# Measuring single-ended signal by differential ADC

#### Ivan Govnov

Joined Nov 27, 2020
10
Hello, Buddies! In first, pardon for my poor English. Secondly, I need help with single-ended signal measurement by using ADS1120, because I have already broke my brain. Introduction:
2) 5-v current sensor with ratiometric output with 2.5 bias on output with small drift. So, as example, 2.5 v means 0 A, 0.5 v - -50A and 4.5 v - +50A;
3) Desired current frequency is 50...60 Hz, high-order harmonic is not very significant
Questions:
1) What would be better: simply bias AIN(-) by half of supply, and use 5-v supply as reference (further - Vsup). So, output of transducer in 2.5-v would mean code 0, 4.5-v (2/Vsup)*2^15-1 and 0.5-v - (-2/Vsup)*2^15. Also, I can use divider to get 2 V of supply in REFP0 pin to get full-range.
Or, another, I can use 2 opamps to convert single-ended signal to differential. First opamp should be configured as follower, second - as inverting with unity gain. Because all opamp would be in one package, it may be expected that drifts and offset would be rather similar and rejected by differential ADC. So, it is no need to use expensive low-offset opamps. Or no?
Example:
2) Is it true, that filter on REFPn pins must be similar to ADC input filter for real ratiometry? Also, by using divider on AIN(-) pin, is it significant to match resistance of filter on AIN(+) and AIN(-). In other, for example, output resistance of divider is 750 ohm, resistor of AIN(+) is 5.1k, so I need add to AIN(-) pin resistor near 4.3K, not 6.2K? In fact, output resistance of transducer unknown, but minimum recommended load is 4.7K.
3) I like to measure 2 transducers by single ADC: so, I put sampling rate to 2KSPS and simple switch signal by internal multiplexer and then start conversion (in single-shot mode). So I obtain 1KSPS per channel - this is near 20 samples per 50-Hz period. Is it look acceptable to calculate phase shift between 2 currents and frequency? (the load is 3-phase motor, but measuring only 2 phase may be enough to detect abnormal situation).
With respect!

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,088
You have 2 pins that can be muxed to select an external reference that draws 10 nA, So, pick 2 resitors to divide by 2.

Don't use Vcc/2; Your reference is Vcc/2. The instantaneous value of Vcc divided by 2.

it's imposible for say an automotive sensor that uses 5v to get to 5V or 0V. Use the transfer function of the sensor.

I didn't look at the datasheet, but I would suspect that you would get a 2's complement (+-number) that you would multiply by amps/bit to get the +-50A and it comes out in the wash.

Last edited:

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,088
I added some stuff in the post above.

Alegro didn't do a good job. You do get nominally 20 mV/Amp

On PDF page 22 note that the output cannot reach Vcc.

#### nsaspook

Joined Aug 27, 2009
11,758
I see a problem with your question 3. Using a single ADC will compromise the ability to accurately calculate the current phase difference from two sources. Most controllers designed for 3-phase motor applications have at least 4 independent ADC devices, each with a separate S/H channel to handle simultaneously sampling at least two phases for voltage and current. If your applications is only looking at fault conditions instead of power control then a single ADC method should work.