General debate thread about audio measurements

[amirm]

@amirm then from same akm datasheet
''It is recommended for SACD format book (Scarlet Book) that the filter response at SACD playback is an analog Low pass filter with a cut-off frequency of maximum 50kHz and a slope of minimum 30dB/Oct. The AK4490 can achieve this filter response by combination of the internal filter (Table 32) and an external filter (Figure 42).''

No test show if the lpf of a dac is optimised for cd or sacd or high res.

That says nothing about "high res." It is about SACD only (not PCM). DSD encoding creates a lot of noise that is pushed out to ultrasonic range using noise shaping/DSP. If you play DSD content even with the above filter on, you are "listening" to tons of ultrasonic energy. So if you think ultrasonics are bad, you better stay way away from SACD!

Speaking of high-res, you should watch this analysis I did a while back on what is inside high-frequency music you purchase: (watch full screen)

 

[Calexico]

The problem is that you write more than you read. Did not not see this graph in the Topping D50s where you started this argument?

Even if I had not post the above measurements, audibility of these filters is highly questionable.

I saw this but it's incomplete. It's only for cd samplerate.
At 192 khz the cut off frequency is around 96khz. Thats's why i'd like to see if frequencies between 20khz and 96khz for 192 khz generate noise in the 20khz 96khz range (before the digital filter)
On datasheet we see that the cut off frequency is shifted according to fs/2
That's why i 'd like to see noise for high res. Do high frequencies that are no more filtered at22khz generate ultrasonic noise before and after the cut off freq of the filter.?
Is it clear enough??

 

[Calexico]

@amirm
That's the kind of tests i'd like to see for different samplerates
https://www.audiosciencereview.com/forum/index.php?attachments/0minus4-18i20-png.8053/
On your test we can see that the upper frequencies where the noise grows is hidden because you stop too early at 30khz.
And for high res audio i'd like to see if a 40khz freq generate thd / noise or not.
For 192khz i'd like to see if a 80khz freq generate thd / noise or not
If not i will feel more confident to feed dacs with 44.1khz samplerate source.
(Because i don't know how they perform at higher samplerates as no test shows if dac perform better or not at higher samplerate.)

 

[solderdude]

That's the kind of tests i'd like to see for different samplerates

or can you ?

something like this:

or this:

What do these quite typical plots tell you ?

 

[Calexico]

or can you ?

something like this:

Yes and i'd like exactly the same with a 80khz sinewave
To be sure that high res freq doesn't introduce more noise

 

[Calexico]

I think those tests can reveal differences that are not less important than others shown and it show us the differences of filter technics between constructors.
I'd like to see it because i think older dacs had much more ultrasonic noise than now (particularly pure delta sigma ones)
Then we can see with newer generation of dacs the evolution of the noise filtering.
If all noise is always lower than 100 db it's very good.
If a 30khz sine produce distortion at -60db it's good to know it because it means it's better to stay on cd quality.
If 30khz test is perfect ok dac are really perfect.

 

[solderdude]

You 'wish' to see an 80kHz 0dB sinewave reproduced by a DAC and think that is valuable ?

If it was ... don't you think everyone would be measuring this and would be a standard test in any measurement suite ?

Perhaps you would like to see the spectrum of a 44.1k signal up to 5MHz (courtesy Miska, Signalist) of ADI 2 ?

 

[amirm]

That's the kind of tests i'd like to see for different samplerates

From the start, my goal has been to be minimalist when it comes to measurements. There are people who post measurements as if they are paid for each one. I am not one.

When I review gear, good members can ask for more measurements and give good reasons. If so, then I will provide them on a case by case basis. You have not falling in that bucket of good members so don't keep hopes high that I will do that if you ask. Hope you understand.

 

[KSTR]

Lack of proper reconstruction filter in a DAC causes aliasing in the audible band. That is why it is there.

Sorry, but this is incorrect. A DAC without any or with an insufficient reconstruction filter causes imaging (components mirrored above Nyquist), not aliasing. Aliasing (components above Nyquist mirrored down) happens in ADCs without bandlimiting filter.
If the subsequent electronics (and speakers) were distortionless, this imaging, producing ultrasonic frequencies, wouldn't matter much. But with distortion they will produce IMD components, some of them down in the audible range, and that's why the "reconstruction" low-pass filter is needed (and this filter must as well be distortionsless which isn't trivial), besides the wish to eliminate inaudible stuff anyway.
See eg Fig3 in this AppNote by Maxim: https://www.maximintegrated.com/en/app-notes/index.mvp/id/3853

 

[Calexico]

From the start, my goal has been to be minimalist when it comes to measurements. There are people who post measurements as if they are paid for each one. I am not one.

When I review gear, good members can ask for more measurements and give good reasons. If so, then I will provide them on a case by case basis. You have not falling in that bucket of good members so don't keep hopes high that I will do that if you ask. Hope you understand.

I don't mind if you don't do it for me. Do it for the music.

 

[gene_stl]

Amir , thank you for the YouTube video and channel. (not to mention this whole wonderful site!)

