Hi, I am researching how to do multitone nonlinearity test for amplifiers and there seems to be not too much information on how to quantify the measurement. I understand simple single tone (THD) and two tone tests (IP3/TOI) are common and are relatively easy to quantify results, but multitone seems a lot more difficult since there are intermodulation and harmonic products all over the spectrum (and beyond). However, a multitone signal can be implemented such that it is much more representative of the signals the amplifier has to deal with (such as NRZ or PAM4 data). This is why I want to implement it. Does anyone have any idea on how I would quantify this multitone test measurement. It would be nice to have a similar metric to THD or IP3 where it is relative to the powers of the input tones. Thank you in advance!

 

Does anyone have any idea on how I would quantify this multitone test measurement.

You could look at all the IM distortion products with a spectrum analyzer and calculate them as a percentage of the fundamental amplitudes, similar to how two-tone tests are characterized.

 

You could look at all the IM distortion products with a spectrum analyzer and calculate them as a percentage of the fundamental amplitudes, similar to how two-tone tests are characterized.

That's a good idea. Any suggestions on the best way to space the input tones across a certain band of interest. It wouldn't be ideal to have the tones be linear combinations of each other, otherwise you wouldn't know if the output power at an intermod frequency is from. Also, the more tones you have to get a more accurate representation of the signal, the more complicated the ratio becomes. I guess a script can parse through all the linear combinations of the tones within the band of interest?

 

Could sweep them independently , them being as many as you want measure interaction of. Depends on what creation/capture tools you have. Sky's the limit and yes spectrum analyzer .

 

Could sweep them independently , them being as many as you want measure interaction of. Depends on what creation/capture tools you have. Sky's the limit and yes spectrum analyzer .

Interesting suggestion, but wouldn't that not be representative of the realistic signal since there would be no intermodulation?

 

Interesting suggestion, but wouldn't that not be representative of the realistic signal since there would be no intermodulation?

Intermodulations are caused by non-linearity and phase shift in the amplifier. Using swept tones, these can be measured with a spectrum analyzer.

 

Intermodulations are caused by non-linearity and phase shift in the amplifier. Using swept tones, these can be measured with a spectrum analyzer.

That can measure harmonic distortion, but not phase shift, and not IM distortion which requires more than one frequency.

 

That can measure harmonic distortion, but not phase shift, and not IM distortion which requires more than one frequency.

An FFT gives information about both amplitude and phase. Changes in both over the frequency spectrum of the amplifier can be displayed on a spectrum analyzer and used to determine the quality of the amplifier. There is no need to measure intermodulation distortion separately as it is a product of the above.

 

In the line of unusual and specialized test equipment that performs as advertised, I suggest contacting Rhode and Schwartz, because they quite likely already offer a package to perform the test described, or a test very similar.
That company is less well known, but offers an impressive line of test equipment.

 

Thanks for the information, any suggestions to develop a system with the best characterization of the realistic signal and being able to use a metric to characterize the nonlinearity while making sure the intermod products minimally show up at the input tone frequencies. I have PNAs, AWGs, Spec Analyzers, etc.