EAE tuning method? Custom gauge? Scatter plot?

[spittybug]

If I understand the concept of the enhanced acceleration enrichment, the purpose of the "stuck to walls" and "sucked from walls" parameters is to add or subtract fuel based on what has been commanded by the VE table versus what is shown to have actually been combusted (by the 02 sensor). Is that basically right? In a perfect world, if the VE table was spot on for the right amount of fuel at each load/RPM condition, the transition from one to another would be supplemented or reduced by an amount of fuel that is stuck or sucked, right? Would over richening it a little bit would be akin to selectively enrichening the transition for power versus economy?

So, if you had your VE table dialed in for your advance table and AFR targets by having run VE analyze with gentle driving, and were confident that it was as good as you were going to get it, the difference between your true AFR values and your target AFR values are what needs to minimized.

I've read the documentation on how to achieve this and it is naturally a long, trial and error process. Putting the box on the stim is less straightforward once you've already jumpered it for VR input, so I've been thinking about what kind of in-car tuning could be done without tons of wasted gas. It seems to me that creating a custom gauge for TS would help. If you had concentric AFR gauges, one for target and one for actual, this would make it pretty apparent pretty quickly when the two were off, wouldn't it? Or how about a FIDLE output to a buzzer to sound under the condition of AFRtarget<>AFR actual +/- some %. An audible alert so that you could focus on the RPM range you are in and get a feel for the need to modify the EAE curves as a result.

Is there perhaps the means to analyze a log file, calculate the AFRtarget minus AFR actual (is this a variable that already exists perhaps?) and plot it against RPMdot (increasing or decreasing RPM) with a Z axis of actual RPM? The new scatter graphing in MLV would appear to be a great tool for this if such a variable existed. Wouldn't that give really good feedback to show how the AFR is different from target under acceleration or deceleration? If my initial assumption about this being the key to EAE is correct, I think this would help. I wish I knew more about coding so I could answer my own question here.............. Given that data, the next step would be to translate that into how much extra/less pulsewidth for injection. A curve of delta AFR vs. pulsewidth at given RPM would be helpful.... sorry, thinking out loud.

Ideas?

[muythaibxr]

If I understand the concept of the enhanced acceleration enrichment, the purpose of the "stuck to walls" and "sucked from walls" parameters is to add or subtract fuel based on what has been commanded by the VE table versus what is shown to have actually been combusted (by the 02 sensor). Is that basically right? In a perfect world, if the VE table was spot on for the right amount of fuel at each load/RPM condition, the transition from one to another would be supplemented or reduced by an amount of fuel that is stuck or sucked, right? Would over richening it a little bit would be akin to selectively enrichening the transition for power versus economy?

It's not to add or subtract fuel. It's to tell the ECU how much fuel is getting pulled from the walls by each intake event or added to the walls for each squirt.

So, if you had your VE table dialed in for your advance table and AFR targets by having run VE analyze with gentle driving, and were confident that it was as good as you were going to get it, the difference between your true AFR values and your target AFR values are what needs to minimized.

During transients, yes.

Is there perhaps the means to analyze a log file, calculate the AFRtarget minus AFR actual (is this a variable that already exists perhaps?)

You can do this in the log viewer.

Ken

[spittybug]

Thanks for the really quick response!

What is the purpose of telling the ECU how much fuel is getting pulled or added? What does the ECU do with that information?

Is there a variable that already exists for AFRtarget minus AFR actual? You say that this can be done in the log viewer, but I'm struggling without a definition of all of the variables. Does such a list of definitions exist?

Thanks.

[muythaibxr]

Thanks for the really quick response!

What is the purpose of telling the ECU how much fuel is getting pulled or added? What does the ECU do with that information?

It runs it through the wall-wetting equations and uses that information to determine how much fuel is actually going to go into the engine, and modifies the length of the squirt to make sure that the VE-commanded amount of fuel actually gets into the engine (as opposed to getting stuck on the intake port walls).

That is why it's important that the VE table is correct before tuning EAE. EAE is only going to try to make sure that the VE commanded amount of fuel gets into the engine during transients.

