At 01:32 PM 3/8/01, Tom Middleton wrote:
>I thought it would run rough after reconnecting the power, but the car fired up normally and the runs were conducted as soon as I got everything connected and started up....I was really surprised that numbers went up..?
> Very interesting !
The LH ECU does adapt to certain conditions for an individual engine. This allow the ECU to compensate for things like false air (vacuum leaks), injector imbalance, etc.
It is worth noting that if the engine is in perfect condition, the pre-programmed maps built into the ECU are an exact match for the engine conditions. The more that is wrong with the engine, the more adapting the ECU has to do.
The moral might be - If your engine runs/idles poorly immediately after you disconnect/reconnect the battery or ECU, there may be some problems that the ECU has to compensate for.
If, on the other hand, your engine runs very well immediately after the battery or ECU have been disconnected/reconnected, there shouldn't be much wrong with it!
> ----- Original Message -----
> ----- Wally Plumley wrote:
> The LH ECU does adapt to certain conditions for an individual engine. This allow the ECU to compensate for things like false air (vacuum leaks), injector imbalance, etc.
> It is worth noting that if the engine is in perfect condition, the pre-programmed maps built into the ECU are an exact match for the engine conditions. The more that is wrong with the engine, the more adapting the ECU has to do.
I agree that the more there is wrong with the engine, the harder it will have to work to compensate. But I would add that not only would it compensate for the symptoms you mentioned that are due primarily to age, mileage, and minor problems, but it also performs this task even when a vehicle is brand new, to cover for the inevitable spread among the many component tolerances (injector outputs, MAF sensor signal, valve overlap variations, compression differences, etc., etc., etc.). Naturally, it is
preferable to set each component performance specification to a range that is transparent to proper operation, but frankly, that's not always cost effective (EGR valves, and their effect on everything else, were a good example).
I expect that the baseline calibration (maps, lookup values - whatever we want to call them) were dialed in on "known mean cars". This means that all of the components used to build up those test cars were in the middle of their production specification band. Naturally, there are allowable tolerances in the performance specification of each part, which is what
leads to the spread mentioned above. I have seen brand new cars that idled fairly rough until allowed to learn, and then were silky smooth.
When I was with GM back in the eighties, we would intentionally build up high and low limit cars (all components stacked up toward either end of their tolerance spectrum) in order to check the cumulative effect on emissions, driveability, and fuel economy, and the ability of the system to compensate through adaptive learning (emissions being the most sensitive). The only areas that I can recall that learned included base fueling (using the oxygen sensor), base spark advance (using the knock
sensor), and idle control (using either the ISC or IAC valve position).
If the system was running closed loop, and held a certain condition long enough, it would allow the update of the value being used in that cell through a stored multiplier. When the power is lost due to a battery disconnect, running reset, or such, all of the multipliers would return to null, and thus need to be relearned during the subsequent driving.
> The moral might be - If your engine runs/idles poorly immediately after you disconnect/reconnect the battery or ECU, there may be some problems that the ECU has to compensate for.
> If, on the other hand, your engine runs very well immediately after the battery or ECU have been disconnected / reconnected, there shouldn't be much wrong with it!
And everyone should remember that a system can only learn where it has some sort of feedback sensor like those mentioned above, and is running in a closed loop mode. Closed loop is enabled only when certain running criteria are met (engine up to temperature, etc.). Normally, any operation above certain rpm's and throttle angles kicks it out of closed loop and back into open loop (I don't know what these thresholds are on the various 928's). This is why I posted last year with my question about
the "spirited driving" recommendations, since most spirited driving is at heavy throttle and/or higher rpm's, and I also couldn't understand what type of sensor could monitor driving "style".
Mark Reifenberger- Black 84 S in Atlanta.
>So my question is, does driving style really affect the way the brain learns? If I disconnect the battery and dive it hard, will it be even better? Or is it just a matter of getting the full range of throttle under some variety of conditions?
it a pretty complicated system and there is quite some confusion around about what it does and what not. Here comes what I collected from the factory manuals, the owner's manual and various Bosch fuel injection literature:
The brain stores different types of parameters with some of them continuously updated, some not. The latter are the parameters for the adaptive idle control. Erasing these can (but need not) cause the car to idle very rough after start-up. There are two possibilities to get these figures back: the quick one is to carry out a system adaptation. That takes about two minutes but you need to have the Bosch hammer. The second possibility is to let the car idle for a while to give the brain the chance to collect the necessary figures.
Then there is another set of data which is not affected by the Bosch hammer but forms the basis of the adaptive system: For each parameter there is a set of pre-programmed values and second set of "variable" correction figures for this base set to adapt to match the settings to the individual engine (but not to your driving style) These correction figures are learned during normal operation using sensors (e.g. knock sensors) and closed-loop control.
The problem with closed-loop control is that it typically takes many iterations to reach the optimum point, so the better the figures to start with, the faster optimum will be reached.
One example is the max ignition advance, right before knocking occurs.
During acceleration where the engine runs up quickly through the rev band, it is hardly possible to use the knock sensors for closed-loop optimization, due to the quickly changing conditions. Instead the brain uses the correction figures it found to work in the past and start from there. If knocking occurs, the ignition is retarded, if not, it gets advanced a little more and a new correction value is stored for the respective engine condition.
If these figures are erased, you might find your car to knock a little or be down on power until you have "probed" the full rev band for a couple of times.
Then there is a third set of parameters which also gets continuously updated but to match the engine to your driving style. They are updated with a time constant of about 10 minutes. The brains try to optimize the settings for the operating conditions: if you use only little power, it tunes the settings for max economy, if you open it up, it is tuned for max power.
E.g. if you leisurely cruise around town for a while and then decide to drive a little more spirited, it will take about 10 minutes until the car reaches it's full performance again.
So after disconnecting the battery you should expect the car to idle rough and be a little lower on power for the first miles. If you can't drive it spirited during the first outing, don't worry, sooner of later you will have the chance and all should be back to normal again.
Hope that helps
90 S4 Auto