• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Article: Understanding Digital Audio Measurements

Joined
Nov 10, 2019
Messages
9
Likes
15
#21
Ho Lord! Last days we discussed with my wife about music gear and she wants we buy some equipment for measuring. And yesterday we discussed more than one hour about your measurements and what I have understood and how I interpret the things. I have to share the link and it will cost us a lot of new debates about your marvellous explanations!
The Room Measurement article might be helpful. Practical reasons for measuring. https://www.audiosciencereview.com/...om-measurement-tutorial-for-dummies-part-1.4/
 
Joined
Aug 30, 2019
Messages
60
Likes
82
Location
The Netherlands
#22
This is a fantastic article. Great resource for new readers and people that claim that all DAC's sound the same. I really appreciate that ASR offers such a broad selection of component reviews to expose the levels of engineering quality.

DAC's are mighty interesting, it's at the beginning of the chain. Everything a DAC produces other than the signal is going through your - naturally - imperfect pre-amplifier, signal processor, power-amplifier, crossover...

Thanks for the write up. Happy holidays for all of you and may your new year resolutions be at least 20-bits. It's 2020 after all.
 
Joined
Apr 16, 2019
Messages
32
Likes
15
#23
Thanks for this great tutorial!

I'm curious as to your thoughts on ranking by SINAD vs IMD. The Okto DAC8 is ranked at 118 SINAD vs Benchmark DAC at 112, but in the IMD chart, Benchmark shows better performance at most input levels. Wouldn't IMD be more indicative since it's slightly closer to complex signals and the more stringent test?

Multitone distortion measurements would be even better in that regard. "Statistically it is closer to music than a sine wave or two tones." http://www.aes.org/e-lib/browse.cfm?elib=14381
 

renaudrenaud

Senior Member
Joined
Apr 20, 2019
Messages
364
Likes
557
Location
Paris
#24
It is a true blessing that you and your wife can have such technical conversations - that there is both the desire and technical ability. My wife also cares that her music sounds good, but in her case that means good enough. She, like most people, is perfectly content to listen to mp3 music on laptop speakers (ugh). I suppose that most people listen to music as do many musicians, concentrating on how it makes them feel, or the technical merits of the performance. This is a perfectly valid perspective, although I am too distracted by bad sound to enjoy music that way.

Anyway, visitors to this site are interested in the best possible sound, usually at a chosen price point. How nice it is to have a mate that shares one's passion. Or at least tolerates it.
Life is confortable!
 

SpyB

Active Member
Joined
Sep 3, 2019
Messages
142
Likes
101
#25
Thanks Amir,A very well presented and informative article

That should assit all in their analysis of testing results
 

LTig

Major Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
1,738
Likes
2,447
Location
Europe
#28
I noticed Amir mentioned that when running two DAC’s at the same time, the two different clocks can drift apart over time. I am planning to put together a 5.1 system based on a PC and JRiver.
One option would be to have 3 stereo DAC’s, with JRiver doing the channel assignments for L, R, C, surrounds and subwoofer via USB. In that case, there would be three different clocks (one in each DAC). Could they drift apart enough to cause problems, and make that solution unworkable? Is it better (or maybe required) to have, say, an Okto, which would have all channels on a single clock?
With USB you need a single clock for the DACs because in the typical USB mode the clock in the DAC sets the timing.
 
Joined
Feb 11, 2018
Messages
79
Likes
79
#31
With USB you need a single clock for the DACs because in the typical USB mode the clock in the DAC sets the timing.
Yes, for multi-channel audio conversion you will need a multi-channel DAC or three DACs that can be synchronized to an external master clock.
Or use a surround processor with internal DAC.
If you had three SPDIF outputs there might be no problem (assuming the DACs are identical and use a conversion clock that's extracted from the receiver), as with SPDIF the source is the master clock.
 

Morpheus

New Member
Joined
Jun 5, 2019
Messages
4
Likes
14
#33
With USB you need a single clock for the DACs because in the typical USB mode the clock in the DAC sets the timing.
Hello,
Saw your post and would like to know, if you could answer...what if you feed an USB signal, non multichannel, into an USB/SPDiF converter, then to a Mindsps MiniDiGi that splits the incoming data into 3 diferent streams (wich you can internally perform crossover/equing or) that you finnaly feed into three different DACs/amps ( preferably the same model, but not mandatorily) for multiway active loudspeakers?
Conceptualy, getting the data and manipulating/routing/converting everything in the same unit, using the same reference seems ideal, but would this workaround be a viable and qualitative setup, or would clocking issues or others be relevant..? Thanks.
 
Joined
Feb 11, 2018
Messages
79
Likes
79
#34
Hello,
Saw your post and would like to know, if you could answer...what if you feed an USB signal, non multichannel, into an USB/SPDiF converter, then to a Mindsps MiniDiGi that splits the incoming data into 3 diferent streams (wich you can internally perform crossover/equing or) that you finnaly feed into three different DACs/amps ( preferably the same model, but not mandatorily) for multiway active loudspeakers?
Conceptualy, getting the data and manipulating/routing/converting everything in the same unit, using the same reference seems ideal, but would this workaround be a viable and qualitative setup, or would clocking issues or others be relevant..? Thanks.
If the three outputs are SPDIF, they will all deliver the same clock rate to the DACs.
Non-identical DACs, however, will not necessarily have the same conversion delays. Presumably, this can be corrected by adjusting delays in the dsp.
 

bequietjk

Active Member
Joined
Jan 4, 2019
Messages
278
Likes
280
#35
Wow. The Okto DAC8 scores remarkably well in the multitone and is a DAC that I would LOVE to listen to at some point in my life.

