virtins.com

To obtain the absolute attenuation rates at different frequencies, the Bode plot function should be used. However, substantial differences can be observed between your measured gains and the simulation results from LTspice, especially around 1 kHz, where the measured value is about 36 dB, while the simulation result is around 60 dB. This discrepancy is likely due to the frequency resolution (46.875 Hz) being too coarse to accurately capture the attenuation at exactly 1 kHz, as well as the white noise excitation method for Bode plot measurement not being sufficiently precise here.

By the way, if a *.gpc file is loaded directly for frequency compensation (instead of gain and phase compensation), Ch.B after compensation becomes meaningless because the phase compensation will be incorrectly applied as gain compensation to Ch.B. That said, in this method, only the parameters measured in Ch.A are intended to directly reflect those of the original circuit.

With ARTA I have a very clean graph with Frequency Response and FFT-131072. A clean -60db falling.
I have done overlay the ADC Ch.2 on the ADC Ch.1 and I took this capture, very clean and I could export the one channel (csv, txt)

I try a lot to take an approximately graph like ARTA, at the Mi-Pro with the same parameters.
Frequency Response with pink noise was very bad and wrong (something like waterfall).
Frequency Response with white noise was good (almost -45dB fall) but the noise filter graph don;t touch the reference channel 2, lke ARTA at the same frequency...something wrong and with it.
Finally, I tried with BodePlot again, with Kaiser window and FFT-131072.
This is the best that I achieved, a -50dB fall. I take a compensation file from this and I done two SE THD measurements one with RTX6001 (loopback with notch) and the other with EOS1KV Oscillator (with notch).
I think than I am good now, and my effort must to concentrate how to remove the harmonics of 50Hz (hum).

..continuous from the previous message

It appears that your magnitude frequency response measurement used periodic pink noise (PN Pink) as the stimulus together with wide-band octave analysis in ARTA. This differs from using true white noise excitation with narrow-band FFT analysis in Multi-Instrument. Similar to frequency sweeping methods, periodic white or periodic pink noise methods do not require averaging and therefore converge very quickly. In contrast, true random noise methods require long-time averaging to obtain a smooth curve, especially when the frequency resolution is high. To obtain a smooth and accurate frequency response curve in a short time, you can use the “FRswp” button preconfigured for RTX6001. It uses a logarithmic frequency sweep combined with 1/96 octave analysis.

It is also possible to use a periodic noise as the stimulus. In Multi-Instrument, a periodic white noise or periodic pink noise can be generated using the Multitone mode in the Signal Generator. Please refer to Section 4.3.1.8.2.1 of the software manual for details.

1. For single-channel magnitude frequency response measurements:
Wide-band octave analysis should be used when the stimulus is logarithmic sine sweep, pink noise, or periodic pink noise.
Narrow-band FFT analysis should be used when the stimulus is linear sine sweep, white noise, or periodic white noise.

2. For dual-channel Bode plot measurements, only narrow-band FFT analysis is supported in Multi-Instrument. Direct THD/THD+N measurement in Ch.A with the compensation GPC file loaded and without using UDDP calculation is possible only if the Bode plot is measured accurately and with high precision.