission. I shouldn't even have to say this, since it's irrelevant to my point.

My point being: if I know that a product has issues that show up with certain measurements and I keep only showing measurements that make the product look good (regardless of my intentions or biases), then my approach (with regards to this specific issue, not me, as a whole) is un-scientific and borderline dishonest.

Measurements should speak for themselves. People (preferably informed people) will decide what value the results of those measurements hold for their specific use-case.

That's my take. If you keep reading personal insults in what I wrote, then that's on you.

I would like to believe that user feedback is still welcome here.

I thought it was insulting. With a bit of a threat at the end. And you are still insinuating in this post dishonesty and a lack of scientific rigor. Doing it in the third person doesn’t change that.

Thanks for the clarification.

 

[kyuu]

If Amir added every test that everyone got a bug up their butt about, he'd be spending days on every device. I can only speak for myself, but I think there's already a good balance between thoroughness and, y'know, actually getting things done. Especially for things as generally uninteresting as a DAC.

 

[JSmith]

Reviews here are done and published without charge, for all to consume. Each review takes a lot of time, which is @amirm's own personal time... which he gratefully sacrifices for the good of ASR and without asking for anything in return.

Amir is a very honest broker, evidenced by the fact this site is advertisement free and all industry members need to be tagged accordingly and follow stricter posting rules.

To suggest otherwise is rather disingenuous.


JSmith