REW Beta Release RTA level calibration question on REW 5.20 beta 47

[Antonio Di Motta]

The RTA function is really very important and also very well done and also very fast even on the Raspberry. The ability to calibrate the level is very important, this function allows you to have a real and very precise measuring instrument. They perform several tests in the laboratory, I noticed that RTA also uses the level calibration carried out in the preference / soundcard, this is certainly functional when using the sound card to generate the signal used to make the measurements, but if we want to measure a generated signal externally we must exclude the calibration made with the loop since it corrects both the non-linearities of the DAC section and those of the ADC section and we should only implement the calibration for the ADC audio section using an external generator. This type of calibration would avoid level measurement errors when we use RTA to measure externally generated signals.
Thanks for the attention
Regards
Antonio

 

[John Mulcahy]

True, I will change it so the soundcard cal file is only applied when the signal generator is playing.

 

[Antonio Di Motta]

Do you think it would be useful to provide an advanced level of calibration of the RTA function that can be carried out using an external sweep signal generator to make a fine setting?
In this way, having both calibrations, in loop and only in the ADC section, both could be used depending on whether the internal signal generator is paused or in play.
In addition, having both calibrations available, for difference you would also get the one of the DAC section only and I think it could be useful for future updates.
If you think you want to implement this feature, I am always available for testing.
Thanks for the attention.
Regards
Antonio

 

[John Mulcahy]

If you have an external source that is perfectly flat you can simply make a measurement through that source, offset the result so it is 0 dB at 1 kHz and export it as text to make a cal file. The challenge is generally to come up with a reference whose response is either perfectly flat or perfectly known so that it can be separated from the measured response.

 

[John Mulcahy]

Thinking about this some more, the soundcard cal needs to remain active since its result is a combination of input and output effects that can't be separated. If you can separate the influence of input and output you could split them between soundcard and mic cal files and load only the relevant files when measuring.

 

[Antonio Di Motta]

when making measurements that require the use of both the DAC and ADC sections of the sound card, it is certainly correct to consider the effects of both sections, as regards the frequency response that uses an analog microphone connected to the sound card, it must be taken into account of the DAC + ADC + microphone calibration curve and if I want to be more precise the response of the microphone preamplifier should also be considered.
My use was exclusively of the ADC section for the RTA function, in this case the compensation of the DAC section I think leads to an incorrect correction.
Different situation when using USB microphones in this case a correct calibration should take into account only the DAC section + the calibration of the USB microphone if available.
I understand that I have made a correct reasoning.

 

[John Mulcahy]

The challenge is how to find out the response of the ADC section alone, independently of whatever is driving it?

 

[Antonio Di Motta]

Today I have done a test with my function generator which results between 10Hz and 100KHz. I deleted the calibration file of the sound card and had a frequency sweep from 10Hz to 48Khz generated at a 1Veff level by performing the measurement with RTA and I was able to obtain the response curve of the ADC section. I saved the file with only the level values without the phase values because they are not available, and I performed the correction operation with a spreadsheet and generated the calibration file, I did a quick test but it doesn't work.

 

[Antonio Di Motta]

Here are the files that I used to try to make the correction.
The first file is the ADC response measured with RTA, the second the response text file and the third the correction to be used for calibration.
Using this file the signal level in RTA is very wrong I have to understand if I did something wrong. Compared to the calibration file generated by REW this is much smaller because there are less data, certainly for the setting of the FFT at 32K instead of 64K, but I don't know if this can be the cause of the malfunction.

 

[John Mulcahy]

You need to offset the measurement so it is 0 dB at the desired reference frequency, usually 1 kHz.

 

[Antonio Di Motta]

I was wrong to correct the file, because I had inverted the levels that besides setting to 0db at 1khz I had multiplied the values by -1 so that a -3db attenuation became in the + 3db file this is wrong. However there is still something that does not work perfectly and I have to check, perhaps the response curve made with the external generator is not precise because at the high frequencies some level ripple remains.
I'll do more tests and let you know the results.
Thanks John