I need to measure the voltage between two electrodes EL1 & EL2 accurately and send it to a PC. Voltage between EL1 & EL2 can be +/-2V max. which I thought of using a Sigma Delta ADC. But EL1 and EL2 can be at a max. of +48V from Ground. I would be using Sigma Delta ADCs for the first time, and I got impressed with the high no. of bits in these ADCs at low prices! The accuracy required is better than 1mV. I am considering the following options.

1. Use 2 potential dividers and bring down the voltage by 10, to less than 5V, and measure them using 2 Sigma Delta ADCs and measure the difference. My concern is that these dividers would introduce additional inaccuracies, may be rendering it bad!

2. Use an isolated supply for the ADC and connect it to the uC using SPI and optocouplers. Speed requirement is a lowly 1 to 2 samples / sec.

Please comment on my choice and recommend any other better alternative. I might have to measure the voltage between an additional pair of electrodes, say EL3 & EL4 in the same setup, as a future requirement.

 

There are high common mode voltage difference amplifiers like Analog Devices AD8479 that you may want to look into.

 

Option three, use a difference amplifier such as INA146. They are calibrated such that the fluctuations of the common mode voltage dont inifluence the measured difference.

 

Both pretty interesting parts. I suppose, you mean bring the difference voltage down near ground, and then use the ADC, right?

I saw the INA146 having 70dB CMR, and the AD8479 having 90dB. At 48V common mode, this could be 15mV & 1.5mV. Is there some way of cancelling or nulling this one out, as I need 1mV accuracy. Would they remain constant for the part? Please correct me if I am wrong.

Ok, that flags me to check whether the 48V will be more or less constant, or was that a max. spec.

 

The 48V common mode might introduce an offset of 15mV which you could calibrate out, but the accuracy towards changes in the measured voltage would not be hindered by this. How much does the 48V fluctuate, is it noisy?

 

Both pretty interesting parts. I suppose, you mean bring the difference voltage down near ground, and then use the ADC, right?

I saw the INA146 having 70dB CMR, and the AD8479 having 90dB. At 48V common mode, this could be 15mV & 1.5mV. Is there some way of cancelling or nulling this one out, as I need 1mV accuracy. Would they remain constant for the part? Please correct me if I am wrong.

Ok, that flags me to check whether the 48V will be more or less constant, or was that a max. spec.

Perhaps a logging DMM with USB interface - but make sure its still floating when hooked up.

Most types have timed shutdown when idle, that can be a bit inconvenient.

 

I need to measure the voltage between two electrodes EL1 & EL2 accurately and send it to a PC. Voltage between EL1 & EL2 can be +/-2V max. which I thought of using a Sigma Delta ADC. But EL1 and EL2 can be at a max. of +48V from Ground.

http://www.linear.com/product/LT1990

 

The 48V common mode might introduce an offset of 15mV which you could calibrate out, but the accuracy towards changes in the measured voltage would not be hindered by this. How much does the 48V fluctuate, is it noisy?

I am trying to confirm the high voltage info as to whether it is variable or fixed, as it was mentioned as 48V max.

Perhaps a logging DMM with USB interface - but make sure its still floating when hooked up.
Most types have timed shutdown when idle, that can be a bit inconvenient.

Oh nice, no need of much(?) work, simple and neat! Might need another electrode in future, maybe another DMM!

 

http://www.linear.com/product/LT1990

Oh, good, lots of options. Will go through this.

 

The 48V common mode might introduce an offset of 15mV which you could calibrate out, but the accuracy towards changes in the measured voltage would not be hindered by this. How much does the 48V fluctuate, is it noisy?

Am I right in assuming that this offset would remain constant so long as the common mode voltage remains constant? I am intending to use only Gain 1, so it should not matter much if there is small variations of this 48V, as the 90dB CMR can take care of them, right?

 

Am I right in assuming that this offset would remain constant so long as the common mode voltage remains constant?

Yes.

I am intending to use only Gain 1, so it should not matter much if there is small variations of this 48V, as the 90dB CMR can take care of them, right?

Right.