AFR/EGO Control

This section contains the settings necessary to do closed-loop Exhaust Gas Oxygen (EGO) control on the MegaSquirt III engine management system. Closed-loop EGO control allows the amount of fuel being injected to be changed so that the Air Fuel Ratio (AFR) matches the AFR set in the AFR table.

Settings

Basic EGO settings

These settings are used to control the behavior of the closed-loop EGO algorithm.

• Ignition Events per Step - While closed-loop EGO is active, how often the correction is run is determined by this setting. It is the number of ignition events per correction.
• Controller Step Size - This setting is only used with the "Simple" EGO algorithm. It controls how large each correction "step" is. So if the AFR does not match the desired AFR, and Controller Step Size is configured to be 1%, each time a correction is made, that correction will be 1%.
• Controller Auth - This setting controls the maximum amount of adjustment performed by the closed-loop algorithm.
• Active Above Coolant - Below the temperature defined by this setting, closed-loop EGO will not activate.
• Active above RPM - Below the RPM defined by this setting, closed-loop EGO will not activate.
• Active Below TPS - Above the throttle position defined by this setting, closed-loop EGO will not activate.
• Active below Load - Above the load defined by this setting, closed-loop EGO will not activate.
• Active Above Load - Below the load defined by this setting, closed-loop EGO will not activate.
• EGO delay after start - The time in seconds after engine-start before closed-loop EGO can be activated.
• Algorithm - This controls the type of closed-loop algorithm used:
  • Simple - This method of closed-loop EGO control is well-suited to use with a narrowband O2 sensor. If the current AFR (or EGO voltage for narrowband) incorrect, the amount of fuel being injected is adjusted by Controller Step Size every Ignition Events per Step ignition events. This method often results in the actual AFR oscillating above and below the target.
  • PID - This method incorporates a Proportional Integral Derivative control loop which tuned properly, adjusts the amount of fuel being injected to quickly get to the target, and then maintains the target without any oscillation (when tuned correctly).
• EGO Sensor Type - This setting enables EGO control and allows the user to choose between using a wideband sensor or narrowband sensor.

  The following settings are supported:

  • Disabled - No EGO sensor enabled.
  • Narrowband - Sensor in use is a narrowband sensor.
  • Wideband - Sensor in use is a wideband sensor.

• Number of Sensors - This setting is used to select the number of oxygen sensors in use by the MS3. Up to eight sensors can be configured.

EGO Ports -

The EGO Ports settings allow the user to select the input port used to read the signal from the oxygen sensor. The number of EGO ports available depends on the number of sensors selected.

The following ports are available for the first EGO input:

• Normal EGO
• JS5 (ADC6)
• JS4 (ADC7)
• EXT_MAP (ADC11)
• EGO2 (ADC12)
• Spare ADC (ADC13)
• CAN EGO
• CAN ADC01
• CAN ADC02
• CAN ADC03
• CAN ADC04
• CAN ADC05
• CAN ADC06
• CAN ADC07
• CAN ADC08
• CAN ADC09
• CAN ADC10
• CAN ADC11
• CAN ADC12
• CAN ADC13
• CAN ADC14
• CAN ADC15
• CAN ADC16
• CAN ADC17
• CAN ADC18
• CAN ADC19
• CAN ADC20
• CAN ADC21
• CAN ADC22
• CAN ADC23
• CAN ADC24

The remaining EGO ports have the following input options:

• Off
• JS5 (ADC6)
• JS4 (ADC7)
• EXT_MAP (ADC11)
• EGO2 (ADC12)
• Spare ADC (ADC13)
• CAN EGO
• CAN ADC01
• CAN ADC02
• CAN ADC03
• CAN ADC04
• CAN ADC05
• CAN ADC06
• CAN ADC07
• CAN ADC08
• CAN ADC09
• CAN ADC10
• CAN ADC11
• CAN ADC12
• CAN ADC13
• CAN ADC14
• CAN ADC15
• CAN ADC16
• CAN ADC17
• CAN ADC18
• CAN ADC19
• CAN ADC20
• CAN ADC21
• CAN ADC22
• CAN ADC23
• CAN ADC24

AFR/EGO Sensor Mapping

The AFR/EGO Sensor Mapping settings allow individual injectors to be associated with available EGO sensors.

