Beta Test: Multitone Loopback Analyzer software

[Blumlein 88]

Okay, the display glitch as posted in screen shots above. If I have both devices set for 48 khz, and upper frequency set for 20 khz. I can change the recording device to 192 khz and upper frequency set to 96,000 hz it runs fine. If I then highlight a portion with a center click and drag over just the frequency axis it shows what I highlighted. If I run it again after changing FFT size or sample rate the display scales with the changes. Hitting "reset axis" fixes this. So that makes some sense.

Now the other glitch. If I reset sample rates low like 48 khz or 44.1 khz, while leaving the upper frequency limit at 96 khz. It will run properly at all FFT sizes except 1 million. When set to 1 million it works up until it shows processing, and then closes. If I lower the upper frequency to anything lower than 96,000 hz even to just 88,200 hz it will work. But at 96 khz upper frequency limit or higher it closes. Now it makes no sense to set the upper frequency above what the chosen sample rate will record. It was just an accident I didn't remember to reset it. Maybe the upper frequency limit needs to check the recording device sample rate and not allow those higher settings.

 

[bennetng]

Mine got broken too in the select with the latest .
The fame came to the standard levels

View attachment 220792

Very strange because the problem is fixed on my machine when using v1.0.28

 

[Sokel]

Very strange because the problem is fixed on my machine when using v1.0.28
View attachment 220840

Unistalled it,cleaned up registry and everything and made a clean install.
No fix,the same .It's only cosmetic though,so...

Okay, the display glitch as posted in screen shots above. If I have both devices set for 48 khz, and upper frequency set for 20 khz. I can change the recording device to 192 khz and upper frequency set to 96,000 hz it runs fine. If I then highlight a portion with a center click and drag over just the frequency axis it shows what I highlighted. If I run it again after changing FFT size or sample rate the display scales with the changes. Hitting "reset axis" fixes this. So that makes some sense.

Now the other glitch. If I reset sample rates low like 48 khz or 44.1 khz, while leaving the upper frequency limit at 96 khz. It will run properly at all FFT sizes except 1 million. When set to 1 million it works up until it shows processing, and then closes. If I lower the upper frequency to anything lower than 96,000 hz even to just 88,200 hz it will work. But at 96 khz upper frequency limit or higher it closes. Now it makes no sense to set the upper frequency above what the chosen sample rate will record. It was just an accident I didn't remember to reset it. Maybe the upper frequency limit needs to check the recording device sample rate and not allow those higher settings.

I tried to replicate it,it gets really slow at 1M,that's true,but eventually it finishes.

 

[Rantapossu]

Would it be possible to add a dBr option for the frequency response graph too? Like this (PhotoShopped):

If I'm right, it would enable the channel separation measurements when playing the one channel and recording with the other (L and R or R and L). The channel leakage would be visible in the multitone measurements:

Frequency response now:

Frequency response after disabling the dBr could be something like this (PhotoShopped):

This would be super, if it's possible to make!

 

[pkane]

Okay, the display glitch as posted in screen shots above. If I have both devices set for 48 khz, and upper frequency set for 20 khz. I can change the recording device to 192 khz and upper frequency set to 96,000 hz it runs fine. If I then highlight a portion with a center click and drag over just the frequency axis it shows what I highlighted. If I run it again after changing FFT size or sample rate the display scales with the changes. Hitting "reset axis" fixes this. So that makes some sense.

Now the other glitch. If I reset sample rates low like 48 khz or 44.1 khz, while leaving the upper frequency limit at 96 khz. It will run properly at all FFT sizes except 1 million. When set to 1 million it works up until it shows processing, and then closes. If I lower the upper frequency to anything lower than 96,000 hz even to just 88,200 hz it will work. But at 96 khz upper frequency limit or higher it closes. Now it makes no sense to set the upper frequency above what the chosen sample rate will record. It was just an accident I didn't remember to reset it. Maybe the upper frequency limit needs to check the recording device sample rate and not allow those higher settings.

