Beta Test: Multitone Loopback Analyzer software

[Sokel]

Or with 1x multiplier?

With x1 multiplier sometimes seems like giving a large spike and lower frequencies are weird.

 

[Sokel]

Linear chirp is also weird:

-140dB harmonics for Khadas???

 

[pkane]

Linear chirp is also weird:

View attachment 507302
-140dB harmonics for Khadas???

Quite an improvement!

 

[pkane]

Linear chirp is also weird:

View attachment 507302
-140dB harmonics for Khadas???

Can you try choosing an FFT window (Blackman Harris 7)? With two independent clocks there’s little chance that ‘no window’ will work right.

 

[Sokel]

Can you try choosing an FFT window (Blackman Harris 7)? With two independent clocks there’s little chance that ‘no window’ will work right.

It used to work.
Blackman Harris 7 is even weirder:

 

[Rantapossu]

There still seems to be something wrong with the warm-up time:

3 seconds:

0.7 seconds:

 

[pkane]

There still seems to be something wrong with the warm-up time:

3 seconds:

View attachment 507346


0.7 seconds:

View attachment 507347

Is warm up time of 1 sec or less working? How about 1.5 or 2?

 

[Rantapossu]

Is warm up time of 1 sec or less working? How about 1.5 or 2?

0 seconds:

0.7 seconds:

1 second:

1.5 seconds:

2 seconds:

 

[pkane]

0 seconds:

View attachment 507350


0.7 seconds:

View attachment 507351


1 second:

View attachment 507352


1.5 seconds:

View attachment 507353


2 seconds:

View attachment 507354

Strangely I can't reproduce this. I get the same FR regardless of 1, 2 or 3 second warm up time. Can you please post a screenshot of the whole MTA window with the settings panel visible on the left?

 

[Rantapossu]

Strangely I can't reproduce this. I get the same FR regardless of 1, 2 or 3 second warm up time. Can you please post a screenshot of the whole MTA window with the settings panel visible on the left?

Sure, but it takes some time, because I won't be able to use my measuring equipment for a few hours.

 

[Sokel]

I tried 0.5 and seems ok:

What breaks things though is lowering FFT size and/or set the low freq limit to something other than 3Hz or so:

 

[Rantapossu]

Strangely I can't reproduce this. I get the same FR regardless of 1, 2 or 3 second warm up time. Can you please post a screenshot of the whole MTA window with the settings panel visible on the left?

It's the combination of the averages and the warm-up time!

You can replicate it by doing these steps:

- Delete the "_MultitoneDefault.dw" -file to set the defaults

- Start MultiTone

- Set these parameters:
FFT -> 256k
Window -> No window
Test signal -> Log-Chirp
Averages -> 128
Warm-up (sec) -> 3
Low Frequency limit -> 0
High Frequency limit -> 0

You can either keep the averages set as 1 and keep the 3 second warm-up or keep the averages set as 128 and set the warm-up time to 0 second, but you can't have the both...

 

[pkane]

It's the combination of the averages and the warm-up time!

You can replicate it by doing these steps:

- Delete the "_MultitoneDefault.dw" -file to set the defaults

- Start MultiTone

- Set these parameters:
FFT -> 256k
Window -> No window
Test signal -> Log-Chirp
Averages -> 128
Warm-up (sec) -> 3
Low Frequency limit -> 0
High Frequency limit -> 0

You can either keep the averages set as 1 and keep the 3 second warm-up or keep the averages set as 128 and set the warm-up time to 0 second, but you can't have the both...

Thanks, I'll give this a try. Strange that averages has an effect, as it's using a single average internally... or at least that's what I thought

 

[IsaacOscar]

My apologies if this is already covered in this thread (it's a very long one!!)
But I've noticed that the frequency response generated by Multitone is much noisier than one generated by Room EQ Wizard:
(I separated the green line for visibility, usually they both line up perfectly)

(I had Equalizer APO running, my output and input devices are connected via a simple cable)
My Multitone settings:

My Room EQ Wizard Settings:

 

[Sokel]

My apologies if this is already covered in this thread (it's a very long one!!)
But I've noticed that the frequency response generated by Multitone is much noisier than one generated by Room EQ Wizard:
(I separated the green line for visibility, usually they both line up perfectly)
View attachment 508020
(I had Equalizer APO running, my output and input devices are connected via a simple cable)
My Multitone settings:
View attachment 508024

My Room EQ Wizard Settings:
View attachment 508022View attachment 508023

Are you using the classic sweep at REW?

Better compare with a log sweep through REW generator and RTA, it's also a little more noisy.

 

[morillon]

It's an interesting and delicate subject...because it allows us to extract sweep distortion measurements in acoustics in a more or less relevant way. ;-)
(rew also allows you to multiply the sweeps for these uses... ;-) )

 

[IsaacOscar]

Are you using the classic sweep at REW?

Better compare with a log sweep through REW generator and RTA, it's also a little more noisy.

According to the tooltip I am using a logarthmic one:

 

[IsaacOscar]

It's an interesting and delicate subject...because it allows us to extract sweep distortion measurements in acoustics in a more or less relevant way. ;-)
(rew also allows you to multiply the sweeps for these uses... ;-) )

Hmm so you're saying that Multitone is actually more accurate here?
This wasn't actually bothering me much as I just applied a 1/48 smoothing to the graph in REW. But it made me worry about the accuracy of my measurements (which are terrible anyway, given the low-quality ADC I'm using).

 

[morillon]

Hmm so you're saying that Multitone is actually more accurate here?
This wasn't actually bothering me much as I just applied a 1/48 smoothing to the graph in REW. But it made me worry about the accuracy of my measurements (which are terrible anyway, given the low-quality ADC I'm using).

No... if MT shows itself to be "noisier"... it will have more difficulty extracting high-order distortion profiles.
But if you're not looking in that direction and just FR observation, it's not a big deal...
(On the headphone measurement bench, it's the lack of a way to differentiate between left and right channels that's a bit of a shame and requires some extra work. ;-) )

 

[IsaacOscar]

No... if MT shows itself to be "noisier"... it will have more difficulty extracting high-order distortion profiles.
But if you're not looking in that direction, it's not a big deal...

So REW isn't doing any internal smoothing? The difference is due to multitone's method of measurement being inherently "noisier"?
(I wasn't trying to measure distortion, just PEQ behaviour).