Two-tone IMD bug?

[siarez]

In two-tone IMD measurements, REW frequently fails to detect D3L, D3H and beyond. For example, in the attached image, D3L (7kHz) is clearly above the noise floor, but is not detected.

 

[trobbins]

Is that a custom setup, as default DIN IMD is 250Hz and 8kHz, and the IMD components are 250Hz sidebands of 8kHz ?

 

[John Mulcahy]

The definition of the higher order difference frequencies differ for different signals. For SMPTE, DIN and AES-17 MD d3L = f2 - 2f1. For CCIF and AES-17 DFD d3L = 2f1 - f2 (the various frequencies are tabulated in the help). For custom signals REW uses the DIN/SMPTE definition if f2/f1 is > 7, a somewhat arbitrary threshold, which isn't the case for the signal above. I could change that threshold of course, or use the amplitude ratio instead since modulation distortion signals are usually 4:1. Happy to hear suggestions.

 

[siarez]

@troscat Yes it is a custom setup (screenshot below)
@John Mulcahy Thanks for clarifying. For my use case, treating it like SMPTE/DIN works, so lowering the threshold would do the job.

 

[John Mulcahy]

I've changed it to use the amplitude ratio for the next build, probably safer than basing it on the frequencies.

 

[siarez]

@John Mulcahy I actually did get what you meant by "using amplitude ratios". But as long as it calculates percentage f2 ± 2f1, f2 ± 3f1, ... , I'll be happy