The short answer is: The method also attunate harmonics with similar level.
I have looked at a a ADAU DSP and a CS42448 codec card. It has several DACs. I have tried to use them differential and used BB INA2137 as a line receivers. The INA is similar to SSM2141 but -6dB gain
The balancing works great.
But it is so good that the measured distortion is mostly the distortion of the ADC of the soundcard.
Then I put another line receiver on two other DACS of the CS42448 and get another single ended signal with the same amplitude and phase.
If i invert this signal and put both into a third line receiver, the fundamental is cancelled about 40 to 60dB, but the noise is still there and the remaining harmonics is way lower than one channel straight into tha ADC. (eg, with or without muting of the cancelling signal)
This could be replicated with a line receiver, REW, stereo DAC and a cheap ADC.
But will the result be valid?
Is too much distortion cancelled out by the line receiver?
And is the load valid when the fundamental signal is cancelled out by an other source?
Napkin drawing
Example of ADAU graph in sigma studio
I have looked at a a ADAU DSP and a CS42448 codec card. It has several DACs. I have tried to use them differential and used BB INA2137 as a line receivers. The INA is similar to SSM2141 but -6dB gain
The balancing works great.
But it is so good that the measured distortion is mostly the distortion of the ADC of the soundcard.
Then I put another line receiver on two other DACS of the CS42448 and get another single ended signal with the same amplitude and phase.
If i invert this signal and put both into a third line receiver, the fundamental is cancelled about 40 to 60dB, but the noise is still there and the remaining harmonics is way lower than one channel straight into tha ADC. (eg, with or without muting of the cancelling signal)
This could be replicated with a line receiver, REW, stereo DAC and a cheap ADC.
But will the result be valid?
Is too much distortion cancelled out by the line receiver?
And is the load valid when the fundamental signal is cancelled out by an other source?
Napkin drawing
Example of ADAU graph in sigma studio
Last edited:
Here are some REW measurements with and without cancelling.
I have a 15k LP filter on one of the differential so the phase and amplitude is not quite flat on sweep.
(See I have old picture of 42448 card. C49 and C21 must be shorted for output)
I have a 15k LP filter on one of the differential so the phase and amplitude is not quite flat on sweep.
(See I have old picture of 42448 card. C49 and C21 must be shorted for output)
Last edited:
You might find an older thread of interest: https://www.diyaudio.com/community/...on-compensation-for-measurement-setup.328871/
Only read the first posts. That is reducing the harmonics of the dac?
This is reducing the fundemental to be able to measure the harmonics with a cheap ADC.
So its like mimic a 60 dB notch filter that can follow the measuring frequency.
This is reducing the fundemental to be able to measure the harmonics with a cheap ADC.
So its like mimic a 60 dB notch filter that can follow the measuring frequency.
The above measurement have all 4 outputs to -2 dBFS.
-6dB in each gain stage. 1 output cancelling the other .
The idea of the two stages is to do the normal balancing cancelling in the first stage.
My thinking is that it is not more cancelling of harmonics to be achieved with another balancing/cancelling?
-6dB in each gain stage. 1 output cancelling the other .
The idea of the two stages is to do the normal balancing cancelling in the first stage.
My thinking is that it is not more cancelling of harmonics to be achieved with another balancing/cancelling?
Last edited:
Maybe not suprising to many, but somewhat to me is the fact that the distortion of the INA is less than spec when cancelling fundamental on input.
It promise -100dB THD, but here it has -110 almost up to 1k when the two -2dB FS sines (about 1 volt) is cancelling each other.
For the record I dont know if it is INA noise/THD or DAC or ADC noise/THD, so INA migth be better
It promise -100dB THD, but here it has -110 almost up to 1k when the two -2dB FS sines (about 1 volt) is cancelling each other.
For the record I dont know if it is INA noise/THD or DAC or ADC noise/THD, so INA migth be better
Last edited:
And to be clear.
Can this measurement technique show better results than reality?
I know it can show worse because of the added noise of the INA, the noise of the cancelling signal and the noise of the ADC.
But many ADCs has very good noise but less than stellar THD over -20dBV. So can this be regarded the same as using a notchfilter?
Can this measurement technique show better results than reality?
I know it can show worse because of the added noise of the INA, the noise of the cancelling signal and the noise of the ADC.
But many ADCs has very good noise but less than stellar THD over -20dBV. So can this be regarded the same as using a notchfilter?
Read the first 10 pages now.
Really nice work. But in another league.
An instrumentation amp/line receiver or tre is a bit cheaper;-) And I love to use REW
Se analog devises has also come with some nice modules for measuring
https://www.analog.com/en/products/adaq7767-1.html
https://www.analog.com/en/products/adaq7769-1.html
Guess they could be software improved by pavels method
https://www.analog.com/en/products/adaq7767-1.html
https://www.analog.com/en/products/adaq7769-1.html
Guess they could be software improved by pavels method
I got the answer myself.:-(
If fundemental is cancelled then also odd harmonics are cancelled.
As long as phase is constant.
So method can only measure distortion without constant phase.
That is often distortion caused by components and not the DAC itself.
If fundemental is cancelled then also odd harmonics are cancelled.
As long as phase is constant.
So method can only measure distortion without constant phase.
That is often distortion caused by components and not the DAC itself.
So can only be used to pinpoint that distortion is caused by passive components with lower spac than the DAC chip.
Well. In the days of cheap boards with very good DACs that can be useful.
Well. In the days of cheap boards with very good DACs that can be useful.
Last edited:
And it gives a good indication if the dac board can be used with the harmonic cancelling method in link post #4
Last edited:
I am not sure I understand. The differential receiver just subtracts the two input signals. I.e. all (within the receiver's CMR specs) common mode parts are eliminated, only the differential parts gets converted to the single-ended output. How does it help with measuring distortions of some input sine signal?
The trick is to invert one of the input signals. Now the fundemental and the odd harmonics are cancelled (instead of the even harmonics)
Then the ADC is almost not loaded as the fundamental is reduced according to the CMRR of the receiver
Then the ADC is almost not loaded as the fundamental is reduced according to the CMRR of the receiver
But if inferior passive components are used in the output filter of the DAC the phase of the odd harmonics are not in phase for the two chanels and they are not cancelled
At least it looks like that if I measure a good DAC with crap caps on output. I can not cancel out the harmonics with difference amp.