I have converted my car to LPG, and i've been tuning it today with a wideband lambda controller. It runs pretty okay, but i'm not yet satisfied.
The weird thing is, it accelerated best when the lambda value was tuned to about 0.78. That figure seems a bit too rich for me, so i'm wondering whether the measurement is off. People on the internet claim it doesn't matter for the oxygen sensor what fuel is used, but I beg to differ.
To my understanding, a wideband lambda sensor consists of a (narrowband-like) oxygen sensor and an oxygen pump. The pump is used to keep the lambda value inside the oxygen sensor at 1, and the needed electrical current for that, is translated to the actual lambda value of the exhaust gases.
My point is, LPG has a stoich ratio of about 15.5, and petrol 14.7. So LPG needs more oxygen. If then the oxygen excess or shortage is translated to a lambda value, that value may be different for LPG than for petrol. After all, every LPG 'molecule' needs more oxygen molecules to react with than a petrol 'molecule'. So the same pump current and same oxygen supply can compensate less LPG than it could compensate petrol.
Simplistically thinking, i'd say the actual lambda value given by the sensor differs as much as the stoich ratio of both fuel differs; 15.5/14.7 = 1.054. That means my measured lambda value of 0.78 translates to 1.054*0.78 = 0.82. And that's roughly the best ratio to maintain for maximum power, for petrol at least. I see this as a confirmation of my little theory, but i'm curious if there's someone else that can disprove or confirm it.