I've attached a noisy data plot that I collected today. I just wanted to share the data here and see if anyone else has observed something similar. The curve looks somewhat exponential in shape.

The light load conditions (higher AFM voltage) seem correspond to duty cycles under 20 % while the higher load gets the duty cycle up near 80%. You can see that my car drives the AFM down to the 1/4 of a Volt range. (I have stock ports with pineapple sleeves and mild manifold port matching (and stock ECU).)

I'm still tweaking my data collection hardware, so I think that is why the duty cycle scatter plot looks so noisy. And yes, I did a search to see what if any information is on the forum about this topic and found very little. I'll post some more data when I have time...

 

The AFM output voltage varies exponentially with airflow, hence the shape of the trend. This is because at low load the AFM flap is nearly perpendicular to airflow but at high load it's nearly in line with it. So a small change in airflow at low load moves the flap a lot more than the same small change in airflow at high load.

 

Very interesting.

You could drop one variable by ploting AFM vane recession vs AFM voltage.

 

This data collection looks a bit cleaner. My interface to the injectors is a bit noisy right now. The data is from a few seconds of a rev up to around 7k RPMs. I included a text version of the data too. The columns are Time, RPM (k), AFM (V), TPS (V), AFR, Duty Primary (%), and Duty Secondary (%).