[Sutripta]

Originally Posted by Samurai

Controlling devices using real-time OS is something I have done in the past.

QNX / AIX? Or some other RTOS?

Originally Posted by SPIKE ARRESTOR

There are 2 prime reasons why Torque variations / fluctuations occur in an Engine:- a) Engine variation (variation in Mechanical systems, Friction, Imbalance of Pistons, Slap etc.) b) Fuel system variation (Specially Injectors) which results in "Surge" or "Spike" .

Talking of petrols or diesels?

Overall I'm not getting what you are saying. (About the controlled run with new ECU). Some communication gap somewhere.

Regards
Sutripta

[SPIKE ARRESTOR]

Originally Posted by Sutripta

Talking of petrols or diesels?

Diesel as well as Petrol, I'm talking about Torque fluctuations (more so for Diesels).

Overall I'm not getting what you are saying. (About the controlled run with new ECU). Some communication gap somewhere.

Dada, that is AT specific, I'm not sure if this process is applicable to Hyundai models. What are you not clear with?

Spike

[Sutripta]

^^^
Without getting into too much detail, a mass production item which needs individual handholding is very poor design.

Are petrol injectors paired?

Regards
Sutripta

[SPIKE ARRESTOR]

^^ What do you suggest, how must be the variations in Engine & Injectors be countered?

About Petrol injectors, I think so, not sure though. Tell me.

Spike

[Sutripta]

Originally Posted by SPIKE ARRESTOR

^^ What do you suggest, how must be the variations in Engine & Injectors be countered?

About Petrol injectors, I think so, not sure though. Tell me.

Spike

Hi,
Petrols are not as sensitive to injectors as diesels. So was wondering whether these are paired or not. Don't know.

Let me ask you a counter question. Supposing injectors are not paired. After a reasonable length of driving, can the ECU determine the injectors characteristics, and adapt?

Regards
Sutripta

[bblost]

Spike, from what I have read only the engine parts and not the transmission parts seem to be of interest. Why not just idle the engine. With varying loads.

Make it automatic. Why do you need human intervention.

If transmission is important, simply design a spinning track. The kind we use in exercise machines, which stay in the same place, but allows us to simulate walking.

[SPIKE ARRESTOR]

Originally Posted by Sutripta

Hi,
Petrols are not as sensitive to injectors as diesels. So was wondering whether these are paired or not. Don't know.

Will find this out tomorrow from one of my colleagues who is into Gasoline calibration.

Let me ask you a counter question. Supposing injectors are not paired. After a reasonable length of driving, can the ECU determine the injectors characteristics, and adapt?

Even, if you do not pair, the engine will continue to work "satisfactorily" for a novice, for a trained eye the variations within cylinders will be evident through vibrations. The case mentioned by you is quite similar to what happens when the injector starts getting old and its response time changes, this gets monitored continuously and gets adjusted.

Originally Posted by bblost

Spike, from what I have read only the engine parts and not the transmission parts seem to be of interest. Why not just idle the engine. With varying loads.

bblost, on an idling engine (on vehicle) how will you bring these loads into picture?

Spike

[bblost]

Idling engine with no acceleration, with acceleration, heavy acceleration.

Vehicles on top of a moving track. Like the exercise machines.
Now you can drive the vehicles using a machine, instead of a driver.

[Sutripta]

Originally Posted by SPIKE ARRESTOR

Even, if you do not pair, the engine will continue to work "satisfactorily" for a novice, for a trained eye the variations within cylinders will be evident through vibrations. The case mentioned by you is quite similar to what happens when the injector starts getting old and its response time changes, this gets monitored continuously and gets adjusted.
Spike

But can the ECU detect it? The ECUs eyes and ears are the sensors. How will we find out that ONE cylinder is producing slightly different power.
In the case of diesels, previously the only two things we set/ adjusted was the SoD and quantity injected. Injectors were separately adjusted for pop off pressure, and it was assumed that the lag between SoD and SoI would be identical.
We have to know far more things about the modern injector. And with multiple pilot injection (with each injection handling a minute quantity of fuel), its dynamic behavior becomes very critical. Even with exotic manufacturing processes (like electrochemical milling,) the variation amongst injectors is significant enough to cause emissions problems. This variation can only be measured on specialised test equipment, and I don't think by an ECU on a working engine.

Regards
Sutripta

[Samurai]

Originally Posted by Sutripta

QNX / AIX? Or some other RTOS?

It was a real-time yamos, native to Bell Labs. Had commandline interface like Unix, but behaved liked pSOS. pSOS, now that's something I haven't heard in a long time.

Originally Posted by Sutripta

^^^
Without getting into too much detail, a mass production item which needs individual handholding is very poor design.

Exactly my thought.

Originally Posted by Sutripta

After a reasonable length of driving, can the ECU determine the injectors characteristics, and adapt?

I thought this is what ECMs did. It is supposed to be the brain of a very complex control system, the one that constantly learns and adopts.

