WOT Lean Transient Condition

Anything about Ford EEC tuning. TwEECer and Moates questions dominate, but there's some SCT and OBD-II knowledge too.

Moderators: cgrey8, EDS50, Jon 94GT, 2Shaker

Post Reply
birdthunder
Gear Head
Posts: 3
Joined: Mon Jul 02, 2018 10:53 pm

WOT Lean Transient Condition

Post by birdthunder » Sun May 19, 2019 5:30 pm

I'm posting this to EECTuning and EFIDynoTuning, so FYI in case you read both and get confused. I'll post the solution to both if I figure it out.

I really don't know what to do other than info dump everything I know, because it *could* be related. I realize this is a large post, but I don't think I have a typical issue (I'd be plenty happy to be proven wrong with a simple solution).


1996 Ford Thunderbird
4.6 2v (PI motor, heads and bottom end. NOT a NPI bottom end with PI heads)
Ford 39lb Cobra Injectors (I know there's kinda a debate about if the 4v spray pattern for these causes 2v tuning issues. Surely some Deka's wouldn't fix my problems..)
SVO supercharger; Water/meth spray for cooling
4r70w (from a 2003 crown vic)
EEC-V with CDAN4 strategy (custom tune obviously)
3.27 open diff (I know, I have 3.73 with trak lok sitting around to be rebuilt and installed)
90mm LMAF


So, long story short, my transient fuel is just wacked out. Or maybe its my anticipation / manifold modelling. Or maybe I have a simpler issue and am over-complicating things. I don't know, and need help. I've dug through the CDAN2.txt doc and other tuning resources and still haven't figured it out.

Part of it is how different the car is from stock. I don't have a good baseline to work from or compare to for this setup.

Things aren't THAT bad. The car is plenty drivable. It swings rich when blipping the throttle from closed, but otherwise fuel is mostly fine in CL cruising, with a few spikes here and there from fast throttle closing / opening.




My problem is at WOT (or near), and happens sometimes (maybe 50% overall?), not always.


The typical way for me to 'cause' it is to gently accel to 5-10mph from a stop, and floor it. Lambse will be 0.80 and the WB will read from .83-.86 until maybe a little bit after the car shifts into 2nd. For sure, by the time 3rd gear hits, the WB will match the demanded fuel.

Another mostly consistent way is to be cruising a '2nd gear speed' in 3rd gear. Floor it, and after the downshift, the WB will read .83-.86 when .80 is demanded.

A lot of times it follows that general pattern. If I punch the throttle, it'll be lean and taper towards the demanded lambse over the duration of about a gear. If I roll into the throttle smoothly over a few seconds, it doesn't really happen.

Maybe related, during WOT upshifts, it leans out a little bit as well. I'm not sure how to change this. FN371 is for engagements (shifting from neutral into drive), not for shifts right?


This doesn't happen always though. On the same tune, same gas, same day, I might datalog it and it peg the WB within 1-2% for the whole pull.
So, its lean, sometimes.


I don't think its the MAF curve. I don't think the top of my curve is perfect (hard to tell with all this) but this would cause a consistent/reproducible error, right? Not sometimes nearly perfect and sometimes 6% off.


It *looks* related to what gear I'm in. I know that effects a bunch of other things (like LOAD, how fast its changing, etc.) so that makes me either injector params, transient fuel, or air modelling is messed up. Great...


Injector parameters are probably fine. Injectors slopes are linear at WOT, so I would have a consistent issue at a certain load (or MAFv/RPM pair), right? I mentioned I have the 39lb cobra injectors (that aren't the right spray pattern for 2v I guess).


I don't think I can change FN1321 to fix this because sometimes load is barely changing while its going lean. Load changes before this condition (from cruise to flooring it) so I could possibly change the accel time constant to make transient fuel stay in longer? In the stock tune the accel time constant scaled so that its in the range of a few tenths of a second. I would need a time constant which stays in for like 2 seconds, ~10x as long? I'm not THAT far from stock, am I?


