[kkeener]

I tuning my 94gt with an S-trim to run on E85 ethanol. To make this work the car needs to run at an AFR about 9.7 instead of 14.7.

I am able to do this in Open Loop, The car seems to run well there, but as soon as Closed loop kicks in it tries to bring the AFR up, but not to 14.7.

The Lambse values in closed loop idle now are around 11, so it is not trying to get the number 14.7. I believe this is because Ethanol brings it's own Oxygen to the party so the Hego reads .5ish at 11:1 AFR.

So that is a start, but I need to get the Commanded AFR down a couple more points.

I tried the HEGO Bias, but that has a limit of around -0.0139 I need the number to be -0.118 to get to the correct ratio.

Is there any way to do this, or do I have to run 100% Open Loop with E85?

Here is what I have altered specific to e85 so far:
Minimum Lambda .70 from .79
Multiplied all values in Fuel_table_Stabilized_OL and Fuel_table_base_OL by .659

[2Shaker]

Interesting. The O2 sensors are designed to revolve around 14.7 so it will might just have to be OL all the time. By the way, if you haven't searched yet there were a few on the forum fooling with OL all the time. try to find and I'm sure there will be good pointers.

This does leave me wondering how it is that my 2007 F-150 can be dual fuel... :?

[cgrey8]

I've put some thought into this since I've considered converting my truck to E85 when it is available here in Atlanta. However since it isn't, all I can spout off are theories and known issues to be dealt with...no actual experience so keep that in mind as you read through this.

As indicated by your comments, E85 requires much more fuel for the same amount of air than gasoline does. This means you will either need to tell the EEC that it is no longer targeting 14.64 OR somehow fool it into believing it's still gasoline you are burning. There may be some neat scalar or some other method of setting the EEC's adaptive learning to target some other LAMBSE closer to your 9.7 vs 14.64. However I don't know what that would be. If somebody knows the way to get GUFx or CBAZA to target a different LAMBSE during adaptive learning, then you are home free.

Since I don't know of a way, I wouldn't bother fighting the computer to get adaptive learning (i.e. KAMRFs) to work. I'd just give the EEC what it wants, and that's to see LAMBSEs between 14.5-14.8 during CL conditions. The way to do that is either by lowering injector slopes OR increasing the MAF curve.

Given those choices, I'd choose to adjust injector slopes so my load calcs don't get thrown off. With increasing the MAF curve, you'd also need to increase your CID scalar by the same percentage to keep the Load calc right. Since you'll be messing with injector slopes anyway just as part of the tuning process, it just makes sense to do all your adjustment (legitimate and fake-out) there. So adjust the injector slopes lower so the EEC gives the engine more fuel. You have 30lb injectors (according to your sig), so you may need to upgrade to 42s OR ramp your fuel pressure way up to run E85 especially if you have a beefy setup that's already running those 30s in the 80% PWs on gasoline.

Assuming you are only using your current injectors up to 50% MAX PW@WOT with gasoline, you probably have enough injector to work with. Enter an injector slope that's a good 35-40% lower than actual so the EEC will deliver more fuel for any given amount of air being measured by the MAF. What this will do is get the system to where it "looks" like gasoline is being burned by the engine. You'll still target for mid-14s in the LAMBSEs and KAMRFs will adjust as they are supposed to trying to target 14.64. So at that point, all your tuning will be just like it would be with gasoline. Adjust MAF, transient fuel, acceleration fuel, spark, etc etc just like it was gasoline at this point.

So at this point, you may be wondering about the HEGOs since you mentioned adjusting the HEGO Bias. HEGOs could care less about actual AFR, they only switch when they see rich or lean conditions, regardless of what's being burned. Remember, all they are doing is referencing the amount of oxygen is present in the exhaust vs combustibles. So no Bias adjusting beyond what you'd do with normal gasoline should be necessary (i.e. leave it alone until you get other parts of your tune nailed down).

So how does that play out with a wideband? If you have a WB, the transfer of what voltage equals what AFR will not be accurate with E85. Your transfer is calibrated for gasoline. So if you run E85, what's actually around 9.7:1 (or whatever Stoic is for E85) will register as around 14.7:1 (what is Stoic for gasoline). For the same reasons that the HEGOs could care less about AFR, WBs only indicate a wider range of rich/lean conditions. So you'll need a new transfer to accurately reflect what voltages are analogous to what E85 AFRs. But if it were me, I'd keep the gasoline transfer. Remember, by lowering the injector slopes, you've made it appear that gasoline is what's burning in the engine now. This way, you can still fine-tune your injector slopes/MAF transfer against the WB. Since the LAMBSEs will be reading "gasoline" AFRs to maintain Stoic in CL, it just makes sense to keep the WB reporting the same numbers for Stoic.