[Jaggu]

Ok i think it would be dependent on the ECU being used. I came to know recently that there is a hell lot of difference between a X manufacturer and Y manufacturer ECU recently. Right from design philosophy to various parameters, how they are read and controlled. Much more if its a diesel vs petrol ECU. All depends on the cost that the manufacturer strikes up with the automobile company

One thing i can confirm, in a petrol engine injectors are just plug and play. Have seen it being done in many cars.

GB talking to ECU needs to be trained? hmm interesting topic. Carry on

[star_aqua]

Originally Posted by Sutripta

But can the ECU detect it? The ECUs eyes and ears are the sensors. How will we find out that ONE cylinder is producing slightly different power.

yes, the ECU can detect it. the process is called as ZERO FUEL CALIBRATION. the drift in the injectors can cause the imbalance in the power produced by individual cylinder. it is carried out during the overrun phase where the vehicle is driving the engine and no demand from the driver. a sample injections are made into each cylinder individually and the resulting change in cylinder specific speeds are compared and adjusted in order to balance the cylinders for generating equal power.

regarding ATs, the calibration is required for achieving the good shift qualities which is based on the rpm and torque. the calibration is just a fine tune in case of transmission change.
for e.g if the vehicle is going down the slope, the torque is very less and shifts also should be possible at lowest possible rpm. where as when the vehicle is going up hill, the shifts should take place at higher rpms or it should not if the torque is not sufficient.
yes, learnt data can be transferred to the new transmission through tester.
learning will not be required for those transmissions which talk to engine ECU.

[Jaggu]

Originally Posted by star_aqua

learning will not be required for those transmissions which talk to engine ECU.

@Spikee the answer is here ^^

@Star you are really a genius dear

[bzr77k]

Couple of things (my $0.02 ):

1. As already mentioned by somebody, all mass-produced parts are subject to certain tolerances in their characteristics. If these characteristics, can be controlled by design, to not affect the performance, then there is no individual pairing required. But, if these tolerances, cannot be reduced due to cost/process/technology reasons AND if these tolerances can affect the system, then the ECUs are calibrated and the algorithms use a correction factor to account for these tolerances.
Each part (in this case, the injector for example... also some sensors) have tolerances which cannot be limited to the required amount and a correction factor needs to be introduced. This correction factor can be 'learned' into the system in two ways: i) by way of closed loop monitoring by the ECU itself (e.g. drive cycle), ii) by way of measuring the correction factor and programming it into the ECU (for e.g.: by reading the injector code)
The specific parameters that need to be compensated depends a lot on the system design and component selection (usually driven by cost/technology)
This applies to AT ECU's too. Although, i am not aware of the specific parameters that need to be learned.

2. In the Vehicle Assembly line, this correction is part of the process. The vehicles are driven on a Chassis Dyno to simulate the various load conditions. During this cycle, the closed loop parameters are programmed AND the engine/vehicle performance is monitored

- Immobilizer programming is a little different as there is a design-intent need to have unique ids. Unlike the above mentioned cases

Hence, when an ECU is reflashed, a drive-cycle with controlled loading is essential, as well as the flashing of measured parameters. These are typically possible only by specific OEM-designed tools and are not possible with the A.S.S. tools

[vina]

wow man, a techy !

Very interesting, but I'll ask you a few stupid questions just to make sure I understood:

Originally Posted by star_aqua

yes, the ECU can detect it. the process is called as ZERO FUEL CALIBRATION. the drift in the injectors can cause the imbalance in the power produced by individual cylinder. it is carried out during the overrun phase where the vehicle is driving the engine and no demand from the driver. a sample injections are made into each cylinder individually and the resulting change in cylinder specific speeds are compared and adjusted in order to balance the cylinders for generating equal power.

For my understanding (please let me know if I misunderstood)

I'm assuming this needs to be done, because the sensors sensing the specific speeds are (relatively) slow, and unless you inject exactly one cylinder at a time, data from multiple cylinders will get averaged out defeating the whole purpose.

Now if that were so, then I would assume that running a cycle where such conditions (car driving the engine) are deliberately entered can help calibration - is that correct?

also if that is the case then wouldn't the vehicle benefit the most during the run-in period? Most of the drift in many parts will happen right at that time? Kindly confirm.


Another thing: tolerance is different from drift so if only tolerance were the problem then I would think a learning algorithm wouldn't be required - you calibrate once and your are good to go. For drift you'll need calibration continually. Isn't that right?

Originally Posted by star_aqua

regarding ATs, the calibration is required for achieving the good shift qualities which is based on the rpm and torque. the calibration is just a fine tune in case of transmission change.
for e.g if the vehicle is going down the slope, the torque is very less and shifts also should be possible at lowest possible rpm. where as when the vehicle is going up hill, the shifts should take place at higher rpms or it should not if the torque is not sufficient.

okey ....

Originally Posted by star_aqua

yes, learnt data can be transferred to the new transmission through tester.
learning will not be required for those transmissions which talk to engine ECU.

this last bit I didn't understand at all. What do you mean by "transmissions which talk to engine ECU"