Originally Posted by
BennyHHR
I had questions as to the accuracy of my AFX when bought almost two years ago. Did a bunch of testing, including voltage issues, setting up average histos, without a confident solution. Even had exchanges with a NTK engineer on the issue.
First, realize that these and most others, compute lambda and convert that to a gasoline AFR value, even though you may be running an ethanol mix fuel. Stoic is usually considered to be 14.68, yet for some reason NTK considers it to be 14.57, so there's some of the difference.
After considerable testing with open loop tuning and seeing the resulting FTs when back in closed loop, looking at the histos for averages, etc., what I did was write a custom PID for the AFX. Their formula is: AFR = (V * 1.4) + 9. HPT's equivalent formula used in the supplied PIDs is: AFR = (V / 0.7143) + 9, which is the same. After doing considerable tests and some math, in the custom PID I used 9.33 instead of 9 for the adder in the formula.
From that point on, after open loop tuning was done switching back to closed loop always resulted on near 0 FTs, which tells me that the 0.33 "correction" resulted in accurate data. (The gage still reads a little low though.)
That seems to closely coincide with your testing, ktoonsez. You might want to try a modified custom PID for AFR and see what happens.