Use of generator pre-distortion in REW

[jan.didden]

The REW generator can be pre-distorted with settings in the Distortion Controls window.
I want to use this to try to null out a specific harmonic frequency in a measured spectrum.
I import the actual measured harmonics from the RTA in the Distortion Contols window with 'Set from RTA'.
That works neat - the Distortion Controls window shows the magnitudes and phase of the harmonics I see in the RTA display.
For instance, for the 2nd it shows -46.8dB and 62deg.
Now I want to set this to pre-distort the generator such that it cancels the 2nd harmonic measured in the RTA.
How do I do this? Leave the magnitude and invert (add 180 deg) the phase? I tried but it doesn't seem to work.
Any pointers?

thanks,
Jan

 

[John Mulcahy]

Just click the Set from RTA button, nothing else required.

 

[jan.didden]

John, OK, I do that, then cycle the gen on and off to load the new settings, then run the RTA again.
But the difference at the 2nd harm freq is minor, a few 10th of a dB, the phase flips from -62 to +125.
I'm missing something.

Jan

 

[John Mulcahy]

Changes apply immediately, there's no need to turn anything off. Follow the help:

The harmonics dialog can be used to produce a pre-distorted signal to counter the distortion of a measurement system at a particular frequency and level. To use it in that way make an RTA measurement of the system measuring itself, running the generator at the desired frequency and level. Coherent averaging is recommended to minimise the effects of noise on the recovered harmonics. Once stable harmonic levels and phases are seen on the RTA press the Set from RTA button. Note that all the generator distortion harmonics should be off before making the RTA measurement, the Clear all button can be used to ensure that. THD improvements of around 20 dB may be achieved, but the settings are only valid at the frequency and signal level at which they were measured and will not remain valid if the measurement system distortion varies (as its temperature changes, for example).

 

[John Mulcahy]

Here's an example:

 

[jan.didden]

That is quite impressive!
I am trying to wrap my head around the phrase 'make an RTA measurement of the system measuring itself'- which I guess is a loop-back.
In my case I want to measure a clean acoustic source with a microphone, null out the microphone distortion using this technique, and then measure the distortion of a loudspeker.
It probably (apparently) doesn't work that way.

Jan

 

[John Mulcahy]

A typical use would be electronic measurements of distortion at a frequency, first measuring with a loopback then inserting the device under test. It isn't applicable to the scenario you described unless you were to go through the process at each individual frequency at which you wanted a result, which would be very time consuming. Using the "time is money" principle it would be better to buy a mic with lower distortion.

 

[jan.didden]

There are no 'mic with lower distortion'. Nobody specifies their mics at say 90dB SPL. All specs read like '<3% distortion at 148dB SPL'.
I've been measuring 6 mics so far, from vintage 1/2 inch B&K to 1/4 inch modern high power electret touted as 'very low distortion'. They vary very little at 90dB, between -70 and -85 dB.
That is why I am on this project, I have ESL panel measurements that just show the mic distortion!

That said, I would be happy even if I could use this at just a single frequercy!

Jan

 

[John Mulcahy]

The limitation would be the transparency of the "clean acoustic source", since its distortion would end up being applied in opposite phase when you substitute the speaker you want to measure.

 

[John Mulcahy]

It might be more practical to find an environment where the mic could be much further from the speaker.

 

[jan.didden]

In fact, I am simulating a clean acoustical source by using two small speakers, each emitting a different tone. (I have two instances of REW running).
Since the two tones do not intermodulate in the air, any intermodulation distortion is from the mike alone.
I note the level and phase of one IM product, say f1-f2, which is a 2nd order product..
Then I do a traditional harmonic distortion measurement with the 2nd harmonic falling where the IM product was, and with the same exitation level, thus exciting the same 2nd order nonlinearity..
With this measurement I get a 2nd harmonic that is composed of the 2nd harmonic products caused by the speaker plus the mike.
By subtraccting the mike 2nd harmonic measured with the IM measurement from the HD rersult, I am left with the speaker HD only.
At least, that's the plan ...

Jan