The following injector channels can be associated with a sensor:

• MS3X InjA
• MS3X InjB
• MS3X InjC
• MS3X InjD
• MS3X InjE
• MS3X InjF
• MS3X InjG
• MS3X InjH
• V3 (mainboard) Inj 1
• V3 (mainboard) Inj 2

The following EGO selections can be made for each injector channel:

• EGO1
• EGO2
• EGO3
• EGO4
• EGO5
• EGO6
• EGO7
• EGO8

Tuning

Simple Algorithm with Narrowband Sensor

A narrowband sensor is only accurate at exactly stoichiometric mixtures for the fuel being used (14.7:1 for gasoline). At around 0.5 volts, the mixture is stoichiometric. For leaner mixtures (above 14.7:1 for gasoline, above 1.0 lambda) the voltage dips slightly below 0.5 volts. For richer mixtures, the voltage goes above 0.5 volts. This behavior means that it is not possible to hold an exact mixture when running closed-loop with a narrowband sensor.

Because of this, the best algorithm to use with a narrowband sensor is the "simple" algorithm.

The simple algorithm adjusts the mixture richer if the sensor reads lean, and leaner if the sensor reads rich. It adjusts Controller Step Size percent every Ignition Events per Step. This can lead to a small oscillation in O2-based correction once the AFR reaches close to stoichiometric.

The following steps are recommended when tuning the simple algorithm with a narrowband sensor:

1. Ignition Events per Step - When first tuning the engine, this should be set to a fairly low number (4-8) so that if the AFR is very far off, it is corrected quickly. Once the engine is better tuned, this number can be switched to a higher number to gain more stable correction behavior (8-16 or more).
2. Controller Step Size - When first tuning the engine, this should be set to 2% so that when correcting, the engine reaches stoichiometric quickly. Once the engine is well tuned, this should be reduced to 1% to gain more stable correction.
3. Controller Auth - When first tuning the engine, this should be set to 20% or higher. Care must be taken to watch how the algorithm is correcting. In some situations, it is possible for the sensor to read very lean when really the engine is running very rich. Once the engine is tuned, this should be set between 5% and 10%.
4. Engagement Settings - Most of the remaining settings control how and when the closed loop algorithm is engaged. Engagement with a narrowband sensor should happen when the engine is nearly fully warm, 500-1000 rpm above idle, below 80% throttle, below about 80% load, just above the lowest load seen when barely pressing the throttle, and at least 30 seconds after the engine starts. These settings are because the sensor must be hot to operate, must not be used at high load due to the fact that the engine should be operated rich of stoichiometric, and must not be used at very low load because the oscillations will cause the engine speed to oscillate.

Simple Algorithm with Wideband Sensor

Tuning the simple algorithm with a wideband sensor is essentially the same as tuning it with a narrowband sensor with the caveat that the AFR target table is used to set the AFR target. It is still recommended that the EGO algorithm not be used at high throttle position/load due to the fact that the accuracy of the wideband sensor decreases dramatically with pressure and temperature changes caused by high load.

PID Algorithm with Narrowband Sensor

When using a narrowband sensor with the PID algorithm, all the same recommendations for settings given in the section describing the Simple algorithm should be followed.

Additionally, since it is nearly impossible to keep the narrowband sensor from oscillating, it is recommended to start by tuning the 'I' term until the target is reached with minimal oscillation. Once this point is reached. It is recommended that very little (if any) 'P' term is used since the 'P' part of the PID algorithm causes instantaneous reaction, and the response of the sensor is not proportional to the distance from stoichiometric.

PID Algorithm with Wideband Sensor

When using a wideband sensor with the PID algorithm, the same steps as when using a narrowband sensor can be followed for tuning the 'I' term.

Additionally, since the response of most wideband controllers and sensors is linear with AFR, a larger 'P' term can be used to help correct for fast changes in AFR. Caution must still be used however since there is a significant delay between the amount of fuel being injected changing and MS3 registering an AFR change as a result.

Finally, a small amount of 'D' term can be used to help slow response during very fast changes. This helps reduce overshoot of the target.