Now again, will this work? I have no clue. I haven't tried it. But if I ever get to the point where I want to convert my truck over to E85, this would be my plan of attack. But I'd love to hear some "real" feedback from someone else whose actually done this and knows what works and what doesn't.

[cgrey8]

...This does leave me wondering how it is that my 2007 F-150 can be dual fuel... :?

I don't know this for sure, but I think the flex fuel vehicles are using wideband HEGOs as running HEGOs. Since the EEC is designed to work with 100% gasoline, E85, and whatever mix of the two happens to be in the tank, it just makes sense that the EEC have a wider scope of detection on exhaust.

I think even non-E85 vehicles are moving over to wide-bands for emissions reasons. Think about it, if the EEC is off by a certain amount, it knows by how far off so it can adjust the LAMBSEs much faster to get things back in-line. With narrow-bands, it only knows if its rich or lean, but has no clue just how rich or lean things really are. So for lean tip-ins, we currently have to correct the tune to alleviate what's causing the tip-in to be lean. But if the EEC was monitoring a wideband, it could see whether it was just barely into 15:1 or if it suddenly experienced a spike into the mid 16s and needed to compensate in a hurry to get things back to stoic.

And again, since HEGOs don't really care about what's being burned, they just report the rich/lean condition as it exists, it doesn't matter whether it's E85, gasoline, or some other Exx equivalent.

[Cougar5.0]

On the Innovate it's a simple matter of reprogramming the output/transfer curve to read the correct AFR for E85. This would be good for a sanity check if you really feel the need to "see" that the ratio is indeed correct when using the E85.

I was just thinking - the ECU uses lambda anyway and we multiply by 14.64 in the definition files to "see" the AFR in actual air/fuel ratio. It seems to me that you could write a new definition file for E85 that would simply change the multiplier from 14.64 to 9.7 (or whatever stoich is for E85). I'm sure I may be overlooking something obvious here, but it ought to be that simple. I agree with cgrey8 that simply changing the injector slopes by the difference in lambda ratio should be enough to get started though.

[cgrey8]

I didn't think about that, but yeah Change the calculation in the BinaryEditor definition to whatever AFR is needed to meet stoic for E85. That doesn't stop you from having to drop the injector values to meet stoic, but it does at least get LAMBSEs to display/datalog out to actual AFRs instead of gasoline-equivelant AFRs.

At some level, I still believe there is some unexposed value the EEC is using during the calculation of injector PW that represents AFR for gasoline. Think about it, the EEC gets a MAF value indicating how much air has come in. At that point, it has to calculate how much fuel goes with that much air. Once it knows how much fuel it needs, it runs that past the injector slopes to get a PW that it figures will deliver that much fuel. Point is with that scalar exposed, you could change that from the equivalent 14.64 (actual value in the EEC may not be 14.64) to an equivalent 9.7. Then modify the definition file in BinaryEditor to multiply short term fuel trim lambdas by the same scalar value to get what we know as LAMBSEs, (as Cougar said). The result, everything should work with actual injector sizes...no trickery.

At this point, you could configure BinaryEditor with the wideband transfer for measuring AFR of E85 exhaust, and you are set.

So, anybody know where in the BIN the EEC is calculating measured air into fuel requirements?

[kkeener]

cgrey8,
I did upgrade my injectors, I just forgot to get that into my sig. I am running 42lbs now.


I was hoping to not run OL so I could have the efficiency of CL

I was worried about fudging the Injector Slopes because I thought it might throw something else off, not sure what that might be.

Cougar5.0,
I was hoping the Nominal AFR was what was used in that calcuation, but when I put in 9.7 (Stoich for e85) the car wouldn't start. I didn't really look into why.

Anybody know what that Nominal AFR scalar does?

I've never done anything in the BinaryEditor, any idea what I would need to change?


So going the injector slope route, what do I actually multiply the slopes by? 9.7/14.7=.659
but the lambse values seem to be stablizing around 11, so do I go with
9.7/11.x=.8xx

[cgrey8]

If you are getting 11s for LAMBSEs, then I'd go with 11/14.7= ~.75 for a multiplier to multiply the injectors by. I thought it was closer to 35% myself, but perhaps not UNLESS you still had some residual 100% gas in your tank when you filled it with E85 in which case you may have something closer to E70ish in the tank.

