stoichiometric vs. AF sensors

[XfireZ51]

Excellent explanation on WBO2 from Innovate website at:
http://www.innovatemotorsports.com/resources/news3.php



How Wideband Sensors Work (and why Narrowband meters don't work for measuring AFR)
In product literature many times a narrowband oxygen sensor (4 wires or less) is also called an EGO sensor. Wideband sensors (5 wires or more) are often called UEGO sensors. We call them NBO2 and WBO2 sensors to make it clearer.
To understand how a WBO2 works one must first understand how Narrowband sensors (NBO2's) work. There are 2 kinds of NBO2 sensors. By far the most common type is the Nernst cell sensor described here. A (rarely used) other type is resistance based. It has a jump in resistance at 14.7 AFR instead of outputting a voltage. A NBO2 sensor consists of a porous ceramic with electrodes of special metal compounds on either side. One electrode is exposed to outside air, the other side is exposed to exhaust gas. The free-air side is grounded (one wire or 2 wire NBO2, or fed to the ECU as signal ground), the exhaust side electrode is connected to the signal wire coming out of the sensor.
When there is no oxygen in the exhaust gas, but unburned hydrocarbons, hydrogen or CO, these oxidizable molecules combine with oxygen ions migrating through the porous ceramic on the exhaust side electrode. This process creates a voltage between 0.07 to 1.1V between the electrodes. This process is not linear, but acts like a switch. Voltage is there when there are oxidizable molecules, and no (or very small) Voltage is there when there are none. The switchover point is where there is no oxygen in the exhaust, but also no oxidizable molecules. The half-way voltage is typically set to 0.45V for a normal oxygen sensor. For gasoline this would be at 14.7 AFR or the stochiometric ratio. The stochiometric ratio is where there is exactly the right amount of air for the amount of fuel. Typical engines produce more power when richer, because the combustion process is far from ideal and excess fuel assures that all air is consumed.
If there are unburned combustion products but also oxygen molecules in the exhaust, the combustion products rather combine directly on the electrode surface with the oxygen in the exhaust instead of the oxygen coming through the ceramic without creating a voltage. So even though there is fuel there, the sensor will read as if a lean condition exists. (My bolding(Dominic) but this explains the issue of a rich smell when larger cams with greater overlap are installed. Some of the fresh air charge from the intake side escapes through the exhaust "tricking" the sensor thereby causing the ECM to add fuel. That's why people run O/L with larger cams so as to take the O2 sensor out of the pulsewidth calc). These sensors require a minimum temperature of about 300C to work. Single wire sensors relied solely on the heating by exhaust gas. At idle that may not be enough. Multiwire NBO2 sensors have a built-in heater that helps keep the sensor at operating temperature at idle and also speeds up the heat up process to minimize open loop startup time.
The output voltage of a NBO2 sensor changes slightly (higher) the richer the gas is. Many people believe, that by calibrating this curve and linearizing it, one can make a cheap AFR meter using a NBO2 sensor. This does not work. Look at the following curve. This curve was not actually measured but calculated from the material constants and appropriate equations governing NBO2 sensors. In the real world the NBO2 output signal looks pretty close to this curve.


The green line in the graph is at about 880 mV. Because the offset of the curve changes with EGT, at 900C the 880mV output would mean about 10.3 AFR. At 500C the same voltage means 14.1 AFR. So without knowing precisely the sensor head temperature, there is no way of relating the output voltage to a specific AFR. An output voltage above 450 mV just means richer than 14.7, below means leaner. The numbers on a NBO2 meter are just paint. The only correct one is 14.7.
Because of these characteristics ECU's with NBO2 sensors use the sensor only during cruise and idle. They operate by checking if the voltage is below 0.45V, richen up until it is above 0.45V, then lean out, then richen up again and so on. Therefore the mixture oscillates around 14.7AFR. This AFR is optimum for minimum emissions. The oscillations are caused by the operation of the ECU, not by the sensor as many people believe. At WOT, as mentioned before, the engine wants a richer mixture to make power. Because the NBO2 output cannot be used there, all ECU's (not WBO2 based) will ignore the NBO2 signal and purely meter fuel by pre-stored maps relating TPS, MAP/MAF and IAT. There are products that claim to modify/replace the signal of one or more NBO2 sensors at WOT or boost to richen up the mixture. These products can't work because of the simple fact that ALL NBO2 based ECU's just ignore the signal at boost/WOT.

Wideband sensors
For many years manufacturers sought a method to extend the range of exhaust gas sensors to cover the entire range of engine operations. In the early 1990's NTK patented the pump-cell sensor now known as WBO2 or UEGO sensor. The first ones (NTK L1H1) were used on lean-burn Honda engines because engine operation could not be controlled by a NBO2 signal on the lean side of 14.7 (see curve).
It was quickly discovered that these sensors also work in a rich gas environment. Many modern turbo engines require tight control over the air/fuel ratio to keep emissions at minimum but nevertheless produce enough power. These applications keep the engines just shy of the onset of knock. The Bosch LSU4 series sensors were designed with that application in mind and are widely used by OEM's in turbo engines.
WBO2 sensors combine a regular NBO2 sensor and what's called a pump cell in one package. The pump cell is kind of the opposite of a NBO2 sensor. It can pump oxygen ions in or out of the sensor cavity. An electrical current through the pump cell transports the oxygen ions. If the current flows in one direction, oxygen ions are transported from the outside air into the sensor, in the other direction oxygen ions are transported out of the sensor to the outside air. The magnitude of the current determines how many oxygen ions/second are transported, just like the electrical current through a fuel pump determines the fuel transport rate.
Both, the NBO2 part and the pump cell, are mounted in a very small measurement chamber open with an orifice to the exhaust gas. The pumping rate of the pump cell is very temperature dependent. Therefore the sensor head temperature must be tightly regulated through a built in heater. A WBO2 controller (like the LM-1) monitors and regulates the heater to keep it at a constant temperature. In a rich condition the WBO2 controller regulates the pump cell current such that just enough oxygen ions are pumped into the chamber to consume all oxidizable combustion products. This basically produces a stochiometric condition in the measurement chamber. In that condition the NBO2 sensor part produces 0.45V. In a lean condition the controller reverses the pump current so that all oxygen ions are pumped out of the measurement chamber and a stochiometric condition again exists there. The pump cell is strong enough to pump all oxygen out of the measurement chamber even if it was filled with free air.
The task of the WB controller is then to regulate the pump current such that there is never any oxygen nor oxidizable combustion products in the measurement chamber. The required pump current is then a measure for the Air/Fuel ratio.
A basic diagram of a WBO2 controller is shown below:

The PID part in the WB controller regulates the pump current based on the NB-Signal, trying to hold it at a steady 0.45 Volts by varying the pump current. PID stands for Proportional/Integral/Differential and is a commonly used method for a feedback regulation system. Because of manufacturing tolerances the pump current can't be used directly as AFR measurement. For the same AFR different sensors require different currents. Therefore every sensor has built into its connector a calibration resistor called RCal in the diagram. The voltage drop over this resistor is actually measured by the controller (V = I*R). The sensor manufacturers trim this resistor during the manufacturing process so that the controller sees the same voltage drop for a given AFR.
This is how analog WB meters work. They are called analog because the input/output signals of the controller are smoothly varying voltages/currents. The PID controller can be implemented in a microprocessor or as analog electronic circuit using amplifiers, transistors and so on. Implementing it in a microcontroller does not make it a digital system. The LM-1 operates the WBO2 sensor differently. Its (pat. pend.) working principle will be explained in a future article.
Until next time... Keep On Tuning!
-Innovate Motorsports