Is there a variable that already exists for AFRtarget minus AFR actual? You say that this can be done in the log viewer, but I'm struggling without a definition of all of the variables. Does such a list of definitions exist?

Thanks.

There is an AFR target variable and an actual AFR variable. You can use the log viewer to subtract one from the other.

Ken

[spittybug]

I successfully created a variable I call DeltaAFR which shows me how my AFR (as sensed) compares to my target. It helps show me how much I'm off, and it is a fair bit! It is defined as [AFR Target 1]-[AFR] so positive values mean I'm richer than target. Mainly I'm showing rich, but with occasional very lean spikes when I come off deceleration and then go open throttle. How much of a lag is there (time wise) between Target AFR 1 and AFR I wonder? My LC-1 is located a foot from one bank of cylinders, right near where the crossover pipe from the other side of the engine joins in. It's probably 3 - 4 feet from that bank. Any good way to determine that delay?

I'm not changing anything yet though. This shows me what I need to do is turn off the EAE and gently tune again. A couple of my target AFR transitions could be better placed too. Then I'll run this analysis again and it should show me close to 0 for the differential I hope. THEN I'll turn on EAE and see where I go astray. That should hopefully give me guidance as to where to "add or suck".

It runs it through the wall-wetting equations and uses that information to determine how much fuel is actually going to go into the engine, and modifies the length of the squirt to make sure that the VE-commanded amount of fuel actually gets into the engine (as opposed to getting stuck on the intake port walls).

My language was less than perfectly chosen... when I said that fuel was added or subtracted, I was trying to ask if indeed the pulsewidth was being modified to account for the fuel that is sticking or coming off... the overall net fuel delivered doesn't change, but more is being added when the ECU is being told that it is getting stuck on the walls and some is being removed when told it is coming off the walls; all to ensure that the amount we want (VE table value) is indeed what makes it into the engine. Did I say that better?

Using the scatter plots I've looked at DeltaAFR against RPM and I think I will find it very useful once I do the above. Is there a way to set the X and Y scales on both scatter plots and the normal log graphs? I'd like to center DeltaAFR around 0. These kind of analytics are very powerful and really help to visualize what is going on. Thanks for the good work!

[muythaibxr]

I successfully created a variable I call DeltaAFR which shows me how my AFR (as sensed) compares to my target. It helps show me how much I'm off, and it is a fair bit! It is defined as [AFR Target 1]-[AFR] so positive values mean I'm richer than target. Mainly I'm showing rich, but with occasional very lean spikes when I come off deceleration and then go open throttle. How much of a lag is there (time wise) between Target AFR 1 and AFR I wonder? My LC-1 is located a foot from one bank of cylinders, right near where the crossover pipe from the other side of the engine joins in. It's probably 3 - 4 feet from that bank. Any good way to determine that delay?

There isn't currently a great way to determine that delay in the MS2/extra firmware. What I often do is purposely turn off all AE, and stomp the throttle, then look at the amount of time between the MAP starting to increase vs the AFR starting to spike. I know this won't get an exact number, but it's good enough for a ballpark.

I'm not changing anything yet though. This shows me what I need to do is turn off the EAE and gently tune again. A couple of my target AFR transitions could be better placed too. Then I'll run this analysis again and it should show me close to 0 for the differential I hope. THEN I'll turn on EAE and see where I go astray. That should hopefully give me guidance as to where to "add or suck".

Good plan.

My language was less than perfectly chosen... when I said that fuel was added or subtracted, I was trying to ask if indeed the pulsewidth was being modified to account for the fuel that is sticking or coming off... the overall net fuel delivered doesn't change, but more is being added when the ECU is being told that it is getting stuck on the walls and some is being removed when told it is coming off the walls; all to ensure that the amount we want (VE table value) is indeed what makes it into the engine. Did I say that better?

Yep.

Using the scatter plots I've looked at DeltaAFR against RPM and I think I will find it very useful once I do the above. Is there a way to set the X and Y scales on both scatter plots and the normal log graphs? I'd like to center DeltaAFR around 0. These kind of analytics are very powerful and really help to visualize what is going on. Thanks for the good work!

