[chetanhanda]

Originally Posted by esteem_lover

I think you guys just got a little crazier, i don't believe there is any artificial intelligence there in the normal ECU based engines. They are more like, push me i will run, or else i walk kinda machines. Right ?

..you dont believe that adjustments are done and stored according to driving style ?
So definately all of this also is crazy ? right ?

Originally Posted by chetanhanda

I think what learning is being referred to here is the storage of the fuelling and timing corrections according tothe driving style.
The ECU stores these values and corrects them according to your driving style, in a way "learning" the way you drive.
After a certain period the ECU does not do major corrections to these values, it simply stores them as a constant and applies the correction +/- to this constant.
The time period required for re-classifying these corrections as a constant according to driving style and not as short term fuel correction is what clevermax wants to know.

[esteem_lover]

Originally Posted by chetanhanda

..you dont believe that adjustments are done and stored according to driving style ?
So definately all of this also is crazy ? right ?

Hey, I am not the hooly gooly of these ECMs, but please do enlightnen me & the other simple foks about how these 'adjustments' are done & 'stored' in our Indian made cars. It would be of great help to many starting with me, no offence meant.

@chetan...read what you edit posted. Are you seriously saying that these things are incorporated in our simple simon ECU cars ?

[chetanhanda]

Originally Posted by esteem_lover

Hey, I am not the hooly gooly of these ECMs, but please do enlightnen me & the other simple foks about how these 'adjustments' are done & 'stored' in our Indian made cars. It would be of great help to many starting with me, no offence meant.

@chetan...read what you edit posted. Are you seriously saying that these things are incorporated in our simple simon ECU cars ?

Dude I'm also not the "hooly gooly" etc..so u can relax on the "simple people" and "enlightenment" sarcasm ok ?

I dont know what the other Indian cars do but I wouldnt call them simple simon also the way u just did.

Let me see If I can get some fresh data/screenshots over the weekend from a Honda P2P Obd2 ECU. I dont have my older screenshots ..

[ananthkamath]

Every OEM ECU has something called fuel trim. This is the "learning" thing that is being referred to in this thread.

The purpose of the feature is not to learn your driving style at all, AFAIK. It is to ensure perfect calibration for each individual vehicle by taking the minor manufacturing variations out of the loop, so to speak. In other words, a sensor may be 1-2% different than one that came off exactly the same assembly line, or an engine may have fueling demands minutely different from another exactly similar one.

Obviously the ECU has fueling maps inside it that are calibrated by the manufacturer on several test vehicles and burnt in the memory. Because each car has slightly different needs as stated before, ECU's (OEM ones, atleast) have something called a short-term and a long-term fuel trim. The purpose of the trims is to apply corrections to the map based on the closed loop feedback of the oxygen sensor (rich or lean). This is done over a period of time (I dont know how much), and when the trims are fully updated, the ECU is said to be calibrated for that particular vehicle.

The same strategy might be applied to ignition maps via use of the knock sensor or other methods that are obviously overkill for this discussion.

That's all there is to it, gentlemen. No more than that.

I think Chetan and RDKarthik will also have valuable inputs on this. Come on, Chetan, pop out some of the cool OBD stuff . Data, man, Data.

[Shan2nu]

The purpose of the trims is to apply corrections to the map based on the closed loop feedback of the oxygen sensor (rich or lean).

I'l tell you what. I used to have the engine check light coming on after the O2 sensor went bust. After a while it started switching between "on" n "off" modes. When the light was on, the engine would underperform and cut off at 7100, when the light was off, it would blast it's way through the revvs and had no limiter (which i mentioned to Karthik also).

However, for the past 1.5 months, the engine check light has not come on even once, inspite of the O2 sensor not being replaced, the engine is performing better than ever and the cut off is back at 7100!!!

So what do i make of this? Was the engine recognizing that the O2 sensor had gone (which made the check light come on at first), then started fiddling between on/off modes and finally when it wasn't replace, found a way around it? Or did my O2 sensor magically repair itself? LOL

Sometimes, these ECUs do things you just can't figure out.

Shan2nu

[chetanhanda]

Originally Posted by ananthkamath

Every OEM ECU has something called fuel trim. This is the "learning" thing that is being referred to in this thread.

The purpose of the feature is not to learn your driving style at all, AFAIK. It is to ensure perfect calibration for each individual vehicle by taking the minor manufacturing variations out of the loop, so to speak. In other words, a sensor may be 1-2% different than one that came off exactly the same assembly line, or an engine may have fueling demands minutely different from another exactly similar one.

Obviously the ECU has fueling maps inside it that are calibrated by the manufacturer on several test vehicles and burnt in the memory. Because each car has slightly different needs as stated before, ECU's (OEM ones, atleast) have something called a short-term and a long-term fuel trim. The purpose of the trims is to apply corrections to the map based on the closed loop feedback of the oxygen sensor (rich or lean). This is done over a period of time (I dont know how much), and when the trims are fully updated, the ECU is said to be calibrated for that particular vehicle.

The same strategy might be applied to ignition maps via use of the knock sensor or other methods that are obviously overkill for this discussion.

