PID works extremely poorly without a good close base input. Given the variety of valve types/sizes/configs I doubt any type of "scaled other value" could properly fit the needs of a generic solution.

To modify your sentence, "as long as you're extremely close, PID will keep you there", what you said isn't really true, no. Also, PID is a general purpose control algorithm and non-optimal regardless of the purpose.

Found an interesting comparison on the topic:

http://ac.els-cdn.com/S0967066199001902/1-s2.0-S0967066199001902-main.pdf?_tid=c8a71c90-f878-11e3-a2fa-00000aab0f6c&acdnat=1403268519_046cba9dc985d85a0c5db69c84ac77c2

I think it is getting to far in the future ... PID .. Fuzzy Logic .. first let´s talk about the pro´s, con´s and the future of table based idle.

Maybe we should do 2 more topics with "Idle entry conditions" and "Measures of good idle" ???

Indeed far into the future, but I think the first couple pages of that paper were very useful to understand the challenges.

Also found it interesting that they use PI loops not PID.

At least now I won't ask quite so stupid questions. Initially thought an PID loop alone could handle the task, but they are only good in a linear system. There are just too many variables like alternator load, AC compressor on / off, atmospheric conditions and engine conditions.

The first one seems interesting, the second seems obvious? For given engine temperature RPM and AFR are constant with a low standard deviation. You could additionally say that, without closed loop behaviour, it resisted changes in RPM due to load. Start the first one for sure! :-)

FWIW, I absolutely HATE modern engines, especially diesels, that try to maintain idle RPM under driveline load while lugging the engine. Horrible. And in some cases actually dangerous. Old dumb engines are MUCH nicer in such situations, especially if they behave well to low RPM levels. It's not uncommon for me to drag an engine down to 500 while in slow traffic then let it pull itself out again with light throttle before continuing to accelerate normally. It's OK for a driver to stall! That's THEIR fault!


Those things can be incorporated into the feed-forward value in some cases. eg, AC is typically turned on by the ECU AFTER the ECU ups idle speed/airflow.

Fred.

Measures of good idle:
viewtopic.php?f=54&t=2469

Idle entry conditions:
viewtopic.php?f=8&t=2468