I totally understand why the display shrinks when you set higher FFT rates. Multitone uses FFT bins as the unit on the horizontal axis, internally. When you've zoomed in to a portion of the plot, it remembers how many FFT bins fit on the horizontal axis. Then, when you switch to a higher FFT rate, the number of bins stays the same... but the frequency range displayed gets smaller due to smaller bins. Easy enough to fix by pressing Reset Axis but I can make this automatic, so you don't have to.

I'll check at higher FFT sizes, but I suspect the app is just running out of memory. What settings did you have (play/record rate, FFT size, averages, and average overlap) and how much memory is available on this PC?

 

[pkane]

Unistalled it,cleaned up registry and everything and made a clean install.
No fix,the same .It's only cosmetic though,so...


I tried to replicate it,it gets really slow at 1M,that's true,but eventually it finishes.

Strange.

 

[pkane]

Would it be possible to add a dBr option for the frequency response graph too? Like this (PhotoShopped):

View attachment 220846

If I'm right, it would enable the channel separation measurements when playing the one channel and recording with the other (L and R or R and L). The channel leakage would be visible in the multitone measurements:

View attachment 220847


Frequency response now:

View attachment 220848


Frequency response after disabling the dBr could be something like this (PhotoShopped):

View attachment 220849

This would be super, if it's possible to make!

I can add this option for the frequency response graph, no problem. I'm not sure how what you propose would measure channel separation, since the first channel selector applies to the output device (DAC) and the second one to the recording device, ADC. You'd need a two-channel recording to compare the two channels in the output device (which is something I may add in the near future) to measure this.

 

[Sokel]

What is that test which I get 0,0 ?
Jitter?Is it that bad?

 

[pkane]

Unistalled it,cleaned up registry and everything and made a clean install.
No fix,the same .It's only cosmetic though,so...

I wonder if this has to do with higher DPI setting on your PC.

What is that test which I get 0,0 ?
Jitter?Is it that bad?

View attachment 220927

You've excluded the measurement of the main (fundamental) tone by selecting only frequencies between 20 and 20k (the white part) -- the main tone is at 24k. That's outside the measurement frequency range. And J-test isn't meant for measuring TD+N, so that display isn't very useful in any case.

As far as jitter, there is some on the plot shown as side-bands around the fundamental. If you zoom in around the fundamental, you can see the level of jitter-caused distortions as the amplitude of these side-bands. Seems fairly low, so shouldn't be a problem but it's hard to see the actual amplitude without zooming in.

 

[Sokel]

Ok,looking good now.

Thank you!

 

[Sokel]

I wonder if this has to do with higher DPI setting on your PC.

Where do I change that to test?
The only dpi I see handy is the mouse one.

 

[Rantapossu]

I can add this option for the frequency response graph, no problem. I'm not sure how what you propose would measure channel separation, since the first channel selector applies to the output device (DAC) and the second one to the recording device, ADC. You'd need a two-channel recording to compare the two channels in the output device (which is something I may add in the near future) to measure this.

Thanks!

In my case both DAC and ADC are in the same device and the amplifier, the cable or anything else I want to measure, is between the outputs and inputs of my DAC/ADC.

When I feed 0 dBFS signal to the left output channel at the certain frequency and measure -80 dBFS signal at the right input channel, I know the amount of leak at that frequency.

When I do this without connecting anything else than a quality cable between inputs and outputs, I have a baseline measurement for my system. Then I switch the cable to the worse one or add a amplifier between the outputs and inputs of my system and measure again to see if the measurement gets worse.

Of course the frequency response won't be a straight line (Like it isn't in my L->R measurement graph above, there is over 20 dB difference between certain frequencies), but I will at least know if the cable or amplifier makes the channel separation worse.

 

[Sokel]

A,ok,got it.
It's higher,yes.117%.
I'll test and report.

