Thought the guys who do this might find it interesting.
Courtesy of EFI University.

am not finding the slopes on the rich side to correlate well to the slopes on the lean side.

As a result, sensors calibrated using free air, then run on a tuned engine on a dyno, will show 12.0 when another sensor will show 12.7 ECT.
I have taken 3 different LSU4 sensors on the same controller and have seen as much as 1 AFR error between them (each having undergone a free air calibration prior to test).

I am wondering if the trim resistors are a superior calibration method than free air calibration all things considered


Hi,

The trim resistor is functionally EXACTLY the same as a free air calibration. Basically its a scaling factor not for the sensor curve, but for the inverse of the sensor curve (Ip over 1/Lambda). The only difference is the trim resistor is determined at 8.29% O2 in N2 vs. 20.9% O2 in N2.

But you are right, for an analog feedback wideband that measures Ip and trim resistor, the free air calibration is not as good as the trim resistor. The reason is the 1/Lambda over Ip curve becomes non-linear at higher pump currents. For a digital method, where the result is always linear with pump O2 flow, this non-linarity disappears and the resulting slopes ARE following a fixed relationship.

Just as a tip:
Plot the Lambda/Ip curve of the sensor not as Lambda/Ip, but recalculate the curve as K over Ip, where K = (1/Lambda) - 1. You'll see what I mean.