Page 3 of 10

Posted: Wed May 07, 2008 1:47 pm
by mops
But here's catch that I discovered....
if you are starting with added to walls to high and sucked from walls too low (essentially ecu thinks the puddle is bigger than it actually is), increasing 'added to walls' does not have desired effect, because ecu thing there's whole lot of fuel coming from the walls there it actually isnt. This is actually contradicts tuning tips 'for more AE incerase fuel added to walls'

Posted: Wed May 07, 2008 2:09 pm
by Peter Florance
Or is this a better test?
Here I moved the throttle around and tried to keep it centered around 3000 rpm

Thanks!

Posted: Wed May 07, 2008 2:12 pm
by muythaibxr
mops wrote:But here's catch that I discovered....
if you are starting with added to walls to high and sucked from walls too low (essentially ecu thinks the puddle is bigger than it actually is), increasing 'added to walls' does not have desired effect, because ecu thing there's whole lot of fuel coming from the walls there it actually isnt. This is actually contradicts tuning tips 'for more AE incerase fuel added to walls'
Correct, the algorithm will do many different "odd" types of things if added to and sucked from walls values are too far apart from each other.

Ken

Posted: Wed May 07, 2008 2:16 pm
by muythaibxr
Peter, that's better, but what I'd do, is press the throttle down, leave it there until EAE recovers to 100%, and AFR recovers to its flat state, then let off, but only do the whole thing once, so you can go from a flat AFR, to the throttle press transient, then flat again, then lift throttle transient.

After doing that, you can easily see what the AFR is doing in response to each individual throttle movement with no extraneous data to clutter the AFR reading.

The rate that you pressed the throttle at was pretty good though.

Also, my recommendation would be to tune that area until the AFR is as flat as possible (the "around 3000 RPM area") then move to tuning the RPM curves once you feel comfortable that the AFR is as flat as you can get it.

I cannot emphasize enough though that your VE table must be correct at steady-state for EAE to properly do its job.

Ken

Posted: Wed May 07, 2008 2:19 pm
by Peter Florance
muythaibxr wrote:Peter, that's better, but what I'd do, is press the throttle down, leave it there until EAE recovers to 100%, and AFR recovers to its flat state, then let off, but only do the whole thing once, so you can go from a flat AFR, to the throttle press transient, then flat again, then lift throttle transient.

After doing that, you can easily see what the AFR is doing in response to each individual throttle movement with no extraneous data to clutter the AFR reading.

The rate that you pressed the throttle at was pretty good though.

Ken
Ok, got it.

Now, do I have to put it flat to the floor or can I move it 1/2 to 2/3 just to get a feel for how the adjustments are going to work?

Thanks Ken!

Posted: Wed May 07, 2008 8:07 pm
by Peter Florance
Here's step-wise movement.
I think it looks ok except for the lag?

Posted: Thu May 08, 2008 8:01 am
by muythaibxr
Yeah, there was a slight lean spot on the press, and a little bit of richening on that really slow lift.

Assuming your VE table is correct, I'd bump up the added to walls slightly to begin with.

Re:

Posted: Fri May 09, 2008 6:22 am
by Peter Florance
muythaibxr wrote:Yeah, there was a slight lean spot on the press, and a little bit of richening on that really slow lift.

Assuming your VE table is correct, I'd bump up the added to walls slightly to begin with.
I added about 5% to added to walls and the richness was still there, but smoother?

Just to see what it would do, I increased sucked from walls by about 5% (to tell EAE more fuel is sucked from walls) and it seems to have helped.

It actually felt pretty good, better than my-well tuned RPM based AE on MS1. And I haven't done any rpm corrections; this is only at 3000 rpm or so.

Will post screenshot and datalog when attachments are working

thanks!!!

Re: Quick EAE question...

Posted: Fri May 09, 2008 10:19 am
by Peter Florance
eae_test_05_09_08.png
Here's this morning's test.

I drove it again at lunch and it feels really really good.


Note I didn't change the default rpm corrections below 2500 rpm and only did my tuning in the 2700-3300 rpm.
Having said that, the default rpm correction curves seem to track pretty nicely off and back on idle (900 rpm)

comments?

Re: Quick EAE question...

Posted: Fri May 09, 2008 10:35 am
by muythaibxr
Yeah, I'm assuming that the lean spike when you went to closed throttle was on purpose, overrun fuel cut or something?

It looks like going into that state, there was a slight rich spot, so you could probably get away with slightly increasing sucked from walls there.

Re: Quick EAE question...