Reading through all the CDAN2.txt stuff, it makes me wonder if my manifold filling / anticipation stuff is messed up. This is MY understanding of what manifold filtering and anticipation roughly are. Not necessarily correct, please let me know if its not.

From the MAF (an upstream measure of flow) and some horrific math, the EEC figures how much air is in each cylinder. It's inherently involved during transient conditions. For example:
- if MAF flow is constant, then mass flow through manifold and thus cylinders is constant. Mass/time at MAF = mass/time at cylinders.
- if MAF flow is changing, then the MAF reading at that instant won't accurately describe the mass of air in the cylinders.
Basically, when you floor it, the manifold is suddenly exposed to higher pressure (from the inlet) and starts filling -> the MAF feels more airflow. But the cylinders don't start getting more air at the same rate; thats dependent on how fast the manifold fills to equilibrium. So you have to model manifold filling to determine how much air is in the cylinders at that instant, given the current upstream flow and knowledge of the previous flow (which should be in the manifold / cylinders by now).
When you decel, MAF feels flow decrease first, but cylinders are still using air from manifold (since its over equilibrium pressure), so MAF feels decel before cylinders actually get less air. Again, you have to model the manifold to deal with this.
...Is roughly what I think manifold modelling is referring to.

In CDAN4, this is done with FN052M/N and a few other parameters. I don't honestly understand these to well; they represent some physical characteristics of airflow or pressure through the MAF and manifold. I'm focused in one this manifold modelling stuff so much because of how different my car is from a stock 1996 TBird. Seems like a supercharger would have dramatically different filling properties, if for no reason other than shape / volume / recirculation / etc.
With this being a positive displacement SC, does it affect things more? Talking out of my ass, but with air recirculating in the SC, would the manifold 'stay' at (nearly) equilibrium pressure and better correlate the instantaneous MAF signal to cylinder air mass? Could my manifold modelling be 'buffering' airflow changes at MAF longer than it should? How do I describe that in FN052M/N?

Either way, my issues doesn't really happen a set RPM because it doesn't always happen. These might have a more complex effect than 'constant at an RPM' but I don't know what it is / how to tune it.

I'm using FN052M/N values I got from a spreadsheet that MitchB posted. They're supposed to be from the a stock 2004 4.6, so I thought they fit my car better than the values from the stock NPI 4.6.
I made this change a while ago, and if I recall correctly, tried both stock values and the MitchB ones. Neither fixed my lean at WOT issues. Might be placebo, but in general the custom ones made stuff more accurate and consistent, so I thought it was a step in the right direction.


Anticipation is a separate function, but tangled up in manifold modelling. Mine is on; could that be my issue? I've read that you have to rework a bunch of transient fuel stuff to disable this, so I haven't tried it (and gone through the process of getting transient close again to test if WOT fuel is consistent then).


Manifold volume is set at 6. Short version is that 6 is whatever I left it on trying to find what worked best.
I made a post at viewtopic.php?t=22633 explaining that, as far as I could tell from math, you SHOULDN'T scale this to your overall scaling % (mine is 80%). By scaling a different function, you actually scale air modelling correctly (the computed filtering constants AIR_FK, air_k_base, etc. will be unchanged), whereas using manifold volume introduces error. If I'm not full of shit, then I did it the 'correct' way (so my manifold volume is effectively a 'unscaled' value)... My point is that my scaling should be completely unrelated to my issue, especially since.


My inlet temperatures get pretty high. Like 200-230+ easy, heatsoaked and driving spiritedly. Normally I would spray water/meth, and not let temps get this high. I'm trying to dial in fuel though, so I haven't been spraying anything (I need to switch to just water so I can spray and still dial fuel). It got me thinking though...

There's parameters like FN1052 which have to do with...manifold modelling. Some kind of adjustment factor based on, in part, inlet temp. I've wondered if any stock tunes have reasonable values in the table for ACT values of 200+? You would never see it on a stock car. I've tried my stock table and one from a PI engine'd car (I think) and didn't really notice a difference. I could try dummying these out so that ACT doesn't have as much of an affect, but that's the point of the table. I'm also not sure how this affects fueling output: from the math, it seems like it affects manifold filling, so maybe it could cause a 'tapering error' like I'm seeing, based on changing load / air flow?