Don't know... Phil Tobin (who wrote the log viewer) might be able to help you there. You can ask that in his section of this forum:

viewforum.php?f=100

Ken

[spittybug]

I've got myself a pretty good driving tune. I took a long log today and have attached it along with my .msq No bucking, no hesitation, drove very nicely. I can improve the after start and the warm up, but I'm not worried about that right now.

In looking at the log however, I'm a couple of % richer than I want to be, fairly consistently. I notice too that there are occasional transitions where I go very lean momentarily. This tune was done a lot by hand by taking logs at certain RPM levels and monitoring AFR values. I then changed the VE values and did it again..... All of this was done with car parked. Those runs were giving me AFR deltas (difference between what my AFR table calls for and what I actually get) that were +/- .5% As this log shows however, once out on the road, my delta is far greater. The VE analyze values for the VE tables were considerably lower, and making me go lean in many places.

Should I be adjusting the base VE table in this case or the EAE curves?

Is there a mathematical starting point to determine how much % wise to lower the VE table? I know how much AFR I wish to drop, I know a starting VE level and I know injector ratings....seems this ought to be calculable but I'm just not there yet.

Thanks!

[spittybug]

Made things a little bit better again today. Without a dyno, I'm just driving and logging and checking my AFRs.
I tried EAE with the lag turned on.... DISASTEROUS.... rendered car totally undriveable. I have previously tried to go up on the number of squirts (set at 2), but with no luck; it just kills the car.

Transitions from very low map levels to high still make me go off the chart lean for a quick spike. I don't feel it in the car, but I see it on the logs. It must be affecting acceleration performance too. I tried upping the % added to walls at the lower half of the MAP range, but all that seemed to do was make me richer in steady state; didn't help the acceleration spikes.
I don't have standard AE turned on as I found it difficult to avoid herky jerkies. Perhaps I need to set a mapdot threshold >100 and then play with that enhancement curve? Or am I doing something wrong with the EAE?

[muythaibxr]

EAE is difficult to get right on large fast throttle movements. Just add a tiny amount of normal AE (with a normal-sized threshold and the value at that threshold 0 with a gradual curve up to where you want AE to actually do something) to "get EAE started."

The reason for needing to do this is that MAP doesn't change until around the same time the amount of air going into the engine changes, which leaves the MS in a state of reaction slightly behind actual conditions. The only way around that is to do something TPSdot or MAPdot based (AE) and use that to bump up the PW in-time with the MAP changes instead of lagging behind.

Ken

[spittybug]

Thanks Ken, I'm going to give that a try. It really only is these momentary spikes that I'm trying to resolve at this point (at least without dyno testing for the ultimate in tuning), so I'll just set the threshold high enough to not always kick in and then slowly increase the curve to the point where it minimizes spikes but doesn't cause bucking.

I'm going to use mapdot since I want to be able to share my efforts with the stock Delorean conversion guys who don't have TP sensors. I also find the TPSdot too noisy in my past experience.

[muythaibxr]

I'm going to use mapdot since I want to be able to share my efforts with the stock Delorean conversion guys who don't have TP sensors. I also find the TPSdot too noisy in my past experience.

Strange. I've always found MAP to be a lot more jittery than TPS.

Ken

[spittybug]

Hummmm. Not a good first attempt with the AE. I set the threshold at 150 and curved it from 150 to 350, 0 to 5 ms adder. At high kPa values, I get real herky jerkies and my AFR goes really rich, so I know that curve is too aggressive. I notice however that the AE kicks in even though the 150 threshold isn't met.... huh??? Visible in logs.

BUT, in digesting the logs I have come to realize that the times I go off the chart lean is when the load drops like a rock down to the teens kPa level (momentarily). I wasn't taking into account the fact that the AFR being recorded has a bit more delay on it than the MAP does. I was seeing the RISING map as making me go lean, where in reality it was the DROPPING map preceding it. The pulse widths go to zero despite there being VE values in the bins that should result in at least a 2ms pulse width. It is that 0ms PW for a beat or two that is killing the AFR value. I double checked and my decel fuel amount is set to 100% so that isn't it. My O2 sensor controller authority is set to 0 right now, so he isn't modifying it. No over run fuel cut enabled. I'm assuming therefore that EAE must be doing it?

