EXCERPT FROM PRODUCER
"The LM-1 uses a very different approach that by its nature avoids the pitfalls of the PID loop and is inherently linear and independent of the
sensor response curve.
Think of it like of a fuel injection. You could theoretically run a fuel injector with a controlled current so it opens only a little for low fuel flow and more for higher flow. This would be very difficult, non-linear and
inaccurate, and also would vary widely between injectors, but would be possible. Or you can switch the injector on and off and vary the on-time vs. off-time precisely. The flow rate during on-time is fixed and the average flow per cycle is linearly related to the on- vs. off time. This
approach is simple, inexpensive and can instantly adapt from cycle to cycle precisely and predictable. The LM-1 uses a variation of the second approach(and using some more elaborate math
).
Looked at it another way, the LM-1 uses the sensor as the time-controlling element and comparator device of a delta-sigma-A/D design. The analog input of this A/D is not a voltage but lambda itself, the digital output is direct and follows instantly (delayed just by the sensor response time itself) without any feedback regulation or settling time. The reason the LM-1 is relatively inexpensive is that the firmware (and embedded math) has a zero production cost...
*********************************************
*********************************************
Some people who have modified the original DIY-WB to run with Bosch have found that it works, but I doubt they actually measured the output compared
to known lambda gases. We have tried that approach early in our development and found errors of up to 0.9-1.0 AFR when the exhaust temperature or flow varies because of the inherent temperature gradient and thermal inertia in the Bosch between the heater and the cells (cells are cooled or heated much faster than the heater element).
***********************************************
***********************************************
The Bosch specs assume control of temperature by the sense cell resistance(impedance). Although the sense-cell voltage (input to the PID regulator)is fairly independent of temp., its resistance is temperature dependent. The pump current (and therefore measured AFR) and pump cell resistance depend very much on the temperature of the pump cell.
In the conventional
PID approach measuring the pump cell resistance is very difficult to impossible. That's why the sense cell is used ... The LM-1 approach CAN
measure the pump-cell resistance and can therefore control the pump cell temperature directly. This avoids the inherent error due to the possible temperature gradient between sense and pump cell.
A customer of ours has tried the LM-1 last weekend on a NASCAR race car side-by-side (two side-by-side bungs) with a very expensive professional AFR meter (> $10k). He found that both showed exactly the same AFR but the LM-1 was responding instantly while the expensive unit took up to 1 second to finally settle on the same AFR after some overswing time (where it oscillated around the actual AFR). This is very typical of a non-optimized PID loop.
This customer (a professional race car electronics manuf.) now thinks that the LM-1 solution is the only one that can be used for closed loop control or even just AFR logging of NHRA dragsters, because with the others the race is over before the AFR settles in
."
+++++++++++++END EXCERPT+++++++++++++++
Very illuminating!
