What is the effect of my reference's variation on the output of my AD5421?
Table3 gives specifications with the internal reference of the AD5421 being used. For example the TUE (total unadjusted error) with internal Rset and internal Reference is ±0.157%FSR over -40°C to 105°C temperature range. This would represent the total error of the current output.
The specifications in table1 of the AD5421 datasheet assume an "ideal" 2.5V reference. If you chose to use an external reference you would need to calculate the effect of this reference on the output. For ease of explanation I will assume the internal current setting resistor (Rset) is being used.
The TUE specification with internal Rset is ±0.126%FSR. If we use a 2.5V reference with an absolute error 0.04% and a temperature drift of 3ppm/°C, and the tempearature range is -40°C to 105°C, the worst case TUE will be given by:
TUE error of AD5421 + Absolute error of reference + Temperature drift (in%) * temperature range
= 0.126%FSR + 0.04% + (0.0003% * 145)
Thus with this external reference the TUE error will be 0.2095%FSR. Note that this means with the internal Rset the accuracy is better with the internal reference than an external 0.4%, 3ppm/°C reference.
This result will not always be the case, as the reference on the AD5421 is very accurate. For example the AD5420 would benefit from a precision external reference, as the internal reference is specified as 0.1%FSR and 10ppm/°C.