decel question

[Belgian1979]

That's a very good possibility.

There is an algorith that lets you compensate for wall wetting and such. Haven't tried it, as this is difficult to say the least.

What for ignition timing ? Right now I've set the timing a fair amount of time before the intake valve opens so as to give the fuel time to vaporize. Maybe I should set the time shorter ?

[Rick Finsta]

As RednGold86Z said it's the fuel puddle evaporating. Your best options are either tuning the EAE or leaning out the cells on the VE table that you spend in decel but not under power or in cruise. The problem the latter creates is you won't have enough acceleration enrichment to make up for the lack of fuel in the intake especially with the then too-lean VE. So while it solves a problem for a dragstrip car it wouldn't work for a road/circle track car. Even in a drag car you'd have issues going lean after a hiccup in a run or perhaps even during shift recovery - not good.

Got an .msq we can look at? Or did you cross post this on the MSExtra forums?

RednGold86Z do the FiTech units run X-Tau or Enhanced Acceleration Enrichment (or other wall-wetting compensation) algorithms? If so I'd be interested in how you did the self-learning for that since that is one thing that has not yet been implemented in TunerStudio (the tuning software in the Megasquirt world). It's not hard to bandaid a close tune with an aggressive PID loop for the EGO (O2) correction but it bothers me that so far it has me licked.

[Belgian1979]

cannot upload the msq as the extension is not allowed here. I have not asked about this on msextra.

Anyway, fuel puddling could indeed be a factor. Injector placement on my manifold is as worse as it can get due to space constraints.
Haven't even tried the X-tau or EAE settings. They are complicated.

[RednGold86Z]

As RednGold86Z said it's the fuel puddle evaporating. Your best options are either tuning the EAE or leaning out the cells on the VE table that you spend in decel but not under power or in cruise. The problem the latter creates is you won't have enough acceleration enrichment to make up for the lack of fuel in the intake especially with the then too-lean VE. So while it solves a problem for a dragstrip car it wouldn't work for a road/circle track car. Even in a drag car you'd have issues going lean after a hiccup in a run or perhaps even during shift recovery - not good.

Got an .msq we can look at? Or did you cross post this on the MSExtra forums?

RednGold86Z do the FiTech units run X-Tau or Enhanced Acceleration Enrichment (or other wall-wetting compensation) algorithms? If so I'd be interested in how you did the self-learning for that since that is one thing that has not yet been implemented in TunerStudio (the tuning software in the Megasquirt world). It's not hard to bandaid a close tune with an aggressive PID loop for the EGO (O2) correction but it bothers me that so far it has me licked.

The FiTech transient fueling is a home brew, that matches X-Tau's output, but doesn't use the desired fuel as the input - it just accumulates and decays. X-Tau can oscillate and the result could be a steady output. My way is stable. If X Tau (without trigger thresholds) were to try to do the impossible (i.e. get tons of fuel from the injectors up in the throttle body to the cylinder as fast as the air can move, when the engine is stone cold), it's possible to have crazy fluctuations in the pulsewidth, with the hope that the right amount of fuel gets to the cylinders.

I've made X Tau, and tried it, but it was just too much of a non-intuitive tuning process to grasp easily. My way is more of a wall film builder and reducer (though when compared side by side with X-Tau, the injected PWs are exactly the same when in a stable ratio of x and tau). My way tries to build or decay a wall film over time - without much regard to the calculated flow into the cylinder. Looks exactly the same when the squirts come out, in most situations. The calculation goes both ways - adds fuel when wall film needs to get bigger, and subtracts PW or even turns off the injectors if the film needs to be reduced. During a severe enough situation, the injectors can be shut off even before the throttle closes - the wall film supplies all of the needed fuel (and often more than what's needed). A true X-Tau is also hard to tweak to be asymmetric - if you want accel fuel to have a rich dip, you'd get a lean dip on decel, unless you told it to make the rich dip in the desired AFR. It also never forgets, and a little error in one way can have a large "calculated" wall film growing slowly, but when the throttle is closed and that wall film wasn't actually that big, it can lean way out, because a small error in PW when the PWs are small is actually a very big error. A small error in PW when the PWs are big, is a small error that can be covered by closed loop.

For the throttle body injection stuff (large cold wet manifolds), with the discharge rings with a volume of fuel in the channels to replenish, there are some tweaks to the above stuff that were a bit ad hoc, and patched on top. Port injection doesn't have most of that problem (though the manifold can build a bit of upstream film during an extended WOT). Direct injection has basically none of that.