Just to be sure of things I did a free air calibration of the O2 sensor this morning, prior to today's tests.

Another thing, why does my kPa go over 100 when I'm NA with no boost? I live in Houston, essentially at sea level and I have the initial MAP reading setting the barometric correction. I'm a rear engine car, so there is no "scoop" or other means to ram air.

Sorry for so many questions, but I'm the type of person that really likes to understand what is going on so as to be able to better diagnose things in the future. Thanks for helping.

[muythaibxr]

The PW can go to 0 from EAE if the code thinks a lot more is getting pulled from the walls than added during a transient. That is normal behavior. I personally take advantage of that to make entering overrun fuel cut smoother.

As far as kPa values over 100, I can't answer that aside from you must have an interesting intake setup!

Ken

[spittybug]

So if I wish to prevent going to 0 PW as I rapidly drop from high MAP to low, how do I adjust the EAE curves? (he asks out loud)

Since I'm going lean, it thinks more is being sucked off walls than truly is and reduces the PW accordingly. That means I either add more on the "added to walls" or remove some from "sucked from walls", right? Is it preferential to do one over the other? Does the whole curve move down the Y axis or am I playing with one end or the other? I'm going from very high to very low MAP value (essentially the whole range of the curve), that's why I'm not sure where/how to adjust these curves. Do my curves look way off? I'm using pretty close to the default ones. Are there "better" ones floating around out there?

As far as my intake is concerned, nothing odd! K&N air filter inside the fiberglass side pontoon with 4" hose to mouth of throttle body. Pontoon does not "capture" or funnel air in any way, it is just a hollow fiberglass body part that the Delorean's stainless steel panels bolt onto.

[Peter Florance]

Made things a little bit better again today. Without a dyno, I'm just driving and logging and checking my AFRs.
I tried EAE with the lag turned on.... DISASTEROUS.... rendered car totally undriveable. I have previously tried to go up on the number of squirts (set at 2), but with no luck; it just kills the car.

Transitions from very low map levels to high still make me go off the chart lean for a quick spike. I don't feel it in the car, but I see it on the logs. It must be affecting acceleration performance too. I tried upping the % added to walls at the lower half of the MAP range, but all that seemed to do was make me richer in steady state; didn't help the acceleration spikes.
I don't have standard AE turned on as I found it difficult to avoid herky jerkies. Perhaps I need to set a mapdot threshold >100 and then play with that enhancement curve? Or am I doing something wrong with the EAE?

Highly recommend more squirts, even if you have to retune VE table.

[muythaibxr]

As Peter says, more squirts will help if you can manage it. Additionally, I would adjust the SOW curve down to solve your problem. To figure out where in the curve to do that, you should do some smaller throttle lifts from/to certain loads and adjust only the areas that need it down.

Ken

[spittybug]

I'll give it a try.
Not sure how to actually go about altering VE table to get 6 squirts to work, but will read up. When tried before, it immediately kills car. (Oddfire, 6 cyl, port injection, 19# injectors).

Don't know if it would have had any affect, but the EAE tables didn't go to as low a MAP value as my engine dives down to. Does "falling off the bottom of the curve" perhaps do something like zero out the PW? The Y axis scale of those curves is "Correction %". I assume this is the upward or downward correction to the PW? If so, how can it actually go as low as zero?

[muythaibxr]

Fallig off the curve just means you can't tune that.

Ken

[spittybug]

Doesn't look like I can go 6 squirts since I would fall down to 3ms per squirt (total REQfuel = 18ms). If I go to 3 squirts, it goes to the recommended minimum of 6ms and the idle pulsewidth will probably be right around 2ms. I'm running simultaneous on the banks.

Will try shortly.

[Peter Florance]

Doesn't look like I can go 6 squirts since I would fall down to 3ms per squirt (total REQfuel = 18ms). If I go to 3 squirts, it goes to the recommended minimum of 6ms and the idle pulsewidth will probably be right around 2ms. I'm running simultaneous on the banks.

Will try shortly.

I idle at 1.8msec
Opening time becomes important