Beta Test: Multitone Loopback Analyzer software

[Sokel]

Just keep running 16M FFTs, it'll die in no time

I actually did a 32M FFT, as well. The first time took about 5 minutes to complete, the second time only about 15 seconds.

I already do,i have a log-chirp 0-60Hz in the background for 5min now,it will end tomorrow probably

 

[pkane]

You have no phase control. The correct waveform would look like this in MT (I dragged the file onto it, stereo channels intentionally phase inverted):

View attachment 259297

So I added the option to specify phase of individual sine components in the test signal. With this you should be able to construct any of the sine-wave inter-sample over tests in the paper.

This way:

Or this one:

You can even specify phase for multiple tone test signal:

 

[MC_RME]

Seems not fully correct. For example the 11.025 kHz file must give 0 dBFS peak level and look like this:

Interesting how different it looks to when my old WaveLab renders the file:

In the same way WaveLab shows the other example (7350 Hz posted by me) as full square wave, with a step.

 

[pkane]

Seems not fully correct. For example the 11.025 kHz file must give 0 dBFS peak level and look like this:

View attachment 259425

Interesting how different it looks to when my old WaveLab renders the file:

View attachment 259426

In the same way WaveLab shows the other example (7350 Hz posted by me) as full square wave, with a step.

View attachment 259427

Ah, yes, I see I missed one place where the phase needed to be included in the computation.

The 11025Hz@45 looks like this now (a 3dB play gain is required, since Multitone test signal specifies sine amplitude and 0dBFS would not generate oversample overs!):

 

[pkane]

In the same way WaveLab shows the other example (7350 Hz posted by me) as full square wave, with a step.

View attachment 259427

That 7350Hz plot seems wrong in WaveLab. It shows 8 samples per cycle. At 7350Hz this would make the sampling rate 7350x8 = 58.8k???

Here's what the one from MT looks like (6 samples per cycle = 7350x6 = 44100Hz):

The 11025Hz one also looks wrong in WaveLab. It shows 0 time interval between two samples in the vertical transition. That can't be right, as that is a square wave, not a sine wave. A sine wave will have four equally spaced samples. Like this:

 

[MC_RME]

Wavelab shows 6 samples as well...it just connects the samples differently. Do you have Reaper? There you can see both versions: WaveLab style with less zoom, and as soon as zoom is increased so that Reaper marks the sample points it turns to your version

 

[MC_RME]

And here is the 11.025 in sample view again. I think I understand the other effect, but not this one - how/why does MT show skewing/slope?

 

[MC_RME]

And here is the 11.025 in sample view again. I think I understand the other effect, but not this one - how/why does MT show skewing/slope?

View attachment 259445

I think I got it...WaveLab does not show the real sample values at all points. The way its display works any sample on an edge could be either the lowest or highest value. You perform a direct connection between the samples which gives/keeps the right values for any given sample.

 

[pkane]

I think I got it...WaveLab does not show the real sample values at all points. The way its display works any sample on an edge could be either the lowest or highest value. You perform a direct connection between the samples which gives/keeps the right values for any given sample.

Exactly what I was thinking

 

[MC_RME]

I downloaded the latest version (1.0.67) but I only see FFT up to 4M. Where can I select 32M?

Select the FFT Size field and type 32M...

 

[Blumlein 88]

BTW zooming to +-5Hz the horizontal scale does not look altogether correct.

View attachment 259495

Paul will reply I am sure. Yes, what you are seeing is something of a bug if you zoom in far enough. I don't recall, but basically I think Paul said it would require a considerable rewrite and those levels of zoom in are uncommonly needed.

 

[MC_RME]

And I fully agree with that. No information gained at <1 Hz per Div.

 

[Blumlein 88]

Also if you back up the zoom level just before it looks like that, you can then use the SHIFT key+left click on the mouse to create a pointer which will show you hz down to several decimal places.

 

[pkane]

Well, I have to disagree with that. Maybe not needed at 12kHz but definitely useful let's say at 20Hz. Similar scaling at e.g. 30Hz seems to work slightly better.

It should be easy to add a few decimal places to the axis labels. Not sure why it’s important, but it can be done.

 

[pkane]

A new test version 1.0.68: https://app.box.com/s/ue7ll9xmvwogst817x2l1xg09opvgy47

Many changes in this one, some major, so please test:

• Add: support for phase specification in test signal format
• Change: better label selection for linear frequency axis
• Change: add a significant digit to frequency axis labels
• Change: bandwidth selector changed to work consistently across all chart types
• Change: allow mixing of different sampling rates and FFT sizes from history on the same plot

For those wanting to test intersample overs, here are three presets (I'm not including these in the installer):

https://app.box.com/s/w5sehf9ow20ym5inh1tm3h2l010w1mcw

Download the zip file and extract the three preset files. Copy them to C:\ProgramData\Multitone Analyzer\Presets folder on your computer. Then, restart Multitone -- the three Intersample presets will show up in the drop down, in addition to all the ones included in the installer:

 

[Sokel]

A new test version 1.0.68: https://app.box.com/s/ue7ll9xmvwogst817x2l1xg09opvgy47

Many changes in this one, some major, so please test:

• Add: support for phase specification in test signal format
• Change: better label selection for linear frequency axis
• Change: add a significant digit to frequency axis labels
• Change: bandwidth selector changed to work consistently across all chart types
• Change: allow mixing of different sampling rates and FFT sizes from history on the same plot

For those wanting to test intersample overs, here are three presets (I'm not including these in the installer):

https://app.box.com/s/w5sehf9ow20ym5inh1tm3h2l010w1mcw

Download the zip file and extract the three preset files. Copy them to C:\ProgramData\Multitone Analyzer\Presets folder on your computer. Then, restart Multitone -- the three Intersample presets will show up in the drop down, in addition to all the ones included in the installer:

View attachment 259521

You're great,thanks!

 

[Sokel]

Does this looks ok?

 

[pkane]

Does this looks ok?


View attachment 259530

What does the spectrum look like?

 

[Sokel]

What does the spectrum look like?

Like this:

 

[pkane]

Like this:

View attachment 259532

Looks like a raised noise floor, but overall, doesn't look that bad. Then again, I'm no expert on intersample-overs