Oscilloscope Vrms, wrong!

[lemon]

Probably the issue with FR cutoff when I applied a compensation file, was due that I haven't restart the software Mi-Pro.

I changed the values of compensation file with positive values ​​instead of negative ones that I had.
But I seem that this applied of both channels and not only to Ch.A, at the "Spectrum Analyzer Processing.." the menu "Apply to" is inactivate.

[lemon]

Let's see what happens if I applied a compensation notch file.

I create a .fcf file with the values as I show at the attachment and then I loading it via [Spectrum Analyzer Processing]>"Compensation 1".

All the band under 2K has gone, I see at the manual that all frequencies that falls outside the defined range will be given a gain value of –1000 dB.

At the start the .fcf file had only 2..8 values of harmonics, after I created this file with some frequencies from 10-1000Hz.

Right click on spectrum analyzer with only the A channel of notch filter has no enabled the "Generate Frequency Compensation File (Flat)" to try with this way.

Probably the issue with FR cutoff when I applied a compensation file, was due that I haven't restart the software Mi-Pro.

I changed the values of compensation file with positive values ​​instead of negative ones that I had.
But I seem that this applied of both channels and not only to Ch.A, at the "Spectrum Analyzer Processing.." the menu "Apply to" is inactivate.

OK, I resolved them.
The inactive "Generate Frequency Compensation File (Flat)" wants to stop the capture and of course works but it isn't useful in my case.

Here a new capture and I have complete some other measurements THD+N%, THDN_dB, am I right?

THD+N% = [THD_A(%)]*POW(10,([f1RMS_A(EU)]-[f1RMS_B(EU)])/20)/[f1RMS_B(EU)]
THDN_dB = 20*log10(([THD_A(%)]*POW(10,([f1RMS_A(EU)]-[f1RMS_B(EU)])/20)/[f1RMS_B(EU)])/100)

Now, I have on the Notch Channel, with the Method-1
THD% = 0.000016 %
THDdB = -135.90 dB
THDN% = 0.000023 %
THDNdB = -132.83 dB

[VirtinsTech]

For THD (20–20 kHz, without compensation), Method 1 measured a higher value (0.0000099%) than Method 2 (0.0000079%). This is expected, as Method 1 includes a greater number of harmonic orders (2nd-20rd vs 2nd-9th)

For THD (up to the 8th harmonic, without compensation), Method 1 measured a slightly lower value (0.0000078%) than Method 2 (0.0000081%). This is also expected, as Method 1 includes harmonics only up to the 8th order, whereas Method 2 includes up to the 9th order.

The compensation file appears to be correct. We have recreated and tested it here without any issues. Please check it again, particularly for any unexpected or non-displayable characters in the empty spaces.

[VirtinsTech]

[Spectrum Analyzer Processing]>"Weighting">"Apply to" is for frequency weighting profile selection only. On the other hand, it is possible to specify the compensation factors for individual channels. If the compensation file is mono, it applies to both channels. If it is stereo, it applies to each channel individually. By the way, the compensation factors at the harmonic frequencies should be positive for your notch filter, according to the Excel sheet you provided earlier.

“Generate Frequency Compensation File (Flat)” should be used after a frequency response test to generate a compensation file that flattens the measured frequency response.

[VirtinsTech]

Method 1 for THD+N

(1) THD+N in dB
[THDNDB_A(dB)]+[f1RMS_A(EU)]-[f1RMS_B(EU)]

(2) THD+N in %
POW(10,([THDNDB_A(dB)]+[f1RMS_A(EU)]-[f1RMS_B(EU)])/20)*100

or

[THDN_A(%)]*POW(10,([f1RMS_A(EU)]-[f1RMS_B(EU)])/20)

[lemon]

Thank you again for the help.

The compensation file is right, now. I make it for two channel as seems at the attachment capture.

I don't know if the compensation file must include the values before harmonics (10Hz-900Hz),

I have done the compensation file from the Frequency Response with White Noise. For example I set one SE output of RTX6001 to 1Vrms and press the "FRwhite" from the "Panel Buttons". The FFT is the same with this of THD measurement.
The one cable goes into the Notch and the Notch Output to the left ADC, the other cable goes direct to the right ADC of RTX6001.
Then I compare the two responses on the Spectrum Analyzer and note the differences.
First, I noted only the harmonics h2..h9 but I saw that there is a step at the left part of Spectrum...then I decide to include and the left values from 1KHz, (10, 20, 50, 100, 200, 300Hz).

With this compensate file, the Ch.A with the notch is as the attachment file.

What is the right, finally? The compensation file must be have only the harmonics, or all the values?

[VirtinsTech]

Actually, in the above discussion, the compensation factor at the fundamental frequency is assumed to be 0 dB (i.e., no compensation). If the compensation file is constructed manually or generated from the measured frequency response, with absolute gain values, to flatten the true frequency response of the notch filter, then all parameters, including THD and THD+N, calculated from the compensated spectrum will directly reflect the true values of the original circuit. No UDDP calculation is required.

[VirtinsTech]

The absolute compensation file can be generated automatically using the transfer function measurement function in Multi-Instrument, with the input signal fed into Ch.B and the response fed into Ch.A. The generated compensation file has a *.gpc format (i.e. including absolute Gain and Phase). Similar to FCF files, the GPC file can be loaded as a frequency compensation file.

For THD measurement, spectrum data below the fundamental frequency are not used, but they are included in THD+N calculations. Also, note that peak energy is not derived from a single FFT bin, but from several FFT bins around the main lobe. Therefore, the frequency compensation should span at least slightly beyond the cutoff frequencies.

[lemon]

Nice Information, well expained but:

Actually, in the above discussion, the compensation factor at the fundamental frequency is assumed to be 0 dB...No UDDP calculation is required.

Is it meaning that UDPP THD calculation isn't required and the DDP THD_A is adequate? Look at my next answer about this...

The absolute compensation file can be generated automatically using the transfer function measurement function in Multi-Instrument, with the input signal fed into Ch.B and the response fed into Ch.A. The generated compensation file has a *.gpc format (i.e. including absolute Gain and Phase). Similar to FCF files, the GPC file can be loaded as a frequency compensation file....frequencies.

1) You mean the Bode Plot probably.
OK, here is the capture of this setup, for the purpose of this I use the RTX6001, the Generator of Ch.A goes to notch and the output of this to ADC Ch.A, the Gen of Ch.A goes direct to ADC Ch.B.
I save the "Spectrum Analyzer Generate Frequency Compensation File (flat)".
Then I run the THD and compensate with this file.
I have attach the capture of this.
Is that OK until now?
At the UDDP THD measurements I have, THD_20...20K: 0.0003381%, THD_h2..h9: 0.0003372% and with DDP THD_A I have_A: 0.12743%
For me the true for the Ch.A is THD_20...20K and THD_h2...h9 and not the THD_A, am I right?

[lemon]

2) At the last if I use as compensation file "Spectrum Analyzer Generate Frequency Compensation File (flat)" that from FRwhite then I have a totally different values of THD...the differences on compensation file is bigger!
Which method is right (1) or (2), probably the (2) is too very good to be true!