I've made a learn algorithm for the accel fuel a long time ago, but it's not implemented yet. The user has to tweak accel and tweak cranking fuel to suit their engine. I do use a very fast closed loop that starts as soon as the sensor temperature feedback says it's ready, and somewhat fast learning, and a confidence of learn table to slow it down when it sees a stable settled result for a learn cell. This learning of the steady state fueling is inside the engine management code - and is stored in a simulated EEProm section of memory on power down - it's permanent until changed by the next drive or commanded to be reset. Disconnecting the battery won't affect it.

I've probably said too much, but maybe this can help paint the picture better of what goes on behind the scenes with EFI.

[Rick Finsta]

Thanks for the detailed response.

The X-Tau (Ford?) strategy was utilized in the original MS code but has been replaced with EAE (Enhanced Acceleration Enrichment), which was derived from Toyota's work on the subject if I recall correctly.

Your strategy sounds *similar* to the EAE setup that we use in the MS2/MS3 world, then. It isn't a transient strategy per se, it's on all the time tracking the amount of fuel in the intake and adjusting PW to suit. I'm convinced that we can find an easy way to autotune it if we find the right scatterplot/histogram setup. For now, you basically have to just drive around and go through several iterations changing the different curves. The problem I have run into is duplicating test conditions - do you increase the "added to walls" coefficient or decrease the "sucked from walls" coefficient? To know you have to try it both ways under the same conditions, then throw in another variable and see if it behaves as it should. So minimum four tests under the exact same conditions... it's hard enough to duplicate a delta TPS now I've got to try and duplicate a delta MAP or Delta RPM in the exact same load conditions while not hitting a whitetail dear or bicyclist? If I had a dyno and hundreds of hours I'd bet I could get it done but on the street I think the key will be a lot of data (scatterplots/histograms).

I understand you have a need to protect your IP so I won't pry further.

[Firedome8]

Hi new here looking to learn, on decell do most EFI systems turn off fuel ( DFCO ) or is that not common and needs addressed in VE table. Could that maybe solve your popping issue?

[mk e]

The M1 motec is an accumulate and decay as well. I kind of copied that system (but added a little more flexability) into the enginelab ECU I have now...it looks good on a simulator but I haven't tried it on an engine yet. That code is open and available to anyone who wants it.


Back to the real thread.

Popping means there's fuel and air in a combustably mixture in the pipe and an ignition source too. You need to get rid of 1 of them to cure it. I've had the best luck with advancing the timing....45-50 degrees is what I generally do under hard decel so the burn is complete after the exhaust valve opens. Then air leaks....I had one driving me crazy that was mostly the cat test pipe bolts had come loose so it was sucking in air but not loose enough to make any noise.

[Belgian1979]

I run 44° in the decel area, so that's already close.

I'm not sure where it's exploding in the pipes. Headers are closed to the heads, but further downstream, there are slip fit connections with slits in the pipes and pipe clamps. These do not totally seal up.

I have a feeling that redngoldz got it correct. I think it's massive build up of fuel on the wall that gets drawn into the cyls during decel. I know the engine draws a massive amount of vacuum under heavy decel. Idle is at 68 kPa, decel is low 20's.
FWIW, immediatly after coming of the decel and opening throttle again it goes really lean and it's hard to get enough fuel in there to get a decent fuel ratio.
I have been struggling with the AE all the way from the beginning. The first thing I noticed was the AE not coming in only on TPS at low TP openings. I used mapAE to compensate for it.

I have been looking at the EAE feature, but I do not see a good way to tune it and the manual doesn't give much explanation either. That stuff is really complicated.

Anyway, I already destroyed an old exhaust muffler with a serious pop at high rpm. Granted it was an old muffler. I now have a new system with 3" pipes, and a straight through muffler, but the explosions are enough to having to let the engine really go down to very low rpm before reengaging throttle. Annoying as hell.

[Rick Finsta]

I just put new exhaust on my junk and had to weld up all those little slits in the slip-fit joints. I get why they are there (for clamped applications) but what a PITA with the TIG torch!

I got your PM and replied, I'll take a look at the tune and see if anything pops out (no pun intended).

[Belgian1979]

Thanks.

