Braineack

iTrader: (62)

you'll need to put AE back in to deal with quick throttle stabs.

I think Ken reworked the EAE code a bit in the newest alpha.

y8s

iTrader: (8)

reworked AE, not EAE.

Braineack

iTrader: (62)

ah

ScottFW

iTrader: (4)

I spent some time last month redoing my EAE. When I first started messing with it I had left normal AE on, so I got to do it over again. It's now dialed in pretty well, probably just going to do one more session of logging & adjusting and then add back some normal AE for the quick stabs, but it's pretty smooth as is.

The best way I found to tune it is on a highway with a decent uphill stretch so rpm can remain fairly stable, then very slowly add in throttle until WOT. Then go back and look at your datalog in MLV, plotting WBO2 and target AFR on the same graph with the same scales. Where you see your actual AFR starting to deviate from the target (assuming a very well tuned VE table in the first place) that's the kPa at which you need to adjust EAE added-to-walls. To tune sucked-from-walls I basically do the opposite- start at WOT and slowly decrease throttle, check the datalog for the kPa where actual AFR starts to deviate from target. A few iterations of datalogging slow throttle sweeps from zero-to-WOT-to-zero got it dialed in quite well.

If it goes rich on accel, decrease added-to-walls.
If it goes lean on accel, increase added-to-walls.
If it goes rich on decel, increase sucked-from-walls.
If it goes lean on decel, decrease sucked-from-walls.

miatauser884

iTrader: (11)

Any rough estmate on % added/sucked from walls per .1 afr? I know all setups will be different, but there has to be a ball park estimate. Im going to try the logs. Right now I do it while driving. Trial and error

ScottFW

iTrader: (4)

You're asking how much to adjust based on how far off target you are? Not sure but if I was "kinda far off" I adjusted that point of the curve by .2, or by .1 if it was "a little off." I never adjusted more than .2 at a time; I'd just do another iteration of logging & adjusting.

I found it basically impossible to simultaneously keep one eye on my wideband gauge, one eye on the EAE gauge to see what it was doing, one eye on engine load to know what kPa range to adjust, one eye on the laptop to make the adjustments, and one eye on the road. EAE is damn near impossible to accurately tune on the fly, at least for me.

ScottFW

iTrader: (4)

When I typed that it had been a while since I looked at EAE. I did some more tuning today and realized I should clarify the above. It's only the sucked-from-walls that I adjust a tenth or two at a time. The adhere-to-walls I will adjust anywhere from 1 to 3 full points at a time depending on how far off it is. Whole numbers for adhere-to-walls. This is probably obvious to anyone who has messed with tuning EAE. Still, these tuning threads should be as clear as possible.

I still have not added back any normal AE for the quick stabs, but the EAE by itself is reacting very well to what I would consider to be fairly fast throttle movements. I'm gonna see how close I can get it without any normal AE, but at this point it looks like it will only be needed for "Tourette's fast" throttle blips.

miatauser884

iTrader: (11)

Would you mind posting your tables? I think I have tuned as well as I can with the logs. I believe the delay in afr reading is now inhibiting my fine tuning. It's a lot better than it was. I believe that I am ready to start tuning the rpm table. What values are you using right off idle? What seems to work with "feel" doesn't seem to jive with my log. Again, I think it is the afr delay between throttle pressed and gauge/ log value at a particular kpa.

The MSextra documentation suggests that the max and min values should stay very close to default, but the values in between will change significantly.

ScottFW

iTrader: (4)

I'm still using 100 straight across the board for rpm and CLT correction tables. But I only log & tune on a warmed up engine around 3000-3500 rpm. Being consistent now will make it easier to not screw everything up when I go to adjust those curves later.

Not sure how different the EAE curves will be on a 1.6 versus a NA 1.8 versus the better flowing 99 head, but I'll post mine for ***** & giggles. You can see I'm still a bit lean in some areas and rich in others, still a work in progress but it's not ridiculously bad. And obviously I haven't tuned above 100 kPa yet, but I have reserved a bit of each table for when I haz teh boostz.







You can see that the second point from the left there (26 kPa, 1.7 in the table) doesn't look like it fits the rest of the curve. Bringing that point up results in my AFRs being lean in that range. Could be an issue with interplay between sucked-from-walls and added-to-walls, could indicate an issue with my VE table tuning, or maybe that's just how it has to be, I dunno.

miatauser884

iTrader: (11)

Thanks, I'll post mine up later. Definitely a difference in values, but the curve is somewhat similar. Now that my 3000rpm curve is getting very close, I'm going to go out and start logging at other rpms. I've noticed that off idle has to be done by feel. I tried a quick adjustment of the rpm curve yesterday based of some of my data, but I reverted back to no rpm correction.

miatauser884

iTrader: (11)

Wondering what kind of values you guys are seeing for AE in MS3 when you have a well tuned EAE table? I was tuning this a little while free revving the engine. Removing the stumble in the 250 r/s area required a PW% of ~50. Seemed high, but I'm not sure how this value relates to ms, which was 1.x when using diypnp.

miatauser884

iTrader: (11)

Here are my EAE and AE data now that everything is driving smooth. It's always in a process of fine tuning, but the car drives very nicely. I was able to tune most of this from logs by taking the ratio of the target afr and the actual afr, then adjusting the tables by the calculated factor. This seems to get close, then fine tuning is done while driving.

I repeat the process to see if everything moves in the right direction. I actually find it more difficult to adjust AE than EAE now. There seem slike there is a wide range of AE values that "feel" fine. Where as EAE values have a very narrow window that works EAE is also really eay to adjust by watching the wideband. AE not so much because there always seems to be a lean spike after the throttle opens, followed by a small rich spike. If I try to tune out the small rich spike, then I can "feel" that the car isn't as smooth.

y8s

iTrader: (8)

A word of warning about this method... it may work ok for "Added To Walls" because that is a percentage of injected pulsewidth, but it probably wont work very well for "Sucked From Walls" since that is a percentage of "WallFuel" which is somewhat unrelated to injected pulsewidth.

Also, the EAE calculations are based on actual injected pulsewidth as I mentioned above. That means that when you start adding regular AE back in, it will become a percentage of that value, not the normal load/rpm pulsewidth.

In other words, at constant speed and load, the EAE contribution will be smaller than when tpsdot triggers enrichment and you are at that same load/rpm point.

miatauser884

iTrader: (11)

So EAE becomes a percentage of (PW + AE) once AE is turned on? Does that mean that once the EAE curves are dialed in, then they need to be scaled to effectively scale out the AE?

That would explain why SFW never seemed to get really spot on with my method. It always came down to watching the AFR gauge in the car. The log method only gets close. It still comes down to watching the AFR gauge and adjusting the graph.

What do you see as a typical AE percent on your table. MS or PW% is fine.

y8s

iTrader: (8)

honestly I have never had good luck with EAE for even mildly fast throttle movements. I'm talking anything over 10-20 tpsdot is too fast for EAE. BUT! With the new tpsdot filter and AE that is in alpha 17, I think that doesn't matter anymore.

on the new AE, I'm using low to mid single digits for sub 200 tpsdot. like 3-5% accel pump AE.