Adventures in EAE (or, Why do I abuse myself like this?)
#21
Retired Mech Design Engr
iTrader: (3)
Join Date: Jan 2013
Location: Seneca, SC
Posts: 5,009
Total Cats: 857
If your spikes are in the 14-15 range, then I suggest you fix it by adjusting the min - max numbers in MLV to make them look smaller. Problem solved.
See, "Henry's Awful Mistake" by Robert Quackenbush.
See, "Henry's Awful Mistake" by Robert Quackenbush.
#22
Point taken. So I've re-focused my efforts on the rich spikes on lift, which drop into the low 12's to mid 11's.
Below is an example of the "RPM Effect". The graphs are an overlay from the same log, but at different times, In the heavier graph the peak RPM to the left of the marker is ~4300 RPM. In the lighter graph the peak RPM is about 3700 RPM. At the knee of the MAP curve, the starting AFR is significantly different; 12.8 for the higher RPM vs 13.7 for the lower RPM. I'm guessing this is solely a result of the VE table (especially for the low-RPM event because the WF is relatively stable - which to me indicate equilibrium between AtW and SfW).
I think something different is happening with the higher-RPM event.There, the WF is increasing and I think this is being driven by the AtW RPM curve, which is increasing by a greater "delta". What's interesting is that the AFR's or both end up at the same amount - 11.7! That is saying to me that I'm sucking too much off the walls at lower RPMs, which is driving the AFRs rich.
I think this is why EAE drives people mad. Anyway, here's the plot that I'm referencing.
Next Steps;
I've "normalized" both of the RPM curves to be at 100% at 3K RPMs, and shifted bot of the MAP curves up or down accordingly. When the rain lets up later this afternoon (I hope), I'll get some baselines before starting to tweak either of the RPM curves.
Below is an example of the "RPM Effect". The graphs are an overlay from the same log, but at different times, In the heavier graph the peak RPM to the left of the marker is ~4300 RPM. In the lighter graph the peak RPM is about 3700 RPM. At the knee of the MAP curve, the starting AFR is significantly different; 12.8 for the higher RPM vs 13.7 for the lower RPM. I'm guessing this is solely a result of the VE table (especially for the low-RPM event because the WF is relatively stable - which to me indicate equilibrium between AtW and SfW).
I think something different is happening with the higher-RPM event.There, the WF is increasing and I think this is being driven by the AtW RPM curve, which is increasing by a greater "delta". What's interesting is that the AFR's or both end up at the same amount - 11.7! That is saying to me that I'm sucking too much off the walls at lower RPMs, which is driving the AFRs rich.
I think this is why EAE drives people mad. Anyway, here's the plot that I'm referencing.
Next Steps;
I've "normalized" both of the RPM curves to be at 100% at 3K RPMs, and shifted bot of the MAP curves up or down accordingly. When the rain lets up later this afternoon (I hope), I'll get some baselines before starting to tweak either of the RPM curves.
#23
Tweaking Enginerd
iTrader: (2)
Join Date: Mar 2013
Location: Boulder, CO
Posts: 1,776
Total Cats: 358
Unfortunately, you won't be able to address the decel rows with an MS2 :(
If you are close from an EAE perspective, you can use MAPdot and TPSdot time based AE to fill in the gaps. Set the scaling slider (if there is one on MS2?) To 50%, zero out the MAPdot at first and focus on TPS. TPS gets the near instant transient, MAPdot works on the transient after TPS.
Filtering is found in the general settings tab.
Again, I do not think you can fully eliminate the rich-on-lift condition. Even with an EAE % of 0% (PW = deadtime + correction) you will still get a rich condition.
If you are close from an EAE perspective, you can use MAPdot and TPSdot time based AE to fill in the gaps. Set the scaling slider (if there is one on MS2?) To 50%, zero out the MAPdot at first and focus on TPS. TPS gets the near instant transient, MAPdot works on the transient after TPS.
Filtering is found in the general settings tab.
Again, I do not think you can fully eliminate the rich-on-lift condition. Even with an EAE % of 0% (PW = deadtime + correction) you will still get a rich condition.
Last edited by Ted75zcar; 06-08-2019 at 07:10 PM.
#25
Tweaking Enginerd
iTrader: (2)
Join Date: Mar 2013
Location: Boulder, CO
Posts: 1,776
Total Cats: 358
Go with the lowest amount of filtering you can while minimizing aliasing. You probably won't be able to eliminate all of the aliasing (as shown in the pic below) while keeping valuable information. I have 70 on map, 75 on RPM, 80 on Lambda for reference.
#26
Retired Mech Design Engr
iTrader: (3)
Join Date: Jan 2013
Location: Seneca, SC
Posts: 5,009
Total Cats: 857
To report. This has been an eye-opener. Looking closely, it becomes apparent that the lean spike after lift is indeed not an EAE phenomenon. I turned on low pulsewidth correction, and returned my high RPM EAE curves back to previous settings. The lean spike has gone away.
I think Ted is right on with that.
Also, if I may tangent, my EAE is good under load, but when free-reving from idle, AFR’s do not track well. Hmm could be VE also, as this hits areas not normally seen. Any comments?
I think Ted is right on with that.
Also, if I may tangent, my EAE is good under load, but when free-reving from idle, AFR’s do not track well. Hmm could be VE also, as this hits areas not normally seen. Any comments?
#27
Yesterday I changed the filtering and, frankly, I can't see much if any difference in the logs. I was expecting them to be more jagged, but they look pretty much the same to my untrained eye. I bumped the filtering values for MAP, RPM and AFR up to 90 (just for the hell of it) and this is what I saw (bold is w/ less filtering);
Anyway, I'll leave it like that.