Great thread, amazing reference and so glad you put this together. Any time I have trouble remembering how the testing process goes, I can come back here and refresh :)

Thanks, amir. I'll be paying much closer attention to the 32 tone test, along with the THD+N Ratio vs Frequency graph that I still need to understand better.
 
Last edited:

Eirikur

Senior Member
Joined
Jul 9, 2019
Messages
318
Likes
462
#36
Thanks @amirm for this very nice introduction article, I'm sure this will help a lot as an anchor for many questions that may arise from your measurement reviews.

Can you help me to visualize and understand the reconstruction filters?
To test for the filter response, we feed the DAC random white noise, which naturally has infinite bandwidth. The response of the low pass filter becomes obvious once we capture the output of the DAC and convert it to the frequency domain using FFT. Figure 16 shows an example of this as I change the filter settings in the DAC.



1577434265779.png

Figure 16: example of different DAC output filters.​

The audibility impact of such filters is likely very low to non-existent so I don’t put a lot of value on this test.
In the presented graph we can see frequency response with white noise signals (*) but how does this equate to proper input signals? [EDIT: Can we assume the white-noise is a fully legal signal with only <FS/2 components?]

What are the (aliasing) effects, if any, due to the different reconstruction filters on legal signals?
For instance, I would expect that a properly formed 44100Hz input would have zero frequency components above 22050Hz when properly reconstructed, is that correct?
The question is then: which of the filters presented constitute a proper reconstruction? Intuitively I would say the blue one (steep decline towards FS/2), while e.g. the green one seems... wrong?

Thanks, and keep the articles and reviews coming!
Regards, Erik

[Edited, thanks Ray!]
(*) White noise can be an illegal signal when the levels are set to high (causing inter-sample overs)
As example of such an illegal signal, I used Audacity to create 44100Hz white noise (0dB) and up-sampled it to 384kHz and indeed you can see FFT frequencies extend all the way up to 192kHz:
1577433413239.png
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
9,028
Likes
6,904
Location
Riverview FL
#37
(*) White noise is "special", essentially it is a constructed illegal signal, right? [Please correct me if I'm wrong]
White noise is random sample values...

https://en.wikipedia.org/wiki/White_noise

As example, I used Audacity to create 44100Hz white noise and up-sampled it to 384kHz and indeed you can see FFT frequencies extend all the way up to 192kHz:
I created 44.1khz 16bit white noise, converted to 32bit float, and resampled it to 384kHz, and had to attenuate the result by 5.791dB to lose clipping, without dither...

1577435403657.png


Looked the same with dither...

Resampling the 16bit without converting to 32bit float, and without losing the clipping looks about like like yours:

1577435686864.png


So, I think the extra frequencies are due to clipped samples after resampling. Full scale white, as generated by Audacity, has "intersample overs"

1577436180829.png



Oops... log instead of linear frequency above...

1577435848917.png
 
Last edited:

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
9,028
Likes
6,904
Location
Riverview FL
#38
Try your white test again...

But create the white with an amplitude of .5 instead of 1 (that's about -6dB), and resample without dither.

1577437106421.png


With shaped dither:

1577437217677.png
 
Last edited:

Eirikur

Senior Member
Joined
Jul 9, 2019
Messages
318
Likes
462
#39
I created 44.1khz 16bit white noise, converted to 32bit float, and resampled it to 384kHz, and had to attenuate the result by 5.791dB to lose clipping, without dither...
Thanks Ray, something funky is going on; my result was created fully within the 24bit domain though.
In the float32 domain like yours it seems like the one you posted... until you save it to a 24bit FLAC and load it again, then it's back to my original FFT!

Try your white test again...

But create the white with an amplitude of .5 instead of 1 (that's about -6dB)
Yup, that's it - now lemme quickly update my questions to Amir before he sees what a dumbo I am ;)
 

bennetng

Addicted to Fun and Learning
Joined
Nov 15, 2017
Messages
875
Likes
708
#40
What are the (aliasing) effects, if any, due to the different reconstruction filters on legal signals?
For instance, I would expect that a properly formed 44100Hz input would have zero frequency components above 22050Hz when properly reconstructed, is that correct?
How about this? A bandlimited "legal" signal at 44.1kHz and Audacity says it has 4.006dB headroom.
Image1.png


Don't use the highest resampling quality:
Image3.png


After resampling to 88.2kHz you can see that the peak still increased.
Image2.png


The fact is digital filters in DAC chips are almost never as good as a high quality software one, and software DAWs are very popular these days. As long as the input signal has a higher frequency limit towards Nyquist, digital filters in DAC chips can still have some effects, despite audible or not.
 
Top Bottom