• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Understanding Audio Measurements

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,835
Likes
16,497
Location
Monument, CO
We measure jitter directly on wideband signals. The DSO (digital sampling oscilloscope) does jitter separation (usually gets it right, not always) and generates histograms, eye plots, spectra, and a wealth of useful tables about the jitter components. Costs about $500k USD, better up your donations boys and girls, Christmas is coming and we could get Amir a new toy! :)

In the end nobody outside the test and standards world (well, and some R&D folk) cares about the jitter itself; we care about what it does to the signals we are listening to (or whatever).

IOW, what Amir said.
 
Last edited:

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
So the signal of -110dB wil be reproduced with amplitude accuracy of +/-0,5dB?

What does those -20dBFS in the title mean?

Why are you not happy? What would you say are acceptable results for this test?
 
Last edited:
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
What does those -20dBFS in the title mean?
The graph needs to be normalized to something to deal with constant offset/bias. So -20 dB is picked by default as that reference. It is loud enough where linearity is not a problem, but low enough as to not include distortion.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Why are you not happy? What would you say are acceptable results for this test?
The waveform at low amplitudes is modulated up and down. So depending on when the samples are taken by the analyzer, you get different results. This problem clearly shows the reason we do linearity tests. The waveform should not be modulated up and down in the DAC.

You have to judge what is acceptable to you. My expectation is that there is less than +-.5 dB of error at -120 dB which is easily achieved even with cheap DACs. Here is Topping D50 for example at $250:

1532449255669.png
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
The graph needs to be normalized to something to deal with constant offset/bias. So -20 dB is picked by default as that reference. It is loud enough where linearity is not a problem, but low enough as to not include distortion.

I see. Does that mean that 0dB on your graph actually equals -20dB, -10dB on the graph equals to -30dB .. and -100dB equals -120dB of maximum output level?
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
You have to judge what is acceptable to you. My expectation is that there is less than +-.5 dB of error at -120 dB which is easily achieved even with cheap DACs. Here is Topping D50 for example at $250:

View attachment 14199

My question was what is acceptable to some "good DAC engineering standards" (if something like that exists).

Noise in my room (as well as probabl in any other room where people live and that is not in the middle of the desert) is at best at around 40dB, so I'm not really worried about hearing anything happening at -110dB while listening at any level my speakers can produce.

Btw, Topping D10 is almost 3 times cheaper than D50.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
4,500
Likes
5,417
Location
UK
Noise in my room (as well as probabl in any other room where people live and that is not in the middle of the desert) is at best at around 40dB
It probably is a low frequencies, but not in the midrange, where the ear is most sensitive. I have measured negative db in my room in the midrange, I've posted the graph before.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
It probably is a low frequencies, but not in the midrange, where the ear is most sensitive. I have measured negative db in my room in the midrange, I've posted the graph before.

Amir measures linearity with 200Hz tone which is not really midrange. :D
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
My question was what is acceptable to some "good DAC engineering standards" (if something like that exists).
As pure engineering, there isn't one. I use -120 as the dynamic range of our hearing is about 116 dB. So a playback "channel" that can convey that, can be shown to be transparent. Anything less than that gets into shades of gray (content, listener, etc.). In other words, this assessment is a mix of psychoacoustics and engineering.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Amir measures linearity with 200Hz tone which is not really midrange. :D

Well it is by most definitions the start of the midrange. But I'm not sure what @Soniclife defined as midrange in his measurements.

It is worth noting though that bass noise floor will tend to mask the midrange too, although not to the extent it masks bass.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Does that mean that 0dB on your graph actually equals -20dB, -10dB on the graph equals to -30dB .. and -100dB equals -120dB of maximum output level?
No. The voltage output at -20 dB i used as a reference. Then when the dB value is lowered in the test, the ratio is computed from what a correct division from output at -20 dB would be. So it is a normalization value. At all time, the ratio is the mathematical division versus actual division by the DAC.
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
No. The voltage output at -20 dB i used as a reference. Then when the dB value is lowered in the test, the ratio is computed from what a correct division from output at -20 dB would be. So it is a normalization value. At all time, the ratio is the mathematical division versus actual division by the DAC.

Sorry, I didn't get that. What is actually the relationship of the -20dB reference level and steps of 0, -10, -20, .. -120dB?
 

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,065
Location
Zg, Cro
As pure engineering, there isn't one. I use -120 as the dynamic range of our hearing is about 116 dB. So a playback "channel" that can convey that, can be shown to be transparent. Anything less than that gets into shades of gray (content, listener, etc.). In other words, this assessment is a mix of psychoacoustics and engineering.

Ok, that sounds reasonable. But noise floor at -120dB seems one thing and amplitude error of +/- 0.5dB of a signal at -120dB seems another, and that last one doesn't seem like a hearable issue at all to me. Btw, wouldn't even a very solid amp have a SNR of 100-110 dB so its noise will anyhow mask those little errors?

Btw, I still think that there is some noise in every room so the things happening under -100dB are not really an issue when listening to the music at home at any "normal" level, but nevertheless, let's discuss it. :)
 
Last edited:

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,427
Likes
3,981
Location
SoCal
I do wonder if linearity should be checked on several frequencies just to make sure some particular behavior at 200Hz doesn't cause the overall conclusion to be based on limited data. Just for the test coverage's sake.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,427
Likes
3,981
Location
SoCal
Sorry, I didn't get that. What is actually the relationship of the -20dB reference level and steps of 0, -10, -20, .. -120dB?

I think the assumption is made that at -20dB the DAC performs in an ideal manner and the output voltage is recorded, the rest of the level measurements are normalized based on that point.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
I do wonder if linearity should be checked on several frequencies just to make sure some particular behavior at 200Hz doesn't cause the overall conclusion to be based on limited data. Just for the test coverage's sake.

That's been answered several times. Bottom line- makes no difference, which isn't surprising.
 

gvl

Major Contributor
Joined
Mar 16, 2018
Messages
3,427
Likes
3,981
Location
SoCal
That's been answered several times. Bottom line- makes no difference, which isn't surprising.

I thought this was based on the Yggdrasil 1kHz test per earlier comments, does this mean it is universally true?
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Actually, at low levels, the analog nonlinearities are minimized. They're most important at high levels.
 
Top Bottom