Is this worth trying? -Idea to improve SNR of ADC and DAC
Christof Eb.
Posts: 1,247
Hi,
I would be happy to receive some feedback about the following idea:
I would like to reduce the noise of my propeller based guitar sound system, which uses the onboard sigma delta adc.
What do you think of the following idea:
The analog-signal shall
a) be led directly into a sigma delta adc and
b) be inverted (its polarity) with an opamp and led to a second sigma delta adc working in parallel.
The difference of the two digital signals will form the digital input of the sound processor.
The same procedure could be done with the output signal.
What is your feeling: Does this fight the main source of the noise?
Well, I am aware, that you don't know my specific setup, but perhaps someone could give me some hints...
Christof
I would be happy to receive some feedback about the following idea:
I would like to reduce the noise of my propeller based guitar sound system, which uses the onboard sigma delta adc.
What do you think of the following idea:
The analog-signal shall
a) be led directly into a sigma delta adc and
b) be inverted (its polarity) with an opamp and led to a second sigma delta adc working in parallel.
The difference of the two digital signals will form the digital input of the sound processor.
The same procedure could be done with the output signal.
What is your feeling: Does this fight the main source of the noise?
Well, I am aware, that you don't know my specific setup, but perhaps someone could give me some hints...
Christof
Comments
I think the same effect could be obtained by going directly into multiple sigma-delta ADCs and averaging the results. The counter outputs will be fairly random w.r.t. with each other, since the input thresholds will always differ by a little bit. By averaging, the deterministic stuff will be enhanced and any independent random variations will be suppressed. However, if the random variations from multiple counters suffer dependencies, the result could actually sound worse.
-Phil
Phil,
I remember someone has already tried to average multible adcs, he reported that this was not so good....
Leon,
yes some codec chip would be good, but I have not yet found one, that is easily available and easy to solder. And I was still thinking, if there could be a significant improvement with a simple clever modification...
In former experiments I had found, that there is a certain DC- band around the mid between the supply voltage and GND, where the result of the adc is completely reigned by noise. So another idea is to add dither to the input voltage to keep it out of that region.
Christof
- Convert the signal as close to the source as possible.
- Use a balanced signal (+signal and - signal) if possible.
- Use shielded twisted pair lines for the analog signal.
- Use a low noise ADC with balanced inputs.
- Oh yes, almost forgot...Only ground the shield at one end - usually the ADC end.
Won't work. I tried it about a year ago and found the noise shows no direct correlation or synchronisation. It just ended up making the noise level higher.
so I will not waste these hours....
Christof