• 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). There are daily 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

Sokel

Major Contributor
Joined
Sep 8, 2021
Messages
1,304
Likes
905
Take a look,9/21 and it still measures 2Hz.

9-21.PNG
 
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
Take a look,9/21 and it still measures 2Hz.

View attachment 232325
This is still OK. The question is, does it finish? There's not much room to do a log sweep between 0 and 31Hz, so only a few measurements will fit in that space. Those that are too close to each other will be skipped, so jumping from 1 to 9 isn't unexpected.
 

Sokel

Major Contributor
Joined
Sep 8, 2021
Messages
1,304
Likes
905
This is still OK. The question is, does it finish? There's not much room to do a log sweep between 0 and 31Hz, so only a few measurements will fit in that space. Those that are too close to each other will be skipped, so jumping from 1 to 9 isn't unexpected.
No,it only jumps from 1 to 4.
From then on measures 2Hz,finishes and starts over at 2 Hz until 21.
It measures 2Hz 16 times and every time it finishes.
It doesn't measure anything else.maybe it likes 2Hz.
 
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
No,it only jumps from 1 to 4.
From then on measures 2Hz,finishes and starts over at 2 Hz until 21.
It measures 2Hz 16 times and every time it finishes.
It doesn't measure anything else.maybe it likes 2Hz.
Hmm, strange. Seems to work differently for me with exactly the same settings. I'll do another quick check, but I think that's an extreme case that isn't worth spending too much time on.
 

Sokel

Major Contributor
Joined
Sep 8, 2021
Messages
1,304
Likes
905
Hmm, strange. Seems to work differently for me with exactly the same settings. I'll do another quick check, but I think that's an extreme case that isn't worth spending too much time on.
Ok,cleaned up everything and now measures 2Hz until step 20 (from 21) where it jumps to 26Hz and then to 31Hz which is the selected end freq.

31.PNG
 

Rantapossu

Active Member
Joined
Jul 21, 2022
Messages
261
Likes
190
No,it only jumps from 1 to 4.
From then on measures 2Hz,finishes and starts over at 2 Hz until 21.
It measures 2Hz 16 times and every time it finishes.
It doesn't measure anything else.maybe it likes 2Hz.

The behaviour is exactly the same when I'm measuring the crosstalk with just one channel:

It tries 40 Hz over and over again:

1661186762821.png


1661186810650.png


1661186844205.png



Eventually it manages to get the measurement at the higher frequencies and plots it.

1661186994954.png


There are a lot of skipped frequencies on the plot when it's ready.

This is because of the "No signal" -detection and when MT doesn't get the strong enough signal, it just tries again when sweeping.

I think that this is actually a bug because it should jump to the next frequency and not just repeatedly try the same one again and again.

Actually I'm not even sure if it just prints the wrong frequency during the sweep or actually measures the wrong frequency.

It would be ok to try the same signal again too if the previous result is not ok , but then it shouldn't increase the sweep counter.

@pkane could the "No signal" detection be disabled on the settings when needed? Sometimes I'll have to use the second channel with full signal just to get past the "No signal" -detection and then hide the second channel plot when the measurement is ready. Disabling the limiter would make some measurements much easier.
 
Last edited:
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
The behaviour is exactly the same when I'm measuring the crosstalk with just one channel:

It tries 40 Hz over and over again:

1661186762821.png


1661186810650.png


1661186844205.png



Eventually it manages to get the measurement at the higher frequencies and plots it.

1661186994954.png


There are a lot of skipped frequencies on the plot when it's ready.

This is because of the "No signal" -detection and when MT doesn't get the strong enough signal, it just tries again when sweeping.

I think that this is actually a bug because it should jump to the next frequency and not just repeatedly try the same one again and again.

Actually I'm not even sure if it just prints the wrong frequency during the sweep or actually measures the wrong frequency.

It would be ok to try the same signal again too if the previous result is not ok , but then it shouldn't increase the sweep counter.

@pkane could the "No signal" detection be disabled on the settings when needed? Sometimes I'll have to use the second channel with full signal just to get past the "No signal" -detection and then hide the second channel plot when the measurement is ready. Disabling the limiter would make some measurements much easier.
That sounds like a noisy result. When measuring sweeps, MT can skip a number of measurements that seem noisy. I think I’ll disable that, and just display whatever result is measured.
 

Rantapossu

Active Member
Joined
Jul 21, 2022
Messages
261
Likes
190
That sounds like a noisy result. When measuring sweeps, MT can skip a number of measurements that seem noisy. I think I’ll disable that, and just display whatever result is measured.

Thanks!

Is your "Noisy signal" -detector different than the famous "No signal" -detector? When doing the crosstalk measurements the signal is still clear (Especially at 1 kHz), but it's level is under the "No signal" detection threshold.
 
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
Thanks!

Is your "Noisy signal" -detector different than the famous "No signal" -detector? When doing the crosstalk measurements the signal is still clear (Especially at 1 kHz), but it's level is under the "No signal" detection threshold.

Different.
 

Rantapossu

Active Member
Joined
Jul 21, 2022
Messages
261
Likes
190
Different.

Ok. Disabling the "Noisy signal" -detector might not be enough to make the crosstalk measurements if the -120 dBFS "No signal" -detector is still active.

