Ideas on Ecu+ Knock Sensing / control
Ideas on Ecu+ Knock Sensing / control
Tom,
I just wanted to open up a discussion about knock detection and correction now that you are working away on new features.
As you have previously mentioned, MUT-II knocksum is only reported at max 100 times persecond and the Ecu+ dsp unit logs voltage at every fire event. When tuning evo's it is all about keeping the ecu happy from a knock sum perspective. From working with several tuners it seems that in many cases the stock ecu detects knock well before an actual knock event may occure.
From what I understand, products like the Exede attenuate the factory knock sensor signal by some percentage, due to this, they are able to consistently produce more power than a re-flash because they are able to run a little more timing than you would with a reflash (and still keep the ecu happy.
One of the problems with voltage based kncok detection (as i see it) is that the level of background noise, and therefore output voltage, changes through the rev range. There are a few products (UTEC, and KnockLite, come to mind) that allow you to program the knock sensitivity by rpm range (http://www.turboxs.com/more_info.php?ID=212). This process involves running the engine through various rpm ranges and establishing a baseline voltage from each range then setting the detection curve rather than a fixed value.
Now, what I would like to propose for Ecu+ is the ability to attenuate the knock sensor output that goes to the ecu AND implement an rpm range based knock detection similar to the products mentioned above.
I think this would allow Ecu+ users to extract a little more power and still have a good safety net to fall back on.
Obviously, is the signal is over attenuated and the user did not setup the Ecu+ knock control strategy properly, there would be a serious risk of trashing your motor.
Thoughts????
I just wanted to open up a discussion about knock detection and correction now that you are working away on new features.
As you have previously mentioned, MUT-II knocksum is only reported at max 100 times persecond and the Ecu+ dsp unit logs voltage at every fire event. When tuning evo's it is all about keeping the ecu happy from a knock sum perspective. From working with several tuners it seems that in many cases the stock ecu detects knock well before an actual knock event may occure.
From what I understand, products like the Exede attenuate the factory knock sensor signal by some percentage, due to this, they are able to consistently produce more power than a re-flash because they are able to run a little more timing than you would with a reflash (and still keep the ecu happy.
One of the problems with voltage based kncok detection (as i see it) is that the level of background noise, and therefore output voltage, changes through the rev range. There are a few products (UTEC, and KnockLite, come to mind) that allow you to program the knock sensitivity by rpm range (http://www.turboxs.com/more_info.php?ID=212). This process involves running the engine through various rpm ranges and establishing a baseline voltage from each range then setting the detection curve rather than a fixed value.
Now, what I would like to propose for Ecu+ is the ability to attenuate the knock sensor output that goes to the ecu AND implement an rpm range based knock detection similar to the products mentioned above.
I think this would allow Ecu+ users to extract a little more power and still have a good safety net to fall back on.
Obviously, is the signal is over attenuated and the user did not setup the Ecu+ knock control strategy properly, there would be a serious risk of trashing your motor.
Thoughts????
About attenuating the knock signal - I'd need special hardware to do that, and the ECU+ doesn't contain anything that could be used for that right now. An alternative would be to make an external circuit to do it, and I'd probably need a bunch of interest for this to be worthwhile.
Doing an RPM-based knock voltage correction table is feasible. I think Jeff Jeske mentioned that in another thread. Post some thoughts as to how it might work from the user's perspective - that'd be a good start on getting it implemented in this or the next feature cycle.
Tom
Doing an RPM-based knock voltage correction table is feasible. I think Jeff Jeske mentioned that in another thread. Post some thoughts as to how it might work from the user's perspective - that'd be a good start on getting it implemented in this or the next feature cycle.
Tom
So the table would just be user defined voltage at various points. In order to find your starting point, you would start up Ecu+ and slowly run the car through the rev ranges to log the knockvoltage and establish the base lines.
Not sure if you would want to do rate or change based detection (spike detection) or absolute voltage or voltage delta from the range setpoint.
I think one setting for the reaction would be sufficient (like what you already have, pull x degrees and y fuel).
Thing is without knock signal attenuation, I'm not sure it would be worth the extra time and effort on your part since the stock Ecu is already good at pulling timing.
If we had this feature though it would allow some of use to work on figureing out the stock knock sensor filters while still having a margine of safety.
Not sure if you would want to do rate or change based detection (spike detection) or absolute voltage or voltage delta from the range setpoint.
Code:
0-1000 0.5v 1001-2000 0.6v 2001-3000 0.7v 3001-4000 1.2v 4001-5000 1.3v 5001-6000 1.6v 6001-7000 1.5v 7001-8000 1.4v
Thing is without knock signal attenuation, I'm not sure it would be worth the extra time and effort on your part since the stock Ecu is already good at pulling timing.
If we had this feature though it would allow some of use to work on figureing out the stock knock sensor filters while still having a margine of safety.