[MadBill]

One way to narrow down the source of the 'pop fuel' would be with a very long downhill decel in a relatively high gear, enough time that the intake system will be totally dry. If the popping stops before long it could be wall film fuel but if it continues there has to be another source.

[digger]

I get light pops on over run so I used fuel cut and it works 100% very quickly. I also get very light popping at very low throttle (~2%) under higher rpm condition eg coasting down hill is only time scenario happens presumably due to higher vacuum and mismatch in runner flow somewhere in the system even though throttles are sync'd very well . I haven't 100% ruled out any leaks yet so I'm interested in this as a similar reason for it.

[Rick Finsta]

One thing I notice about your VE table is it looks a bit lumpy in areas where you would tend to see transients, that is to say it looks like the potential is there that the VE table is adding fuel where the TPS/MAP based AE should be. Do you rely solely on VEAL or are you smoothing things out between tuning sessions? Do you have MegaLogViewer HD? I've got a theory on how to figure out when VE is being used in lieu of AE using the histograms but I have to get the car back out for a bunch of full throttle blasts (which isn't easy to do when it keeps raining and/or hailing nightly...).

Otherwise if you have a nice big log file with a bunch of throttle stabs in the 4000RPM+ range I might be able to test my idea if you'd like to email me. Real testing is going to require I actually induce that error on purpose, though.

[mk e]

I run 44° in the decel area, so that's already close.

Well, I'd try more. I've seen as much as 60* used but never had to go that high myself.

I'm not sure where it's exploding in the pipes. Headers are closed to the heads, but further downstream, there are slip fit connections with slits in the pipes and pipe clamps. These do not totally seal up.

That is almost certainly why a big part of why you have the problem....RTV those slip joints. It may or may not last but it will help pinpoint the issue.

Bill's idea to do a downhill runn was excellent too so you'll know more about where the fuel is coming from.

FWIW, immediatly after coming of the decel and opening throttle again it goes really lean and it's hard to get enough fuel in there to get a decent fuel ratio.
I have been struggling with the AE all the way from the beginning. The first thing I noticed was the AE not coming in only on TPS at low TP openings. I used mapAE to compensate for it.

I have been looking at the EAE feature, but I do not see a good way to tune it and the manual doesn't give much explanation either. That stuff is really complicated.

I used to have a bd habbit of running the mixture rich down near idle to help it come cleanly of idle wihtouth stumping.....well...I still have the habbit to be honest because it usually helps a lot.

You can do the same thing everywhere but that hurts your fuel milage everywhere not just at idle.

I've never tuned the MS AE or EAE so I won't comment beyond that idea it always to dump in lots of fuel at throttle open and taper back as things stablilize so the tables are always about how much extra and for how long. You tune it by doing throttle stabs while you log, but trust the way it feels over what the O2 says....I usually add until it goes to 12.0-13.0 and stay in that range through the accel but current OEM thinking is the mixture should be stable meaning it should stay around 14.7 through the accel.

Fuel kill on decel is a fuel saving mode that usually makes the tip-in leaning worse and there is almost always a hesitation or stumble whne the fuel comes on so I don't use that feature. But it the popping is more a concern to you than a hesitation you could try going into the VE table and set the VE to 0 when MAP is below say 60kPa (if 68kPa is the lowest real number), using the VE table should give you more control of the fuel kill......just be sure to set another row at say 61kPa that has real VE numbers in it becasue the ECU interpolates and you want the fule to be on then off not partly on. This might work better than a TPS based cut since for you tiny changes in TPS make a big change in MAP near throttle closed.

[Belgian1979]

One thing I notice about your VE table is it looks a bit lumpy in areas where you would tend to see transients, that is to say it looks like the potential is there that the VE table is adding fuel where the TPS/MAP based AE should be. Do you rely solely on VEAL or are you smoothing things out between tuning sessions? Do you have MegaLogViewer HD? I've got a theory on how to figure out when VE is being used in lieu of AE using the histograms but I have to get the car back out for a bunch of full throttle blasts (which isn't easy to do when it keeps raining and/or hailing nightly...).

Otherwise if you have a nice big log file with a bunch of throttle stabs in the 4000RPM+ range I might be able to test my idea if you'd like to email me. Real testing is going to require I actually induce that error on purpose, though.

I will send you the log file.

The msq you have there could be a bit bumpy as it is indeed with a VEAL correction after a log. I usually smooth things out. I haven't used the msq currently in the logs as I've done some things in an effort to mitigate the popping.

I don't use MLV HD.