M.I. + ASUS Xonar Ess. STX + Win10

[fotis]

Hello
I'm an old customer and forum user, but I've lost any account & login info and so I opened a new one. Sorry for that.
My current version of MI is: Multi-Instrument Pro 3.9.5.6 + DLG + LCR + UDP.
Is running on a PC under Win10 x64 Pro.
My sound card is an Asus Xonar Essence STX.
I use a hard key (dongle) to activate each time the MI.
Well, I did enough upgrades to my PC from the time I first used MI. If I remember well, the OS was then the first 64bit Windows version, the Asus was installed on an Intel MB and everything was perfect regarding all measurements.
When I changed the MB and the OS to Win7 x64 Pro I lost the good performance, currently on spectrum analyzer bandwidth measurement. Please see the previous bandwidth on the lower pic and the new on the upper pic:
Both measurements are in Loopback mode, with 192K-24bit white noise. I tried any combination with the signal generator and the spectrum analyzer processing without any improvement. The bandwidth always falls at 3Khz on both channels.
I have inspected the sound card for possible damage but I haven't located any issue.
Have you any idea to help me?

Thanks

Fotis

[VirtinsTech]

Thank you for your questions. The two were measured with different method despite the same white noise excitation was used. "Asus band.png" used inter-frame linear average and no frequency weighting while "adc1.png" used inter-frame peak hold and A-Weighting. The setting of the former one is correct. However, it still cannot not explain why the frequency response dropped from about 2kHz onwards. Probably the signal captured in the oscilloscope was wrong?

By the way, with the new version of the software, it may be easier to use periodic white noise as the excitation signal (Signal Generator > MultiTones > Automatic > White Noise and Align to FFT). It excites each FFT bin within the specified frequency range equally and does not requires any average.

[fotis]

As I decided to find out if the problem is caused by the Asus Xonar or the M.I., I removed the sound card from the PC and did a careful visual inspection of the board without finding any damaged parts.
Please note that I haven't used M.I. last 10 years for the reason I explained in my first post.
I repeat that all tests of Asus with M.I. gave excellent results until the upgrade to Win7 (I think during 2013).
Next, I reinstalled the card on the PC and re-checked the bandwidth with the M.I. FFT, but the resulting curve was the same bad as above.
After that, I decided to test separately the Asus Line IN and Line OUT.
I connected an external function generator to Line IN and did an FFT analysis with the M.I.
With the external gen. in sweep mode from 20 to 20,000 Hz and the M.I. FFT analyzer in record mode, the bandwidth shows much better.
The -3dB point moved from 3KHz to 8.5kHz.
Is not so good, but it clearly shows that there is no hardware problem within the Asus Line IN circuit.
For confirmation, I did a slow manual sweep in the same range, but this time by observing the M.I. oscilloscope.
The RMS value of the sinusoidal signal looks very stable from 20 to 20,000 Hz, with a deviation of just +/- 1 mV.
As a result, I consider that Asus Line IN works just fine.
Now I'm trying to figure out a way to inspect and the Line OUT of the card.
Unfortunately, an external FFT analyzer is not available.
In simple listening tests with speakers and a HiFi amplifier, I can't hear any problem and could confirm you that a -3dB drop from 3 kHz and above is very audible!
I think that I will make a test as above, but with the M.I. signal gen this time in Loopback mode, and I will observe if the FFT measured response is consistent with the oscilloscope waveform RMS value.
I will be back with news and some screenshots.

Best

Fotis

P.S. Just now I noticed your reply while I was typing. Thank you very much.

[fotis]

Thank you for your questions. The two were measured with different method despite the same white noise excitation was used. "Asus band.png" used inter-frame linear average and no frequency weighting while "adc1.png" used inter-frame peak hold and A-Weighting. The setting of the former one is correct. However, it still cannot not explain why the frequency response dropped from about 2kHz onwards. Probably the signal captured in the oscilloscope was wrong?

By the way, with the new version of the software, it may be easier to use periodic white noise as the excitation signal (Signal Generator > MultiTones > Automatic > White Noise and Align to FFT). It excites each FFT bin within the specified frequency range equally and does not requires any average.

Thank you for the tip on using the white noise new feature. My mistake, both FFT have been taken with Pink Noise! Anyway, the individual settings are not the problem, I will find them again. The -3dB drop from 3kHz is a very coarse problem and apparently is not related to settings. So I have to examine if there is no sound card hardware issue and in this direction, I still have to confirm with a measurement that the Line OUT is also healthy. If it is, then the problem arises from the side of software. Asus has released enough upgrades and I have installed the latest driver for Win10. I also have installed the latest version of M.I. I don't know if there is any conflict. I will make the test I described above and will post it with pics.
Sorry for the consumption of your precious time!

Thanks

Best

Fotis

[fotis]

OK, the problem of the drop from 3khz and above applies only to the Pink Noise stimulus.
All other, White noise, frequency sweep, etc. provide very good and realistic FFT bandwidth, yet without Intra and inter-frame processing and without window function (using rectangle).
It is remarkable the stability of M.I. yet at 192khz sampling frequency with ASIO.
All other software I've tested these days, present "hung-ups" in such conditions.

Thanks a lot for the new softkey, it makes M.I. run faster.

I'm happy because I got back my FFT analyzer

Best regards

Fotis

[VirtinsTech]

The spectrum of pink noise is NOT FLAT, it follows 1/f law. If pink noise is used as the stimulus for single-channel magnitude frequency response measurement, then the spectrum analyzer must be put under octave analysis mode rather than narrow-band FFT mode. In MI, the octave analysis mode can be set by right clicking the spectrum analyzer window and selecting [Spectrum Analyzer X Scale]>"1/1" octave, "1/3 octave", ..., or "1/96 octave".

[fotis]

OK, thanks.
I use pink noise as a stimulus when I make electroacoustic measurements. From what I know, it is the standard signal for measuring the frequency response of speakers, etc.
In all other cases, I prefer white noise, sweep, etc.
Thanks for the advice.

Best