According to EVK3 HD bias document, physical threshold of positive and negative events are controlled by bias_diff_on and bias_diff_off, and bias_diff is recomanded not to change. I did some experiment on the 3 bias parameters, and finded a stronger relationship between (on_event_rate / off_event_rate) and [(bias_diff_on - bias_diff) / (bias_diff - bias_diff_off)] than (bias_diff_on / bias_diff_off). Is it responsible? Is it reasonable that I set weight for single on event 1 and weight for single off event -[(bias_diff - bias_diff_off) / (bias_diff_on - bias_diff)] instead of -(bias_diff_off / bias_diff_on) when I try to reconstruct the change of brightness?

If you were not sure about it, I wanna simply ask how should I infer actual threshold from bias parameters of EVK3 HD?