• 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!

Beta Test: Multitone Loopback Analyzer software

OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
I know this is off the debug focus, but is there consensus as to when the aes17 notch should be used? Thx

Use it to reduce the effect of noise and distortions that are close to the fundamental in the THD+N and noise/N+D calculations. It's used to remove (notch out) the fundamental in some calculations of THD+N, but MT doesn't use it for this.
 

Moto

Active Member
Joined
Jan 19, 2021
Messages
180
Likes
41
I understand its purpose, but am trying to understand under what circumstances it is appropriate to use.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
I understand its purpose, but am trying to understand under what circumstances it is appropriate to use.

When performing measurements using AES17 standard. According to AES Measurement of digital audio equipment publication:

Overview: Performing a THD+N ratio measurement
The test signal shall be a sine wave. The analyzer shall include the standard low-pass filter. The total signal level at the EUT output is measured and recorded. The fundamental component of the test signal present in the output shall be removed with the standard notch filter, and the unweighted level of the residual shall be measured. The THD+N ratio shall be the difference between the residual level and the total signal level, expressed in dB
 

Moto

Active Member
Joined
Jan 19, 2021
Messages
180
Likes
41
When performing measurements using AES17 standard. According to AES Measurement of digital audio equipment publication:

Overview: Performing a THD+N ratio measurement
The test signal shall be a sine wave. The analyzer shall include the standard low-pass filter. The total signal level at the EUT output is measured and recorded. The fundamental component of the test signal present in the output shall be removed with the standard notch filter, and the unweighted level of the residual shall be measured. The THD+N ratio shall be the difference between the residual level and the total signal level, expressed in dB
Thx Paul, but with measuring devices like the AP, a real notch filter is being used to remove the fundamental as above and in REW and Multitone, a mathematical calculation is being done. Do you do it the same way as REW, i.e. remove 1 octave around the fundamental?
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
Thx Paul, but with measuring devices like the AP, a real notch filter is being used to remove the fundamental as above and in REW and Multitone, a mathematical calculation is being done. Do you do it the same way as REW, i.e. remove 1 octave around the fundamental?

Funny, you're the second person in the last few days to bring up AES17 notch implementation. AP, while using a notch filter, is not exactly a standard since AES doesn't specify the exact parameters of the filter, giving a range of values. MT applies this in the frequency domain using 1 octave around the fundamental, so while this isn't going to match AP exactly, the result will be very similar.
 
  • Like
Reactions: pma

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
I have to think the reason for the notch filter, a time-honored device to extend the dynamic range of a measuring system be it audio, RF, or light, does not apply to a SW implementation? The dynamic range is set by the signal capture (ADC) so what is the benefit of applying a notch filter in SW in this application? Help to find the remaining spurs?

Curious - Don
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
I have to think the reason for the notch filter, a time-honored device to extend the dynamic range of a measuring system be it audio, RF, or light, does not apply to a SW implementation? The dynamic range is set by the signal capture (ADC) so what is the benefit of applying a notch filter in SW in this application? Help to find the remaining spurs?

Curious - Don

I'd say it's primarily for standardization of the measurement protocol than for an actual advantage. There's no advantage I can see to applying a simulated analog notch in the digital domain.
 

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,681
Likes
4,218
Location
Liège, Belgium
I have to think the reason for the notch filter, a time-honored device to extend the dynamic range of a measuring system be it audio, RF, or light, does not apply to a SW implementation? The dynamic range is set by the signal capture (ADC) so what is the benefit of applying a notch filter in SW in this application? Help to find the remaining spurs?

Curious - Don
You may want to have a look at this thread

In practice, the AES17 notch is usefull with digital measurement when the fundamental, through the DUT, an analog notch filter, or from the signal source, shows some spread around its base frequency.

In my case, that was due to the E1DA APU notch filter. I see that also when I use Victor's 1kHz oscillator.

If I measure my RME ADI-2 Pro in loopback mode, the results are identical with or without the AES17 digital notch.
Not at all with the APU or Victor's oscillator.

The AP also uses analog oscillator and notch for high accuracy measurements and, therefore, also requires the same.
I suspect the AES 17 standard includes this to allow use of such analyzers.

The question is: do they compensate for the narrower bandwidth in noise and THD+N calculations ?
Ok, we speak 0.3dB or so, but I'm curious.
As per the AES-17 standard definition, it doesn't seem so.
 
