| ABSTRACT
Simple tests
were performed to investigate the operating characteristics of zirconia
galvanic cells (lambda sensors) in automotive closed loop "three-way" emission
control systems. Commercially available cells were exposed to typical gaseous
components of exhaust gas mixtures. The voltages generated by the cells
were at their maximum values when hydrogen, and, in some instances, carbon
monoxide, was available for reaction with atmosphenc oxygen that migrated
through the cells' ceramic thimbles in ionic form. This dependence of galvanic
activity on the availability of these particular reducing agents indicated
that the cells were voltaic devices which operated as oxidation/reduction
reaction cells, rather than simple oxygen concentration cells. Such operation
explaIns why a cell that is used as a lambda sensor in a closed-loop control
system exhibits a sixfold or greater decrease in voltage output when the
exhaust gas composition changes from a slightly rich condition (lambda
= 0.995) to a slightly lean condition (lambda = 1.005). It also explains
why the voltage of a cell that is located downstream of a properly operating
catalyst normally remains at a low level as the air/fuel ratio oscillates
around the stoichiometric value but increases to a high level when ignition
misfire occurs at a rate that exceeds a certain value.
INTRODUCTION The test data presented in this paper were obtained during an investigatIon of the operating characteristics of zirconia galvanic cells (lambda sensors) in automotive closed-loop emissIon control systems. This investigatIon was prompted by observations of sensor reaction during an ignition misfire test program. During the first part of this misfire program (see Reference 1), lambda sensors were installed in the exhaust ports of a test engine which could be continuously operated at any selected air/fuel (A/F) ratio. |
When
the engine was operated without any induced ignition misfires at a speed
of 1800 rpm, these port sensors exhibited typical operating characteristics
by generating high level output voltages around 800 millivolt (mv) when
the A/F ratio was richer than the stoichiometric value. However, when an
ignition wire was detached from a spark plug, the voltage of the associated
port sensor dropped to its low level value of about 100 mv even when the
A/F ratio was as rich as 13 to 1.
The observed changes in the port sensor voltages when the engine was operating normally could not be explained on the basis of the changes in the concentration of oxygen in the exhaust. As Figure 1 illustrates, when the sensor voltages changed by a factor greater than 6, the oxygen concentrations in the exhaust changed by a factor much smaller than 2. Figure 1 also illustrates the same lack of correlation between the very pronounced step-wise changes in sensor voltages and the very gradual changes in the concentrations of reducing agents, such as hydrogen and carbon monoxide, in the exhaust. As Figure 1 indicates,
there appeared to be a correlation between high sensor voltages and rich
exhaust mixtures under normal conditions with no ignition misfires. Sensor
voltages were always close to the 800 mv upper levels even when the A/F
ratios were slightly less than the stoichiometric value and exhaust mixtures
were slightly rich (lambda less than 0.995). However, very rich A/F ratios
did not result in hIgh port sensor voltages when there was no ignition.
These results indIcated that some physical property of the unburned air/fuel
mlxtures, such as the relatively high
* A lambda value indicates how the actual proportions of oxygen and reducing agents in an exhaust mixture compare to the chemically correct or stoichiometric proportions that would be necessary for the compote reaction of the oxygen with the reducing agents. Lambda Is less than one when the exhaust is rich as a result of an excess proportion of reducing agents. and more than one when the exhaust is lean with an excess proportion of oxygen. |
|
concentrations of hydrocarbons. or the absence of other types of reducing agents, such as hydrogen and carbon monoxide, inhibited the voltage generating capabilities of the sensors. In the second part of the misfire test program, a sensor was installed downstream of the catalyst of a vehicle which could be operated with random ignition misfires in one or more cylinders at any selected percentage rate. When the vehicle was operated at a constant speed of 40 miles per hour and the misfire rate was |
2 percent or less, the voltage of the downstream sensor remained close to the 100 mv lean value as shown In Figure 2. However, when the rate of misfire was 3 percent or more, the voltage of the downstream sensor rose to the 800 mv rich value as shown in Figure 3. These results did not indicate any obvious correlation between sensor voltage and catalyst effluent composition. During this testing, the downstream sensor output voltage appeared to indicate that the effluent of the catalyst was "rich" when the ignition misfire rate was greater than |
|
|
