Electronic Engine Control Algorithm. Is this Correct?
 

Is this pretty much how modern electronic engine control algorithms work?

The engine management control program's goal is to hold the air/fuel mixture at a constant target value as measured by the O2 sensors. The program is written as a closed loop feedback loop with feed forward. A regular feedback control would wait for the O2 sensors to move away from the target O2, then start changing fuel rate to get the O2 back to target. This control is relatively slow. Feed forward is provided by using a mass air flow sensor. The feed forward speeds up control. In fact, if the feed forward were perfect, there would be no reason for feedback.

The manufacturers want a certain target level of richness/leanness in the combustion chamber. They use O2 sensors to measure this, and decide on a target O2 level for the control program. You also have a starting air/fuel ratio target. So, you step on the throttle, the throttle plate opens, and the MAFS sees air flow, and uses the ratio target to set the fuel flow. The O2 sensor reading should be on target, but if it's not, it will change the air/fuel ratio setpoint.

It continues to work like this. As long as the O2 readings are at target, the algorithm maintains things at a constant air/fuel ratio. When the MAFS sees a change in air flow, fuel flow is changed per the current air/fuel target. This continues unless the O2 moves away from its target. At this point, the control program starts to change the air/fuel target.

A bunch of preset tuning factors are needed to coordinate all this stuff going on. These factors are a function of time delays and anticipated rates of change of the variables.

Nope, that's not really it.

The engine management figures out how much air the engine is taking in by using the MAP/MAF/whatever, the throttle position, intake air temp, and other sensors. It crunches those numbers to get an overall "how much air is going in" figure, and calculates a basic injector opening time. Other sensors, primarily the O2 sensor, then provide fine-tuning for that time value.

I do not know how much "authority" the O2 sensor feedback loop has, but it is definitely not the primary metering device. For one thing, most O2 sensors respond relatively slowly compared to the engine revolutions.

There are plenty of "special cases" built in, however. When the throttle goes from closed to more open, the engine adds extra "acceleration enrichment" so the engine doesn't bog or stumble when the throttle position changes. When the throttle is wide open, it adds "full throttle enrichment" for more power (after all, that's why most people push the pedal all the way down!). When the engine is cold, the mixture is also enriched because cold engines don't burn fuel as well as warmed-up ones.

In all of those cases, the engine goes "open loop" and ignores the O2 sensor feedback entirely. It just relies on the basic metering and temperature/RPM/whatever adjustments.

-soD

so what are some symptoms or indications that the mass air flow sensor is not working properly? Thanks! I don't know much about engine control and stuff. Thanks!

When mine went bad, when you would let off the throttle to come to a stop. When you went to go again the engine had stalled.

If you think yours may be going bad, they do make MAF sensor cleaner you can use.

Don

Check engine light, stalling, lack of power, poor thottle response, very poor cold running.

some_other_dave -

In my experience, except for cold-starts, my ECU/PCM almost never leaves closed-loop. Maybe this is another benefit of how we drive. I can datalog, so I should double-check this. Would you make the claim that even when the ECU/PCM reports it's in closed-loop, it may not actually *be* in closed-loop?

What do you think of these explanations? :

Tips on Reading Gauges; Air-Fuel Ratio Monitor
What do you think of this? :

Tech - Closed Loop
CarloSW2

I suspect that is primarily due to the way we drive. If you mashed your throttle pedal all the way down, you would likely see the engine management go open-loop, at least for a second or two. (I know the duration of acceleration enrichment is a tunable parameter in Megasquirt, and is often in the "several seconds" range.)

I would assume that any reporting that the ECU does of open/closed loop state would be accurate, as the state is maintained inside the ECU itself. There may be other factors affecting the indication (e.g., the OBDII bus only updates information every second, and the state is only open-loop for a half-second at a time) but I'd expect it to be reasonably accurate.

--The two articles you quote from match my experience and reading on this subject pretty well. Note that these are all describing the "Narrow Band" O2 sensor; the "Wide Band" type can actually tell how far off of "stoichiometric" you are, which gives more tuning options. (Which is why that type of sensor is often used in vehicles that use lean-burn technology.)

-soD