Other Parts Discussed in Thread: OPA1632, INA333, INA826
Hello,
I Will use an OPA1632 on as input amplifier for an analog to digital converter board. OPA1632 offset temperature drift is +/-5µV/°C. Is this drift stable on a chip ? Target is to calibrate my board over temperature range 0-55°C. Can this offset drift be corrected by a calibration ?
Thanks
Michel
Michel,
OPA1632 offset drift typical spec of +/-5uV/C represents +/-1-sigma of Normal Gaussian Distribution as shown below. This means that 1 out of 3 units will have a drift greater than 5uV/C (1-sigma), 1 out of 370 units will have a drift greater than 15uV/C (3-sigma), 1 out of 1,744,277 units will have a drift greater than 25uV/C (5-sigma), etc. at least for the temperature range from -40C to 85C (for worst case analysis, I would assume the same drift for 0C to 50C.)
With time, all PDS specs undergo a slow long-term shift and this include an input offset drift. The blue bell-shape curve on the above graph (to the right) represents an initial distribution of drift while the green curve shows the same units after 10 years in the field - the maximum drift may shift by the maximum initial value. Since in case of OPA1632, there is no PDS maximum drift spec, you need to use statistical means to set a relative maximum value; for example, if you will accept that 1 out 22 units may be outside of such maximum limit, +/-2-sigma or +/-10uV/C maximum shift should be used. This means that a unit with an initial drift of +10uV/C may become +20uV/C or 0uV/C (+10uV/C+/-10uV/C) while another unit with initial drift of -5uV/C may become +5uV/C or -15uV/C (-5uV/C+/-10uV/C).
The initial offset drift can be calibrated out by measuring a change in the input offset voltage of each unit over desired temperature range BUT there is no way to calibrate the long-term drift since it is a random shift that is not a function of the initial drift - the only way would be to periodically calibrate the individual units as they age.
