Problem with large variation in power detector design for LNA control

Dear All:
I am designing power detector for LNA control ,I have some question qbout the RSSI circuit. I simulate the power detector , I found the process varaiation and temp varaiation effect , The power detector will larger +-3dB variation. Does anyone know how to solve the problems. Thanks

use two DC biased diodes, and take the voltage difference in an op amp. One diode has the bias and RF, the other diode only has the bias.

you can also just use one of the diodes as the feedback for the opamp, assuming the diodes are closely matched, this will subtract (or add depending on how you look at it) the diode offset.