Last edited:

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,827
Likes
5,756
The question is: do they compensate for the narrower bandwidth in noise and THD+N calculations ?
Ok, we speak 0.3dB or so, but I'm curious.
As per the AES-17 standard definition, it doesn't seem so.
Yes,that's about the drop either in loopbacks or measuring other DACs,0.3-0.5db:


with.PNG


with


without.PNG


without
 

Sokel

Master Contributor
Joined
Sep 8, 2021
Messages
5,827
Likes
5,756
Ok,I finally measured my rig from side to side (6 meters cable from xover to ADC,everything as is) ,except power amps (be nice):


SINAD.PNG


SINAD at 1Khz for highs and 100Hz for lows.




FR.PNG


FR and slopes.

Sadly it's too hot for my ADC (see the -15db play gain).

What's impressive though,never been seen in any measurement is that everything can be visible at the same time combined thanks to @pkane:



All.PNG


(I honestly didn't expect it to work so smoothly in history)
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
A notch filter on the analog input makes sense to me as it lets you use more of the ADC's dynamic range. After conversion is when it makes less sense to me, where the signal is already captured and you can always process the data. A digital notch filter can suppress the fundamental but if the spreading around it was captured initially it will still be there unless you broaden the notch, which then suppresses the noise/sidebands along with everything else. Why not just zoom into the area you want to see and ignore the fundamental that way? Too much for my little pea brain to handle, I guess.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
A notch filter on the analog input makes sense to me as it lets you use more of the ADC's dynamic range. After conversion is when it makes less sense to me, where the signal is already captured and you can always process the data. A digital notch filter can suppress the fundamental but if the spreading around it was captured initially it will still be there unless you broaden the notch, which then suppresses the noise/sidebands along with everything else. Why not just zoom into the area you want to see and ignore the fundamental that way? Too much for my little pea brain to handle, I guess.

It makes sense to have an analog filter before the ADC. After the ADC conversion, an analog-type notch filter makes little sense, other than to be compatible with other measurement devices that use one before the ADC.
 

Moto

Active Member
Joined
Jan 19, 2021
Messages
180
Likes
41
So we see spreading at the foot of the fundamental with analog oscillators or analog notch filters but not with dacs. To try to compare results with a consumer adc to an AP it would make sense to use the aes17 digital notch feature in MultitoneAt 1khz that digital notch is not “too” wide but at 10khz , digitally notching out 1 octave seems crazy.
So is the digital 1 octave notch only useful at lower frequencies practically speaking?
I’m sort of back to my original question, what’s the use case for it.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
2 different issues.

1. notch to extract fundamental to get better distortion resolution
2. to eliminate spectrum broadening near fundamental to improve N
 

Moto

Active Member
Joined
Jan 19, 2021
Messages
180
Likes
41
2 different issues.

1. notch to extract fundamental to get better distortion resolution
2. to eliminate spectrum broadening near fundamental to improve N
Ok but does a one octave digital notch in REW or Multitone at 10khz make sense?
 

Moto

Active Member
Joined
Jan 19, 2021
Messages
180
Likes
41
But just by that greater absolute width, isn’t the absolute value of noise energy going to be more greatly reduced than in the 1khz case? So maybe ok for logarithmic calculations but not for absolute value calculations of noise.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
I needed a break so decided to implement the units for the level sweep (@pma). Unfortunately, the setting is still called 'vertical units' in settings :) There's also more to work on in the sweeps, but at least this part can be tested now:

https://app.box.com/s/ue7ll9xmvwogst817x2l1xg09opvgy47


1675820945726.png
 
Last edited:

Guermantes

Senior Member
Joined
Feb 19, 2018
Messages
484
Likes
561
Location
Brisbane, Australia
Ooh, sweeps. So will we be able to do room analysis soon? (only half joking)

This is looking fabulous @pkane -- I'm looking forward to comparing it to our Lindos measurement set.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,626
Likes
10,202
Location
North-East
But just by that greater absolute width, isn’t the absolute value of noise energy going to be more greatly reduced than in the 1khz case? So maybe ok for logarithmic calculations but not for absolute value calculations of noise.

Here's a real world comparison. The differences exist, but are very minor, and most within the measurement variations.

10k AES17 notch vs. 10k no notch, vs. 1k no notch, vs. 1k AES17 notch:

1675824084286.png
 
Last edited:
Top Bottom