Dear Kai,

Thanks for your reply,
Your post has cleared most of my doubts and problems. But still some problems remain. Take an example of the classic precision rectifier circuit that you had posted in reply to my earlier thread. It was using 4 numbers of 22k resistors and one 11k resistor. Any variation in the resistor value due to the tolerance or drift affects the signal and these resistors are so critical that if one of them increases or decreses one of the lobes of the output rectified signal is bigger or smaller than the other. The 22K resistors affect by the proportionate margin for example change in 0.1% affects one of the lobe by 0.1% where as the 11k affects by double margin change by 0.1% affects the signal lobes by 0.2%. Considering the worst case where all the 5 resistors have maximum 0.1% error the effective error in the signal will be approx. 0.6%. And I beleive even the method of using cermets to compensate this is difficult as it will need a cermet with all the resistors and for adjusting these cermets what should be the reference. My DMM doesnt have enough precision to measure the 0.1% tolerances precisely. And since the resistor variations affect only one lobe a compensation at the final amplification also wont help at all. Kindly post your suggestions for the remedies to such problems.

Thanks & Regards,
Prahlad Purohit