O2 + NOx!

[1project2many]

Well, it seems as though the diesel world is starting to gain on gas engines in terms of technology. Apparently most big players are now using a catalyst system to control NOx by injecting urea fluid into the exhaust upstream of a special catalytic converter. The fluid reacts with NOx to break it down to it's original components, nitrogen and oxygen. In order to monitor catalyst effectiveness and to aid in particulate filter regeneration, some manufacturers are using a dual purpose sensor which monitors O2 and NOx levels. While the article I read didn't say the O2 portion of the sensor was a wide band type, it did say the sensor was used to monitor O2 levels for particulate filter regeneration downstream of the converter which suggests that it needs to provide more information than just "stoich" or "not stoich." For those interested in getting a better handle on exhaust emissions while tuning it might be worthwhile to investigate this a bit deeper.

[daleulan]

The NOx sensor's lambda cell is indeed a wideband cell. In most configurations the NOx sensor and its electronics come as a calibrated pair and a CAN bus of some kind is used to communicate the information from the NOx sensor to the ECM. I would imagine in heavy-duty applications the bus would be SAE J1939, for light-duty applications I don't know but using a CAN snooper (like the Vector CANalyzer - IMHO the best CAN snooper out there) one should be able to figure out the protocol if you tap the wires and watch the traffic. VW uses it in their TDI applications, GDI applications may find this sensor useful as well although I am not sure if they are using it there. The sensitivity of the sensor, though, is set up for typical diesel numbers - probably 0 to 500ppm or thereabouts. You can probably get meaningful data post-catalyst on a typical stoichiometric engine but the pre-catalyst NOx numbers of a stoich engine are going to saturate the cell as engine-out stoich NOx is usually between 800 and 3000ppm under most conditions (excluding idle) on any engine I've tested. Generally I find that I get good NOx numbers if I can maintain a postcatalyst stoichiometric sensor voltage at 'some value' which is dependent on the fuel, but for gasoline and most catalysts I have worked with it is about 0.55 to 0.62 volts and generally in this location the best NOx/CO balance happens at exactly stoichiometric. Generally the switching sensor is very accurate in the postcatalyst position and wideband sensors are less accurate so I am not sure if the lambda reading from the NOx sensor would be that meaningful in that position.

[daleulan]

Oh yes, if really good emissions were your final goal of closed-loop control and you wanted to stick with a GM P4, one could probably hack in postcatalyst O2 sensor software into any of the P4-based controllers. Add a PI loop in around a 100ms computational rate. Use an input with maybe a 220k impedance to ground and use that for the postcat sensor. Set the setpoint to maybe 0.58 volts. The ouput of the PI controller is used to offset the target voltages in the existing O2 loop by maybe +/- 150mV or so from the calibrated values. You could actually get away with a proportional term only but a PI controller with a reset time of maybe 30 seconds gives more stable emissions control. I haven't done it since my car is running a Bosch ME7 and my truck is a diesel without O2 sensors at all but it should be possible.