But it can sure help @Sokel with his measurements.
 
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
To promote peace and harmony, and in the interest of relaxing some of the remaining hard limits in Multitone, I've changed the sweeps to:
  • Not to skip measurements, no matter how noisy
  • Not to skip log-based frequency measurement if the previous one was more than 0.1Hz away from this one
  • @Blumlein 88 : also please check the level meter jumps below 5Hz. Do you still see them?
https://app.box.com/s/qbrh3czrvudclkqs9n8lm806k3hzmjx5

You can see some of this reflected in this sweep:

1663800484004.png
 
Last edited:

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
15,915
Likes
26,714
Yes the level readout of the green line is steady down to 2 hz. Just a very minor fluctuation at 1 hz, but essentially looks like you have solved that problem. Also here is a frequency sweep 1 to 21 hz. 10 steps.

1663820608516.png


Here is one 0 to 20 hz 10 steps. Looks like it does not graph the 1st step. Or maybe it simply isn't getting a reading which is more likely. I'm thinking maybe this one should only allow setting down to 1 hz and just leave off the 0 hz.

1663820821987.png
 

Rantapossu

Active Member
Joined
Jul 21, 2022
Messages
261
Likes
190
To promote peace and harmony, and in the interest of relaxing some of the remaining hard limits in Multitone, I've changed the sweeps to:
  • Not to skip measurements, no matter how noisy
  • Not to skip log-based frequency measurement if the previous one was more than 0.1Hz away from this one
  • @Blumlein 88 : also please check the level meter jumps below 5Hz. Do you still see them?
https://app.box.com/s/qbrh3czrvudclkqs9n8lm806k3hzmjx5

You can see some of this reflected in this sweep:

View attachment 232419

Thanks!

The crosstalk measurement is now possible with one channel too (Without using the other channel to just trick the "No signal" -detection):

1663848877125.png
 

Sokel

Major Contributor
Joined
Sep 8, 2021
Messages
1,304
Likes
905
I took Amir's job :facepalm:

 
OP
pkane

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
4,062
Likes
6,910
Location
North-East
For anyone interested, I updated v1.0.49 to generate frequency-modulated test tones, per @pma 's idea for amplifier testing. This can be used in place of multitones, and is a better test signal than even the decorrelated/crest-optimized multitones. This isn't good for testing frequency response, but to look at distortions and noise-floor modulation, this can be hard to beat:

https://app.box.com/s/qbrh3czrvudclkqs9n8lm806k3hzmjx5

There are no FM test tones in the list, but you can type one in directly. Just type this, replacing the three frequencies with the desired numbers:

FM fc/fm/fd

fc = carrier frequency fm = modulation frequency fd = deviation span frequency

For example, to use Pavel's test tone of 1kHz carrier, 200Hz modulation frequency and 900Hz deviation, you'd type:

FM 1k/200/900

As simple as that. You can alter any of the frequencies, of course. For example, to get a set of about 300 test tones spanning 40Hz to a 11kHz, enter this:

FM 1k/40/10k

And of course, you can alter the carrier center frequency to anything desired. For example:

FM 500/20/500

Curious to see what results others will get. Be careful feeding a 0dBFS signal directly into the amp, though, as it is likely to clip or overload due to the low crest factor in this test signal. You can always adjust the Play Gain to lower the level.

1663885819934.png
 
Last edited:

Sokel

Major Contributor
Joined
Sep 8, 2021
Messages
1,304
Likes
905
For anyone interested, I updated v1.0.49 to generate frequency-modulated test tones, per @pma 's idea for amplifier testing. This can be used in place of multitones, and is a better test signal than even the decorrelated/crest-optimized multitones. This isn't good for testing frequency response, but to look at distortions and noise-floor modulation, this can be hard to beat:

https://app.box.com/s/qbrh3czrvudclkqs9n8lm806k3hzmjx5

There are no FM test tones in the list, but you can type one in directly. Just type this, replacing the three frequencies with the desired numbers:

FM fc/fm/fd

fc = carrier frequency fm = modulation frequency fd = deviation span frequency

For example, to use Pavel's test tone of 1kHz carrier, 200Hz modulation frequency and 900Hz deviation, you'd type:

FM 1k/200/900

As simple as that. You can alter any of the frequencies, of course. For example, to get a set of about 300 test tones spanning 40Hz to a 11kHz, enter this:

FM 1k/40/10k

And of course, you can alter the carrier center frequency to anything desired. For example:

FM 500/20/500

Curious to see what results others will get. Be careful feeding a 0dBFS signal directly into the amp, though, as it is likely to clip or overload due to the low crest factor in this test signal. You can always adjust the Play Gain to lower the level.

View attachment 232636
I don't know what it does to amps but E-MU seems to like it.
fm.PNG

Pavel's


fm2.PNG


your example.

Pfff...thought I would explode something :cool:

Edit:

In dbfs probably gives near the best results I get with E-MU (despite the modulation down there)
Normal multitone is much harder...


fm3.PNG
 
Last edited:

Rantapossu

Active Member
Joined
Jul 21, 2022
Messages
261
Likes
190
I can't get a decent result with 8 (or 12, 16 and many other) averages. The last average always spoils it:

1663958587695.png


7 averages works all the time:

1663958655893.png


Is the test signal too short?
 
Top Bottom