Hi, what is the ENOB benefit of driving the ADS131A04 via differential signal?
Using single ended signal I'm already getting 16.8 ENOB.
How much do you expect to gain with the differential signal?
With differential signal the SNR should be better, this improves the ENOB.
….but if the ADC is noisy it could be that i am already on the limits??
Rui,
Generally ENOB refers to a dynamic measure of the ADC which involves SINAD and THD. Instead, I'll just refer to the Effective Resolution as we in the datasheet, as it centers around the noise performance in comparison to the full-scale range. In the ADS131A04 we show this equation for the effective resolution:
In this equation, the 2xVREF refers to the full-scale range. For the ADS131A04, the full-scale range is ±VREF. If you're using a single ended measurement, then you're only using half of the full-scale range (only the positive measurements). This would affect your effective resolution calculation if you're only using part of a measurement range.
Right now, you're getting 16.8 bits of effective resolution with a single-ended measurement. If you used the 2xVREF in the equation, then this is the same calculation as the effective resolution in the datasheet. If you use VREF in the numerator, then you'd get 15.8 bits as the effective resolution.
Whatever you're measuring, it's just better to think about it as how much noise you have in the measurement. If you were to try to increase your signal and make it single ended, you might increase your effective resolution. However, by putting in some additional front end circuitry, you'd could add much more noise, and you might increase your offset error and drift.
Joseph Wu
Rui,
I think that from the noise point of view, the ADS131A04 can achieve 18 bits of effective resolution. Here's a table of the noise from the datasheet that shows the noise performance you should get:
However, as I mentioned before, the definition of ENOB includes AC measurements from the as part of the calculation. For the ADS131A04, the SINAD measurement in the datasheet (fIN = 50 Hz or 60 Hz up to 50 harmonics, VIN = –0.5 dBFS) is given as 101dB. You can calculate the ENOB as:
ENOB = (SINAD-1.76)/6.02 = 16.48 bits
In this case the distortion will limit the ENOB and you won't get 18 bits of performance regardless of single ended or differential measurement.
I would note that if you only used the SNR instead of the SINAD, the SNR (at fIN = 50 Hz or 60 Hz, VREF = 4.0 V, VIN = –20 dBFS, normalized) is 115dB. This would be:
ENOB = (SNR-1.76)/6.02 = 18.8 bits
Both are done with a data rate of 8kSPS.
Joseph Wu