[sailorbob]

You are not going to be able to run closed loop with E85 because AFAIK there is no narrow band exhaust gas sensor that operates at the stoichiometric A/F ratio for E85 (9.765:1).

You will also need to ensure that any wideband lambda sensor can measure an A/F ratio richer than 6.975:1 to allow you to tune for maximum power.

[cgrey8]

Sailorbob, how would the HEGOs know what AFR is? It was my understanding of HEGOs that they measure the content of oxygen vs the content of hydrocarbon in the exhaust. If there's an abundance of oxygen, the the combustion was obviously lean. With a lack of oxygen (and an abundance of hydrocarbon), the mix was obviously rich. So the actual amount of each that was present before the combustion is no longer a factor when what you are looking at is byproduct result of combustion...and specifically how much oxygen was left over after the combustion. So since stoic for E85 is a different AFR, the HEGOs will still switch whenever a stoic exhaust byproduct is detected. Is that not the way it works???

However I can see that the wideband's curve would need to be tweaked to get a more accurate indication of actual E85 AFR since it doesn't deal with such a narrow range of exhaust mix. I can see that the rate of change from stoic (both lean and rich) would be quite different.

I'm not a HEGO design engineer so I'm working completely off vague understanding of HEGO operation. But I'm always eager to learn why I'm wrong if someone can explain it to me.

[sailorbob]

Closed loop operation using a narrowband lambda sensor works by the ecu constantly 'switching' from lean to rich and vice versa about the stoichiometric A/F ratio for normal pump gasoline. If you look at a graph of the voltage vs A/F ratio of a narrowband sensor you will see the voltage changes greatly (relative to the output range of the sensor) for very little A/F ratio change at stoichiometric and ecu looks for this change. Once the A/F ratio is no longer near 14.64:1 the output from the narrowband sensor hardly changes and is of very little use as a control input.

To do the same kind of closed loop operation with a different fuel you would need a sensor that exhibits the same kind of behaviour for the stoichiometric A/F ratio for that particular fuel.

[cgrey8]

I agree with almost everything you are saying. However, my gut feeling is the HEGO would switch at whatever AFR stoic is met at for the combustible being burned.

Following your logic, the narrow-band would continue to exhibit switching behavior around 14.64 AFR even with E85. If you burn E85 at 14.64 AFR, there'd be a MASSIVE amount of oxygen left over in the exhaust because that would be like running gasoline at 20+ AFR. If HEGOs really are sensing oxygen content, it seems to me, they'd pickup on that as a VERY lean condition as opposed to switching.

Please understand, I'm not trying to sound adversarial or argumentative. I'm just presenting my thoughts as best as I can get them out my fingertips. NO disrespect is intended, I promise. If there's something wrong with the way I'm viewing things, I just want to understand what's wrong with my logic and the only way I can do that is to throw it out there and see how it can be debated. Sometimes that means I eat humble pie for all the read, but that's OK as long as I learned something from the experience (saving pride at the expense of ignorance isn't good economics to me).

[sailorbob]

Yes, I see what you're saying and thinking about it, I believe you are right :)

I was wrongly looking at it from the the viewpoint of how the eec works and this is where the problem is going to be. The eec regards a 14.64:1 A/F ratio as a LAMBSE equal to 1.0 and there isn't enough adaptive learning to swing the fuelling to ever get the lambda sensor switching about stoichiometric for E85. This is probably why the CEL comes up on non flexible fuel vehicles running E85.

[cgrey8]

Now we are on the same wavelength, and my way to thinking on that is to lower the injector size so the EEC treats the injectors like smaller injectors than they really are so they deliver the 35% or however much more that's needed.

As for display of the short term lambdas, Cougar5.0 suggested changing the definition file to multiply the short term fuel lambda datalogged by 9.7 (or whatever is stoic for E85) instead of 14.64 so that the LAMBSE displayed and datalogged represents the actual AFR, not what appears to be a gasoline AFR that makes no sense with E85.

And as for the wideband, an E85 transfer is needed to translate the WB voltage into E85 AFRs.

At that point, tweak away in closed loop.

What I was hoping you (sailorbob) could do, since you are our resident BIN-hacker, is to find where in code the EEC is associating 14.64 parts air to fuel so we can expose that parameter and adjust it to be what E85's AFR is. Then no trickery of the injectors would be required. You could enter actual injector flows. At that point, you'd have the basis for a complete E85 tune.

