The Bosch LSU4 wideband sensor is a low-cost sensor meant to provide reproducible lambda reading with a given fuel in a given application during a reasonable sensor lifetime (eg. upto 1000 hours in very good conditions).

Bosch specified a so-called Ip=>Lambda curve that gives a close enough lambda reading with typical gasoline in a typical application.

The lambda reading by this Ip=>Lambda curve deviates from real, depending on several factors:

* exhaust gas temperature (the sensor temp is controlled, but the gas temperature varies)

* exhaust gas pressure (in general, this is not even measured, but rather consistent for a given loadsite)

* Bosch RCal calibration resistor (placed into sensor connector housing between pin2 and pin6 by Bosch, measurable with a DVM, usually 80..200 Ohm value)

* chemical composition of fuel

* other factors like air humidity

Even if you know exact lambda, there is no predefined "best lambda" value. At high power, rich mixture (like lambda=0.77) is safer, but somewhat leaner (like lambda=0.82) at least upto 0.85 can give more power. But richer mixture allows higher ignition advance, that also allows more power. All this is engine (head, compression) specific, and depends on loadsite (RPM, manifold pressure) and even intake air temperature and humidity.

Since there is no absolute truth about lambda, reproducibility is most important, and tuning - though largely simplified by wideband lambda meters - still requires expertise and ears.

For a digital reading, the last digit can differ by +-0.01 : Therefore when comparing 2 digital lambda meters (even under same conditions), or one meter with different sensors, a difference of 0.02 does not necessarily indicate a problem.

----

AFR - Air to Fuel Ratio

Air to Fuel Ratio is an unfortunate, but widely used value, the mass ratio of air compared to fuel. Lambda=1.0 is AFR=14.7 for typical pump gas, although this is continuously decreasing as the alcohol (ethanol) content of gasoline climbs at most places (5..10% is not uncommon).

With high ethanol content, like the more and more popular E85 (with 85% ethanol and 15% gas), the AFR value is much lower.

It is recommended to use lambda value, because lambda=1.00 is much more natural value for stoich than AFR=14.7 , and at least the meaning is the same for different fuels

----

Calibration

* AFR multiplier : only matters, when using AFR reading. HAS NO EFFECT for LAMBDA READING. default: 147 (which means 14.7 reading instead of 1.00). Because of rounding-down internally, 148 might make sense. Also possible to set different value for other fuel, or if different reading is required for some reason (cheating?).

* pump_pw_zero dutycycle : calibrated in factory for "least absolute value pump current" (measured in uA or mV/kOhm). Usually 101+-3. If set higher, device will read slightly richer than real (pump_pw_zero += 2 will make appr -0.01 lambda difference)

* nernstdc_target can be used to set nernst (WBO2 pin1) target DC voltage (in 20mV steps) to +450mV compared to pump- (WBO2 pin5, which is usually appr +4.1V above GND). Default: 229 (0xE5) was used earlier, recently it is calibrated in factory: 229 might result in slightly lower or higher than +450mV. For example if 229 results in +437mV, than 230 will give +457mV which is better (actually, 230 and 231 values seem to be very common). If 229 is used in a device where the optimal is 231, the effect in lambda-reading is appr -0.01