Can't display distortion harmonics from .mdat files in temp folder?

[neodog]

When I open an .mdat from the save folder, and the .mdat from the exact same measurement from the temp folder, I see different data in the distortion pane. In the temp file it looks like only the fundamental and the THD plot is displayed. One side-effect of this is that I don't know how to view distortion data for files that were recovered after an unexpected shutdown. Is there a way to get them to present?

From the save folder:

From the temp folder:

 

[John Mulcahy]

Please attach a file that shows that behaviour.

 

[neodog]

The numbered file doesn't show the harmonics when loaded, the named file is a version of the same data that does show them (the same datasets pictured above)

 

[John Mulcahy]

That's strange. It looks like the temporary copy was saved while the distortion calculation was still in progress, so only capturing partial results. I can't replicate that, but I can see how it could be possible. I'll fix it in the next build.

 

[neodog]

Nice. FWIW, it seems to have happened for every one of the temp files, so it seems I'll be able to confirm the fix pretty reliably.

Is there a way to trigger these calculations again, or do I just need to reperform the measurements?

(Edit: I tried exporting and importing the measurements, it doesn't look like distortion data comes along or gets re-computed that way... seems like the raw sweep recording would be necessary and that maybe isn't saved in the .mdat file)

 

[John Mulcahy]

The distortion data is in the negative time part of the IR, a long way back for long sweeps. That data is discarded once the calculation has happened.

 

[neodog]

I've been struggling to understand statements about distortion being at times t<0, since it's counterintuitive... i.e. surely distortion occurs after the sound that inspired it. But I think this must be a characteristic of the swept impulse analysis approach, wherein the response is analyzed at each frequency slice, allowing that a frequency component can be measured before the tone at that frequency is actually played, which is an unmistakable indication that the fundamental tone didn't cause the sound. Since the sweep goes from lower to higher frequencies, we can presume such sounds are a harmonic distortion of a lower frequency (although, it might be a convenience to presume that the distortion is of the harmonic sort, since others are still possible?).

Is it fair to say that distortion can continue at t>0, but at that point it is very difficult to distinguish from the Fundamental, and that REW doesn't even try? Hence the need to perform a slower sweep if distortion is protracted...

 

[John Mulcahy]

Longer sweeps deliver more total energy and so increase signal to noise. Deconvolution (time domain) or division (frequency domain) of the response by the stimulus causes frequencies that appear in the response before they appear in the stimulus to be moved back into negative time. The benefit of the exponential sweep is that the shift depends on the frequency ratio, so all the harmonics end up clustered individually at progressively more negative time. The linear response of the system is at positive time.