The other thing I've considered is that, due to inlet heat, my WOT spark is low. Like 8.5-10 degrees, maybe less at WOT (when ACT very hot). I don't think it's related, but I guess it can't hurt to mention.
I pulled some spark from the top of my table (until leaning is fixed) so I'm actually getting more like 3.5-6 degrees at (very hot) WOT right now. This happened after the issue existed obviously, so it makes me think low spark is completely unrelated.
Still, should there be a bottom limit that I allow spark to drop to? Normally wouldn't matter cause water/meth spray I guess.




I can't think of many sanity checks to do, but I've made sure I'm not clipping anything obvious like MAF or a FN1036A/B. I've taken so many datalogs of this with different PIDS and can't find anything that looks strongly correlated.

I have a fuel pressure gauge in the car, but I wouldn't be able to notice a few % variation while I'm driving the car at WOT (not sure if the gauge would even show it). Should I get a friend to ride with me and record the gauge to make sure? Pump is 255lph I installed some number of years ago. I have fuel pressure drop outs if I'm low on gas and swing around tight turns at WOT, but have otherwise have never noticed an issue. If I have over half a tank or am driving in a straight line, there's never a drop out. My WOT leaning thing happens in straight lines, with a full tank of gas.


Does anyone have any example tunes of 4.6PI engine cars, or even better, a SVO supercharged one? I know I can't just expect to copy and paste values, but having anything to look at / build from would surely help. I wouldn't mind seeing FN1052, FN052M/N, even just the spark and torque tables from a dialed-in car with similar engine build.

-----------------------------------------------------

Unrelated side question:
I recently got a X-pipe installed. It's about half a car length from the engine. I was prepared to switch to using only one bank's O2 sensor after reading posts online, but I haven't noticed any closed loop issues from reversion/cross scavenging/whatever while still running both banks. Are the issues normally present when the X-pipe is closer to the engine? I would expect wacky fuel trims (short and long?) to start showing up at some RPM range in closed loop. Haven't seen any (new) weird stuff since the exhaust was changed.
My X-pipe is 100% unrelated to the upper portion of this post. I had all of that going on even when I had true duals.



I can post my tune and upload a couple datalogs with examples of it happening later.
Thanks for any help guys, and let me know if I can provide any additional information.

birdthunder
Gear Head
Posts: 3
Joined: Mon Jul 02, 2018 10:53 pm

Re: WOT Lean Transient Condition

Post by birdthunder » Sun May 19, 2019 5:55 pm

I can post more logs later, but I wanted to at least get one uploaded.

"Analog In 7 - Red FireWire|3|224|33|0" is the wide band (lambda).
The tune is CDAN4 strategy.

Noteworthy timestamps of the 'thing' happening:

174 seconds (02:54) WOT, is lean and then tapers off to being a little rich after shift
1429 seconds (23:49) WOT and it tips in rich, goes lean, tapers off after shift (although still a bit rich after)
2083 seconds (34:43) WOT and it leans a bit and tapers off a little
2182 seconds (36:22) WOT and it leans, tapers and goes rich after shift

None of these really show the independent/possibly unrelated leaning during a WOT shift. Which I'm still not sure is independent/unrelated.
Attachments
log.zip
(2.58 MiB) Downloaded 6 times
459C.bin
(112 KiB) Downloaded 5 times

Mitchb
Regular
Posts: 102
Joined: Sat Aug 16, 2008 4:02 am

Re: WOT Lean Transient Condition

Post by Mitchb » Sun May 19, 2019 11:19 pm

I’ll take a look at this and your datalog soon.
The parameters you talk about and others have no meaning really until you graph them so you have a visual representation, make changes and then datalog what the affect was. I am very familiar with your catch code. I may ask a lot of questions.

Mitch

Post Reply

Who is online

Users browsing this forum: No registered users and 19 guests