That's all there is to it, gentlemen. No more than that.

I think Chetan and RDKarthik will also have valuable inputs on this. Come on, Chetan, pop out some of the cool OBD stuff . Data, man, Data.

yes sir !!
Thats exactly what I was trying to explain to our freinds here.
who think its some "Artificial intelligence" ..its just a basic functionality of the ECU not a science fiction thing.

If you drive it peacefully the LTFT values will be different as compared to the values of LTFT which we see under hard accelaration.
Last time I checked after a soft reset via Obd2 within 10 miles my LTFT got stored at 2.3 , after pushing harder these were higher.
Again after reset I drove in mileage mode slowly the STFT was varying but the LTFT values were still not incremented...
After my basic DIY tests I felt it depends on how I rev my engine and drive instead of time or distance factor, the ECU tries to keep incrementing/decrementing 1st STFT and LTFT according to how I crazy I drive e.g- quick short bursts or WOT with vtec engaged etc etc.
Maybe this storage and correction can be called "learning" if viewed from another angle.
At that time I thought 2.3 was high and suspected lean AFR etc, but after discussing my readouts with some guys from Honda .. they said ECU will alow LTFT of max 20 and 2.3 is healthy , I got a xerox of that but I will have to search for it and see if I can manage some data over the weekend.

This correcton is done not just based on the TPS but also based on the health of the engine.. e.g. self correction for lean conditions by increase IPW and trims till no more correction can be done electronically


(Ananth maan ..I will be sending you a CD as soon as I get past the initial set of data..till then will only IAT values do ?)

[Sahil]

I dont think the ECU learns much of your driving style unless its an Automatic then your transmission tunes itself to your driving style.
Atleast in BMW's the car "learns" your driving style for 3500 kms and then locks it in, if you have a heavy foot it tends to down shift faster. There is a trick to reset this by turning the ignition halfway on and then pressing the accelerator for 30 seconds, after which you hear a sound from the gear box which indicates it has been rest. There is a noticable difference there after.

[ananthkamath]

Originally Posted by chetanhanda

till then will only IAT values do ?

IAT values are of little use unless there is a snail under the hood. In the coming months I see the need to very closely monitor IAT (hint hint )

Originally Posted by Shan2nu

So what do i make of this? Was the engine recognizing that the O2 sensor had gone (which made the check light come on at first), then started fiddling between on/off modes and finally when it wasn't replace, found a way around it? Or did my O2 sensor magically repair itself? LOL

I think what you describe is simply a sensor failure. When there is no lambda sensor input the car runs rich. Hence the loss in performance that you noticed. I think in OBD the ECU checks the sensor output to be varying continuously between slightly rich and slightly lean of 14.7:1. If that oscillation isnt happening, it assumes the sensor is bad and goes into open loop (wherein it runs rich to protect the engine).

[Shan2nu]

I think what you describe is simply a sensor failure. When there is no lambda sensor input the car runs rich. Hence the loss in performance that you noticed. I think in OBD the ECU checks the sensor output to be varying continuously between slightly rich and slightly lean of 14.7:1. If that oscillation isnt happening, it assumes the sensor is bad and goes into open loop (wherein it runs rich to protect the engine).

Then why isn't the check light coming on now? I know the car is running rich since the exhaust is a bit smelly but at the same time, performance is awesome.

Shan2nu

[vikram_d]

Originally Posted by Shan2nu

Then why isn't the check light coming on now? I know the car is running rich since the exhaust is a bit smelly but at the same time, performance is awesome.

Shan2nu

I also drove my swift around for a few days with the O2 sensor disconnected and the car used to perform really well.

[bblost]

Is it a good idea to reset the ECU before going on a 650+ kms highway trip.

I am going to try this out. But would love to read some feedback from anyone who has done this.

[Parthasarathig]

Are there any tuning boxes for petrol cars? So far all Ive seen is Diesel tuning boxes. How do we get more performance from petrol cars?

[souravc]

Originally Posted by bblost

Is it a good idea to reset the ECU before going on a 650+ kms highway trip.

I am going to try this out. But would love to read some feedback from anyone who has done this.

Even though this was 3 years back , would you remember your experience on the ECU reset , does it help ? Have a 700 km drive in a couple of days time.
Thanks

[bblost]

Originally Posted by souravc

Even though this was 3 years back , would you remember your experience on the ECU reset , does it help ? Have a 700 km drive in a couple of days time.
Thanks

Sorry Sourav.
But I don't think I did it.

EDIT: I did keep the battery disconnected for close to a month last year.
There was no difference in the way she drives compared to how she was prior to it.

[Parthasarathig]

Originally Posted by bblost

Sorry Sourav.
But I don't think I did it.

EDIT: I did keep the battery disconnected for close to a month last year.
There was no difference in the way she drives compared to how she was prior to it.

You may not notice any difference probably cause the settings may have been stored in the ROM of the ECU and I dont think memory from the ROM goes off without the power.
I however did get my ECU reset and updated I guess from the FORD A.S.S and Its working great now.

Anyone know of petrol tuning boxes? Would love to see how it affects the performance