IMD: module and phase info

[pinox67]

Hi,
A question about the information reported by the RTA.
When I use a Dual Tone test signal (SMPTE, DIN, ... or Custom) is there any way of also detecting the phase information of each intermodulation product via the RTA, similar to what the RTA always does with a single-tone signal?

 

[John Mulcahy]

Is there a fixed phase relationship between the intermod products and something else? For a tone the harmonic phases are shown relative to the fundamental.

 

[pinox67]

In the input signal, the two fundamentals have the same phase. In the output they may have different phases due to linear alterations.
It could be left to the user to choose which of the two fundamentals to use as a reference for intermodulation products and the other fundamental itself. Perhaps this may not seem like very useful information; in reality, the utility is in modelling the device behaviour with non-linear models.

 

[John Mulcahy]

In the input signal, the two fundamentals have the same phase.

I don't see how. They may both start out at t=0 with zero phase but there is no fixed relationship between the signals for any subsequent time segment.

the utility is in modelling the device behaviour with non-linear models

That seems far too narrow a use case to put such a task ahead of the very many others on my todo list, sorry. Probably easier to capture the signals and do the analysis in Matlab.

 

[pinox67]

I think the measurement procedure could be this:

1. Given in the input with two tones (f1, f2) at the same level and phase, it is captured the output by the device for a time window, let's say of 1M samples, with fs large enough to include the higher intermodulation product.
2. The samples are subjected to a Windowing function (e.g. Hann or Blackman-Harris 7) and then it is evaluated the FFT.
3. The levels of intermodulation products at frequencies af1+/-bf2 is detected. The related phases should be corrected to refer to those of the fundamentals f1 or f2, taking into account the frequency difference.

Is this correct?
Also, how can the Spectral Leakage phenomenon due to windowing and sampling be taken into account when detecting the levels of intermodulation products?
I think this effect also occurs in RTA detecting the levels of harmonics for single tones, right?
Can coherent averaging of multiple time window samples help?

 

[John Mulcahy]

What are the relative phases of f1 and f2 in a segment captured after an unknown time delay?

 

[pinox67]

Ok, since the time delay is unknown and the output is captured 'randomly', the phases of f1 and f2 can be any.
Let's say that through other measurements I was able to get an estimate of the Time Delay, to be incorporated into the phase correction in step 3.
Are there any other caveats? How do you deal with the other aspects I have indicated?

 

[John Mulcahy]

REW knows the window type and it's properties and so correctly calculates the component energy from the bins that comprise it.

Coherent averaging cannot be used if the blocks being averaged are incoherent.