 

[Sokel]

Nope,it's the same with 100%
(I'll change back now because I write almost blind )

 

[pkane]

Nope,it's the same with 100%
(I'll change back now because I write almost blind )

View attachment 220930

Can you please uninstall this version (from Windows Settings/control panel) and re-download the installer from the website, then re-install? I see the small glitch that was fixed in the latest update that isn't fixed in your screen shot (the recording device name is truncated and stretched too far left.) Not that it's important, but there was a fix for it yesterday

 

[pkane]

Thanks!

In my case both DAC and ADC are in the same device and the amplifier, the cable or anything else I want to measure, is between the outputs and inputs of my DAC/ADC.

When I feed 0 dBFS signal to the left output channel at the certain frequency and measure -80 dBFS signal at the right input channel, I know the amount of leak at that frequency.

When I do this without connecting anything else than a quality cable between inputs and outputs, I have a baseline measurement for my system. Then I switch the cable to the worse one or add a amplifier between the outputs and inputs of my system and measure again to see if the measurement gets worse.

Of course the frequency response won't be a straight line (Like it isn't in my L->R measurement graph above, there is over 20 dB difference between certain frequencies), but I will at least know if the cable or amplifier makes the channel separation worse.

Oh, I see, you want to measure crosstalk between channels. I thought you wanted to check imbalance.

 

[Sokel]

Can you please uninstall this version (from Windows Settings/control panel) and re-download the installer from the website, then re-install? I see the small glitch that was fixed in the latest update that isn't fixed in your screen shot (the recording device name is truncated and stretched too far left.) Not that it's important, but there was a fix for it yesterday

Yep,that did it.
Strange,because I already used yersteday's version which I installed twice (see the screenshot above,says 1.0.28 )
All fine now!

 

[Rantapossu]

Oh, I see, you want to measure crosstalk between channels. I thought you wanted to check imbalance.

Yep, thats what I meant!

It would be very useful addition to the program to be able to measure the crosstalk.

I have a hunch that I didn't explain my earlier request about ASIO channel selector clear enough too...

I didn't requested multichannel playing and recording (But that would be cool too!), two channels is enough for me now.

I requested the ability to select which one the four ASIO inputs of my device is the left channel and which one is the right channel. Same with the outputs.

Multitone uses the first ASIO channel as the left channel and the second ASIO channel as the right channel, those are analog in my case. Third is in my case the left SPDIF and the fourth is the right SPDIF. I can't access those channels at all without some kind of ASIO wrapper when using Multitone, only analog inputs and outputs.

I'd like to use the a native ASIO optical output of my device and a separate DAC to break the ground loop while measuring expensive amplifiers, just in case.

 

[Blumlein 88]

I totally understand why the display shrinks when you set higher FFT rates. Multitone uses FFT bins as the unit on the horizontal axis, internally. When you've zoomed in to a portion of the plot, it remembers how many FFT bins fit on the horizontal axis. Then, when you switch to a higher FFT rate, the number of bins stays the same... but the frequency range displayed gets smaller due to smaller bins. Easy enough to fix by pressing Reset Axis but I can make this automatic, so you don't have to.

I'll check at higher FFT sizes, but I suspect the app is just running out of memory. What settings did you have (play/record rate, FFT size, averages, and average overlap) and how much memory is available on this PC?

Play and record was 48 khz on both devices. Frequency range was set to 20 and 96,000 hz. FFTs of smaller than 1 million work fine. 8 averages, 87.5% overlap. PC has 8 gig of memory. Watching the resource monitor, memory use never exceeds 57%. CPU never exceeds 12% though CPU frequency does max out at 3.2 ghz for this machine on the more intensive parts of the process. It also does this with smaller FFTs that go on to complete the measurement. Everything looks fine then a couple or three seconds after recording has finished and it is Processing it just closes.

 

[Blumlein 88]

I repeated with all the same settings. With averages set to 1 or 2 it finishes. Averages of 3 or higher it closes the software at the end. So if no one else is having the same issue guess it is my old PC. Still might be a good thing for it to only allow an upper frequency selection within the ability for the sample rate chosen for the recording device.