[kkeener]

I posted my log in the logs section. I did not get a chance to try just altering the injector slopes yet.

http://eectuning.org/node.php?rid=52

The interesting thing to see in the log is the HEGO during open loop is around .8 and Lambse around 9 which is a little richer than I want to be, but close to what I asked for.

Then Closed Loop, the HEGO bounces around .5 and the Lambse shows around 11. So the fact that it didn't try to lean it out much more leads me to believe that the adaptive was looking for a stoich in the HEGO and the lambse is calculating the actual fuel it pushed in there. 14.65 had nothing to do with the calculations.

If that is the case, why was stoich around 11:1 AFR? There was a little gasoline left in the tank, but not much at all. Also, closed loop was kind of sputtering and died a couple times.

[sailorbob]

If you are going to 'fool' the eec by changing the injector slopes I would suggest using the ratio of 14.64/9.765 (i.e. the two stoichiometric values for the fuels). Then yes, you would need to trawl through the definition file and alter any fuel related parameter that has multiplier or divisor with a value of 14.64 to 9.765 to compensate for the injector slope change.

Fixing the code would of course be best but not something I'd be willing to take on (far too many other things to do :D).

[cgrey8]

Well, your logs are confusing.

Have you pulled the battery cable since you filled the tank with E85???? You got KAMRFs learned down to .84 which tends to indicate the EEC is trying to pull 16% fuel, not add fuel. But the LAMBSEs are sure homing in on low 11s to switch the HEGOs. If the KAMRFs were learning, they should rise, but the adaptive update and possibly ACT are probably preventing them from moving in the driveway there. Just resetting them back to 1 by pulling the battery cable would get you a good point or so higher on your LAMBSEs.

Your Adaptive Learning table may have something to do with why the EEC is displaying .84s for idle here.

For now, reset your KAMRFs so they go back to 1 and see what this does to things. If the engine is still running the LAMBSEs too low, start to lower your injector slopes (both high and low) until LAMBSEs are in the 14s (where the EEC wants them thinking it's gasoline).

Once you get the slopes giving you 14s at idle and reving the engine without load isn't immediately dropping the HEGOs lean on rev, you might try loading up the engine a tad just to see how it does.

[Cougar5.0]

A little gasoline could be 2-3 gallons (bottoms of tanks hold deceptively large amounts of fuel)?

BTW, here is a great post on the Innovate forum that talks about alternate fuels and how the meter should always read 14.7:1 no matter what fuel is used: http://www.innovatemotorsports.com/foru ... php?t=1552

One person suggests just leaving it at 14.64 stoich and tuning as if you had gasoline. The % deviations will be the same so you really don't need to do anything except adjust how much pulsewidth is calculated for the new fuel. Just change the slopes and tune from there as if you had gasoline in there.

[kkeener]

I must not have disconnected it long enough. And ya, the fuel is definately blended down a little from the existing gasoline in the tank.

So here is my plan of action:
Top off the tank with E85
Clear KAMRF
Go back to my Gasoline tune then multiply the injector Slopes by:
9.7/14.7 = .6598

Which changes
High: 41.9992=>27.7
Low: 47.3791 => 31.2607

I'll let you guys know how it goes.

[cgrey8]

Based on what your datalog is doing, that plan may get things too rich, but lets see what happens. I agree 2-3 gallons could affect effective AFR by quite a bit especially since alcohol likes to get itself mixed in with the other liquid to reduce the volume. Said in another way, 1 quart of alcohol and 1 quart of water mixed together do not give you 2 quarts of alcohol and water mix. I suspect alcohol works with gasoline in much the same way. There's a technical word to describe that phenomenon, but I can't think of it right now.

I'm really curious to know how this works out for him especially since I've considered doing this same thing when E85 gets more popular in my area. Definitely keep us posted on your progress.

[2Shaker]

Boy, I had to read this one a few times. And Cougar - thanks - that link helped. For a while I thought our friend cgrey8 was way off base (but he was so plolitely insistant and now I get it :-)) In my mind we can summarize all of this to say both the narrow band and WB sensors are measuring Lambda. The 14.7 value is a calculation. We are tuning around Lambda and not AFR. Agree? That's why changing the slopes makes sense and somehow my F150 knows how to set the slopes based the mix of fuel coming through at any given point. (hmmm, I wonder how it does that :? )

[kkeener]

ok, I just had a few minutes, but I tried going back to the gas tune and just multipling the injector slopes .659.

the log is here:
http://eectuning.org/node.php?rid=53

It is pretty close. I think it is mostly fine tuning those slopes.