Next step;
Looking at a comparison of data from the same log where the MAP is the same, but the RPMs are different seems to be telling me that I need to have to trend the SW RPM curve down (instead of up, as it is now). Si, I'm going to "flip" that curve - still centering it at 100% at 3000 RPM - and see what happens. Here's what I'm talking about;
I've decided not to go as "deep" in the idle RPM cells as it is "tall" in the original, but the trend is still upwards from there.
Anyway, I'll leave it like that.
Next step;
Looking at a comparison of data from the same log where the MAP is the same, but the RPMs are different seems to be telling me that I need to have to trend the SW RPM curve down (instead of up, as it is now). Si, I'm going to "flip" that curve - still centering it at 100% at 3000 RPM - and see what happens. Here's what I'm talking about;
I've decided not to go as "deep" in the idle RPM cells as it is "tall" in the original, but the trend is still upwards from there.
#28
To report. This has been an eye-opener. Looking closely, it becomes apparent that the lean spike after lift is indeed not an EAE phenomenon. I turned on low pulsewidth correction, and returned my high RPM EAE curves back to previous settings. The lean spike has gone away.
Also, if I may tangent, my EAE is good under load, but when free-reving from idle, AFR’s do not track well. Hmm could be VE also, as this hits areas not normally seen. Any comments?
#29
So, with my flipped SfW RPM correction, I've noticed something interesting. It looks like the lean peak is being lessened, and the rich dip is coming up. Here's a representative log compared to earlier;
I think I have some leeway to continue upping the SfW MAP curve to see what happens.
I think I have some leeway to continue upping the SfW MAP curve to see what happens.
Last edited by rwyatt365; 06-09-2019 at 06:30 PM.
#31
I shifted my focus to the post-lift rich dip. I was thinking last night about how SfW and AtW worked and thought I had it all figured out. So, this morning I pulled down the AtW MAP curve, and pulled up the SfW MAP curve for the lowest MAP values thinking that would decrease the WF before the throttle lift and lean out the rich dip. All of my thinking was in vain, because my changes had the opposite effect - WF increased slightly and the AFR dipped even richer than before (damn you EAE!!). Below is a representative event from today's and yesterday's logs.
So, I'm going to invert those changes and see what happens.
So, I'm going to invert those changes and see what happens.
#33
So, after I picked my ego up off the ground, I did this;
...and I got this;
The lean spike is less lean, and the rich dip is less rich (marginally, in some places). I think I can now effectively tweak that with the RPM curve.
Thanks Ted!
#34
Ted - I have been following this with the hope that a golden panacea would present itself. But I haven't seen you do a good write up on this yet.
Would you be so kind as to write a little something on how the heck you got such a beautiful AFR response on tip-in/out? From your graph, it looks like the EAE will put in initial fuel and taper it down intelligently as opposed to just a short burst of fuel.
Would you be so kind as to write a little something on how the heck you got such a beautiful AFR response on tip-in/out? From your graph, it looks like the EAE will put in initial fuel and taper it down intelligently as opposed to just a short burst of fuel.
#36
Tweaking Enginerd
iTrader: (2)
Join Date: Mar 2013
Location: Boulder, CO
Posts: 1,776
Total Cats: 358
Ted - I have been following this with the hope that a golden panacea would present itself. But I haven't seen you do a good write up on this yet.
Would you be so kind as to write a little something on how the heck you got such a beautiful AFR response on tip-in/out? From your graph, it looks like the EAE will put in initial fuel and taper it down intelligently as opposed to just a short burst of fuel.
Would you be so kind as to write a little something on how the heck you got such a beautiful AFR response on tip-in/out? From your graph, it looks like the EAE will put in initial fuel and taper it down intelligently as opposed to just a short burst of fuel.
To this end, it would be a bit foolish of me to publish the secret sauce to one of my more valuable products. I have actually already divulged more "IP" in this thread than I ever intended to provide in the open forums. I know this might be selfish of me, but I do need to protect my own interests. I do contribute to some of the other discussions on topics that I have invested in (e85 cold start would be an example) so I don't hold everything back.
On the bright side of things, assuming I decide to take it to the next level at some point, I have identified you as one of the potential beta testers for some of my more special sauce products. It is however still a work in progress, and my real-work activities prevent me from pursuing the moon-lighting opportunities at this time.
#37
Retired Mech Design Engr
iTrader: (3)
Join Date: Jan 2013
Location: Seneca, SC
Posts: 5,009
Total Cats: 857
Ted... I was wondering if that was your plan, when you mentioned elsewhere that you had an off-line EAE program.
#38
DNM - it might be my imagination, but on Ted's graph of EAE1 you can see that there is a sharp bump followed by a tapering down and even a little bump again. If you only use accel enrichment you just get a sharp/quick bump of fuel.
Ted - I may have to take you up on your offer. I've started writing some codes in MATLAB to bring in logs (for AFR at steady state) and really don't know where to start with EAE.
Ted - I may have to take you up on your offer. I've started writing some codes in MATLAB to bring in logs (for AFR at steady state) and really don't know where to start with EAE.
#39
Ted, I appreciate your "divulging" a snippet of your IP in this thread. "Expert opinion" is always welcome.
DNM : I haven't looked at your MSQ yet, but I will!
Other news...not much. I've been trying to apply my new-found revelations to achieve "Ted-like" AFRs. I'm not there yet, but I'm gaining some ground. I'll come back when I get a breakthrough.
DNM : I haven't looked at your MSQ yet, but I will!
Other news...not much. I've been trying to apply my new-found revelations to achieve "Ted-like" AFRs. I'm not there yet, but I'm gaining some ground. I'll come back when I get a breakthrough.