Optimal ADC input resistor for ADAU1701

Hi Fernando,

The Johnson current noise of the total 20K ohm input resistance (18K external plus 2K internal) at room temp & 20KHz BW, works out to 0.13nA RMS. This is 118dB below the 100uA RMS full-scale input current. Overall, the -1701's ADCs are noisier, at -100dB SNR. This suggests that we can vary the ADC input resistance together with its input voltage over a moderate range with little change in the ADC's SNR.

With the standard 18K ohm external resistor, full-scale input is 2VRMS. If your driving stage is capable of more swing than that, it would make sense to increase both its gain and the ADC input resistor, to make that stage and the ADC clip at about the same level. This should provide the best SNR between them. Since the ADAU1701's ADCs are current input, the usual advice to keep impedance low doesn't apply here.

Best regards,

Bob

Hi,

Perhaps an ADI engineer knowing the inner workings of the -170x's audio ADCs could provide a more complete and insightful answer.

Hello Nando.One,

The audio ADCs on the ADAU170x family are current-input instead of voltage-input. They were designed for a nominal external resistor of 18 kOhms, and this circuit was used to characterize the specifications of the device that you see in the table in the datasheet. Varying the input resistor will potentially change the performance of the ADCs compared to the numbers you see in the datasheet, and we haven't fully characterized the device over the full temperature range with all possible combinations of input resistor size. You are of course free to experiment with different configurations.

I personally suspect (along with Bob) that you might be able to get a slightly better signal-to-noise performance by increasing the gain of your buffer, as long as the buffer amplifier has a sufficiently low noise floor, and the input resistor is increased correspondingly.

Brett, thanks for your official input!

Bob

Hello Brett and Robert!

Thank you both for the input. Robert, as always, you're very helpful.

As for now I doubled the gain of my input buffer and doubled the gain of the input resistor, so now it is 39k. Honestly I didn't notice any changes, but it is doing no harm and that's for sure. As the product is ready to sell, I won't mess around with it anymore.

There it is, two SigmaDSPs, one PIC24F, and lots going on on the firmware and software:

Looking good! Congrats on your new product.