How does the ECU "learn"

[hogweed]

I've been trying to get seriously to grips with Motronic technology, and there's one (!!!) thing I don't understand.

Most Motronic 3.1 as fitted to M20s has no Lambda sensor, which otherwise could report the mixture back to the DME, which in turn could adjust it to keep it within limits - so how in God's name does the ECU learn anything with no feedback???

I've posted this several times over the years, but always drawn a blank...

Hogweed

[AlpineCab]

Does it just use a default value if no sensor is present or sensor is damaged?

[Templ8e30]

Apparently the ecu must go through several heat/cool cycles to read all the parameters on the chip.

I personally think the only thing the ecu learns is the idle speed

I agree that without a lambda sensor in the exhaust how can it learn the correct mixture for each combination of speed and load ? it is operating in open loop and has to refer to the base settings on the chip.

Just my 2p anyway and I'm no motronic expert.

Cheers,

Iain T

[Templ8e30]

Unless it operates like an industrial PLC where the program is loaded onto a ROM (read only memory) card and when initially powered up it reads the program off the ROM card and stores it in RAM(random access memory).

The RAM is much faster to read from and write to than ROM, having said that there is a lot of data in a plc program and it can read the lot from ROM in a few seconds.

Hmmmmm.

Cheers,

Iain T

[hogweed]

I'm assured that it uses the map in ROM as its starting point, loaded into RAM, then compares the readings from sensors etc with what's in there, and builds a new copy reflecting actuality rather than theory. This of course is why the car runs lke a bag of shite when you take the ECU out and put it back again - the new table is lost.

BUT as you say, it's open-loop, so how it gets feedback about how the engine's performing is the question... unless it measures the engine's responsiveness to having welly applied by measuring the rate of rpm increase... or something...