Posted: Fri May 09, 2008 10:43 am
by Peter Florance
muythaibxr wrote:Yeah, I'm assuming that the lean spike when you went to closed throttle was on purpose, overrun fuel cut or something?
I'm not sure. I want overrun fuel cut once I'm done tuning. But it wasn't my intention to have it yet.
I'll check my msq
muythaibxr wrote: It looks like going into that state, there was a slight rich spot, so you could probably get away with slightly increasing sucked from walls there.
Do you mean at the low end of the table? 25kpa or so?

Big question; am I working in the right direction at least? Getting the polarity of the logic correct, seems key to me. :)

Thanks!

Re: Quick EAE question...

Posted: Fri May 09, 2008 11:01 am
by muythaibxr
Yeah, that doesn't look too bad... The signal from your wideband seems a bit jittery, but I'm not convinced that's EAE tuning. I think it's just the way that wideband works.

I think you're going in the right direction as well... The 25kPa part is what I was talking about for a slight rich spot before going really lean. I think the "really lean" part was probably just what's in your VE table, and the slight rich spot could be tuned out with EAE.

Ken

Re: Quick EAE question...

Posted: Fri May 09, 2008 11:23 am
by Peter Florance
muythaibxr wrote:Yeah, that doesn't look too bad... The signal from your wideband seems a bit jittery, but I'm not convinced that's EAE tuning. I think it's just the way that wideband works.

I think you're going in the right direction as well... The 25kPa part is what I was talking about for a slight rich spot before going really lean. I think the "really lean" part was probably just what's in your VE table, and the slight rich spot could be tuned out with EAE.

Ken
Thanks Ken
I'll try to increase the low-end suck-from-walls coefficient

The WB is a little noisy. Would it hurt to enable some smoothing in the LM1 (if I can still find the friggin serial cable for it)?

Re: Quick EAE question...

Posted: Fri May 09, 2008 11:42 am
by muythaibxr
You could enable smoothing or lower your lag factor for Lambda.

Re: Quick EAE question...

Posted: Fri May 09, 2008 2:18 pm
by milesinfront
Hey Peter, what engine is this? How many squirts?

Your first datalog looks 100x better than my last, after many hours of work! I'm very jealous... It seems that the standard settings are much more compatible with your engine.

Re: Quick EAE question...

Posted: Fri May 09, 2008 2:29 pm
by Peter Florance
milesinfront wrote:Hey Peter, what engine is this? How many squirts?

Your first datalog looks 100x better than my last, after many hours of work! I'm very jealous... It seems that the standard settings are much more compatible with your engine.

Amazing, someone jealous of anything associated with me. :D

M30b34 lump

It has been VE analyzed to death; I usually only get 1 or 2 cell changes.

Re: Quick EAE question...

Posted: Fri May 09, 2008 2:29 pm
by muythaibxr
Peter posted an msq, you could check that for a lot of the info you need.

Ken

Re: Quick EAE question...

Posted: Fri May 09, 2008 2:37 pm
by milesinfront
muythaibxr wrote:Peter posted an msq, you could check that for a lot of the info you need.
Oh yeah... Doh!

6 alt vs 2 alt... I might try increasing my squirts!

Dam you all for sucking me back into EAE!!!! :twisted:

Re: Quick EAE question...

Posted: Sat May 10, 2008 4:01 am
by Peter Florance
milesinfront wrote:
Dam you all for sucking me back into EAE!!!! :twisted:
You're welcome. That's what we're here for.

Just a thought: take a look at my rpm correction curves, set to for inital tuning. They seem to work ok to allow me to concentrate on tuning the midrange EAE.
I suspect once you go to 6 sqrts and fine tune your VE, it should feel pretty good with default Adheare to and Suck from settings . I haven't changed that far from Ken's values. I suspect it's because they must track some basic physical laws that don't seem to vary as much as one would suspect. Also as Ken as suggested, a little change goes a long way.

I freaks me out to glance down at my LM1 wideband. It moves so little as I drive, I think it's broken. :shock:

This is really great software, Ken.

Re: Quick EAE question...

Posted: Sat May 10, 2008 4:10 am
by milesinfront
Yes I've read about it, but never experienced it first hand:- Upping the number of squirts dramatically reduces the dependence on AE. I had to seriously back of my AE settings...

Down side to 6 squirts is my LM1 readings really get jumpy. I'm not running huge injectors, but they seems to be big enough to make tuning over-run semi impossible. Changing the VE table by one unit is the difference between 17:1 and 12:1 AFR... I'm sure the fact that my O2 sensor is sniffing only one cylinder isn't helping things much... It could also possibly be noise due to the extra injector pulses...

How do I get the initial EAE settings back without re-flashing MS?