When I read about , for example DSD, and all the ultrasonic noise shaping shown on various graphs going way out of audible range my mind asks the question: "even though my ears can't hear it, my tweeter can't reproduce it or if it does at all it is like a straight laser beam in front of it, and my power amp probably won't like it, is this ultrasonic noise present at the output RCAs or XLRs of the D/A converter???? "

I, for one, don't want it there. It can't possibly be a "good idea". At the very least it could cause heating of your tweeter voice coil (if it got that far). It could cause extra heating of your output stage and reduce headroom. Are these typically filtered before the signal is output? I know lots of gear has "wideband design" Sometimes this may just serve to increase the potential noise. Should we have a 2x nyquist roll off to prevent the above concerns or is it already there.

PS What were you using to digitize into "Music Scope"? (Edit: I guess you just load the files in and don' have to digitize them. They are already digitized and this is software only analysis , yes?)

I once witnessed an electronics engineer doing an impromptu "amplifier clinic" A guy whom he did not like brought in a pair of Southwest Technical Products Tiger amps, the 60 or 75 wpc versions. The guy hit them with full power square waves and turned the frequency up. Of course the amps self destructed very quickly. (this occured in 1974. I always felt the guy did that because the kit builder was a very pretentious audiophile) But the point is there comes a frequency where the power devices can't keep up. It seems like these very wide band noise shaping signals might reach that point.

 

[Calexico]

You 'wish' to see an 80kHz 0dB sinewave reproduced by a DAC and think that is valuable ?

If it was ... don't you think everyone would be measuring this and would be a standard test in any measurement suite ?

Perhaps you would like to see the spectrum of a 44.1k signal up to 5MHz (courtesy Miska, Signalist) of ADI 2 ?

Thanks a lot. The -50db sine is an harmonic? Is it good bad or doesn't matter?

 

[amirm]

When I read about , for example DSD, and all the ultrasonic noise shaping shown on various graphs going way out of audible range my mind asks the question: "even though my ears can't hear it, my tweeter can't reproduce it or if it does at all it is like a straight laser beam in front of it, and my power amp probably won't like it, is this ultrasonic noise present at the output RCAs or XLRs of the D/A converter???? "

It is although some have filter settings to roll that off above certain frequency.

 

[amirm]

PS What were you using to digitize into "Music Scope"? (Edit: I guess you just load the files in and don' have to digitize them. They are already digitized and this is software only analysis , yes?)

Correct. Those were music tracks I either bought or downloaded for free. Music Scope just analyzes them. It is one of the few that can handle DSD content.

 

[Willem]

My Quad amplifiers are deliberately bandwidth limited to keep the crap out, the earlier models more so than the more recent ones.

 

[amirm]

I don't mind if you don't do it for me. Do it for the music.

Music has no full amplitude ultrasonic content.

 

[LTig]

[..]I'd like to see it because i think older dacs had much more ultrasonic noise than now (particularly pure delta sigma ones)
Then we can see with newer generation of dacs the evolution of the noise filtering.
If all noise is always lower than 100 db it's very good.
If a 30khz sine produce distortion at -60db it's good to know it because it means it's better to stay on cd quality.
If 30khz test is perfect ok dac are really perfect.

Well, here is the frequency response of my Arcam Black Box 3 (restored back into original state) with white noise @ -10 dB, and the multitione signal, both played at 44.1 kHz samplerate 16 bit and captured by RME ADI-2 PRO fs at 192 kHz samplerate 24 bits.

It's actually not that bad for a first generation bitstream DAC (bought 1991). All spurious signals are below 110 dB.

 

[gene_stl]

It is although some have filter settings to roll that off above certain frequency.

Do you suppose this could explain amps and tweeters mysteriously croaking? I am resetting up my system using some Crowns that have DSP on the input. I think I shall set an ultrasonic roll off just to be on the safe side.

Another question. Would you expect ,a DSD file played through a DSD to PCM conversion to have such ultrasonics?
Can you think of any reason you would want to keep that garbage? Since even the bestest tweeters are pretty useless above 30 kHz and would be like lasers at that frequency wouldn't that be a very good upper low pass or thereabouts?

 

[solderdude]

Thanks a lot. The -50db sine is an harmonic? Is it good bad or doesn't matter?

Where do you see a sine and harmonics ?
I see noise bands about 20khz wide around a 353kHz band (8x oversampling) and a smaller one at 16x oversampling.

How loud do you think your tweeter will reproduce frequencies around 350kHz ?
Will the tweeters react at all ?
Will they produce IM ?

Depending on how much your amplifier distorts and produces IM which, when it were to fold back, will again be lower in amplitude, certainly when using tube amps, will it 'harm' the sound ?

 

[Calexico]

Music has no full amplitude ultrasonic content.

That means it's better to stay cd quality. And that all analog lpf should start as soon as possible with as little roll off as possible.
Edit: or that means that there are only harmonics. Then high res has utility (debatable) if those high frequencies are not hidden in the ultrasonic noise