Question
The AD70 being obsolete, we plan to use the AD7709 and we have two questions :
1) One pin of our RTD is connected to gnd, so is it possible to connect one
resistor in serie with RL1 and one in serie with RL2 to achieve a min common
mode voltage of 100 mV (see datasheeet AD7709 page 27) instead of using Rcm
which is impossible in our case.
2) How can we compute the accuracy of the measurement.
Answer
1) we have not tried this previously but, it should work.
However, the resistor in series with RL1 should match the resistor in series
with RL2 so that no error is introduced due to resistor mismatch.
You need to use low tolerance resistors with low temp coefficients.
2) The AD7709 has an excitation current of 200 uA.
When the RTD is at 300 ohms, the voltage seen by the AD7709 will be 60 mV.
When the RTD varies from 300 ohms to 700 ohms, the voltage will increase by 80
mV.
With a 6.25 K resistor to generate the reference voltage, the reference will
equal 1.25V.
When the gain is set to 8, the analog input range is +/- 156.25 mV.
So, the AD7709 has enough span to tolerate the 60 mV of offset.
However, you could perform a zero-scale calibration to remove this.
This would mean that the RTD needs to be set to a known temperature and a
system offset calibration can be performed.
When the gain is equal to 8, the AD7709 datasheet states that the p-p
resolution is 16 bits when the reference is 2.5V.
With a 1.25V reference, the accuracy will be 15 bits.
This is for a range of +/-156.25mV.
Since you are using a quarter of this range you should expect an accuracy of 12
bits - which meets your requirement of 0.1%.