Although I am still having a problem clearing the kamrf, I disconnected the battery and held down the brake for a minute. I am trying disconnecting overnight now.

2Shaker, here is an interesting article on how Flex fuel sensors work:
http://www.megamanual.com/flexfuel.htm

It would be awesome to hack this sensor into the tweecer.

[cgrey8]

With your KAMRFs stuck, you might want to check your scalars for Adaptive Control min (MINADP) and max (MAXADP) lambda. Pulling the power should return you to 1 on the KAMRFs. It's weird that it isn't.

As for the log, it looks much better. But as I suspected, going the whole 33% overshot you a tad. You might try increasing your high slope some to get the LAMBSEs out of the 15s and back into the mid 14s.

Also something else you can try is to download a CBAZA tune (i.e. J4J1.bin) into one of your bin positions. When you want to clear the KAMRFs, turn the switch to that position and then turn it back to your tune's position. Of course, the wrong tune will make the motor run like absolute crap if at all. But it should clear the KAMs just like pulling the battery cable...just a heck of a lot more convenient.

[kkeener]

Keeping the battery off longer worked to clear the kamrf. also I loaded an A9L bin then back to my tune and that worked also.

New numbers are in, multiplied slopes by .7 the numbers came up pretty good.

I'll try some solid runs and let you guys know how it went, but I did a 20 mile drive on e85. Pretty cool!

[Adam McLaughlin]

What is the price of E85 where you live?

Adam

[cgrey8]

...also I loaded an A9L bin then back to my tune and that worked also...

I thought you were A9L (GUFB). But I see by your updated sig now, you are CBAZA. So yeah, downloading a CBAZA with a GUFx strategy BIN will do the same thing.

Glad it's working out for you. Like I said, I've considered doing this exact same thing at some point and I'm happy to hear that it isn't that difficult to make happen on one of these old processors.

The other things you have to consider with E85 is that it eats up rubber seals. You need special rubber seals everywhere and eliminate non-E85 resistant rubber anywhere else in the system. It's also mildly corrosive to aluminum. But there's only so much you can do about that.

I also understand that alcohol conducts electricity much like salt water does where gasoline doesn't. This affects the exposed leads that clip onto the fuel pump. With them exposed, the fuel is actually conducting electricity. Did you have to do anything with regard to that aspect to convert your system?

[kkeener]

E85 here in San Diego is $2.99 which I believe is on the high end of the nationwide prices. If you compare it to $3.40 for 92 octane you are doing pretty well.

Although even if I do end up spending a little more for it, I may be seeing more power (yet to be determined, but theoretically should be) and Not buying foriegn oil. Buying American made is really important as the dollar is falling like it is. We have a huge trade deficit with other countries that we need to work towards balancing.

Lastly, the Evironment impact of using a renewable fuel that takes in about as much CO2 as it puts out makes me feel good about driving a muscle car.

cgrey8,
I will be checking the seals and rubber after a few tanks, but cars have been mandated to be able to handle up to 10% ethanol since the late 80s so my car has the worst offenders removed already. I found a couple other people on forums that did nothing but bigger injectors and pump in the fuel system and have been runing e85 for 20,000+ miles with no impact on the rubber components.

Alcohol does conduct electricity more than gasoline, but I think most pumps have dealt with that already. I'll let you know if I run into anything.


I'm curious how the smog numbers will turn out. I'm gonna go get a test pretty soon.

[cgrey8]

The smog numbers should be low as long as you aren't running heavily rich.

Generally Nitrogen (N2) in breathing air is inert and doesn't get involved in chemical reactions. However it does under high temp conditions such as gasoline burning lean. Because alcohol burns so cool, there should be almost no NOx emissions. From what I understand, you could have a completely non-functioning EGR and the smog guys would never know it since EGR isn't needed to cool the combustion on alcohol engines.

And yes, you should see a good 5-10% increase in MAX power with alcohol vs gasoline especially with a high compression engine and some good advance.

The 3 biggest downfalls to E85 are:

1. It's corrosive to both rubber and metal (can be overcome)
2. It requires more of it to burn than gasoline (it HAS to be ~25% cheaper per gallon than gasoline before you are breaking even and currently it isn't)
3. Economic effects it E85 has on the cost of food when it's success becomes greater (this has already been seen in other countries that are adopting E85 like Mexico)

Right now engines are still being made dual fuel so they can take E85 or gasoline and any mix in between. This forces engines to still be low compression. Once E85 is more popular and it's nationwide, I expect you'll buy E85-only vehicles which will have compression ratios in the 12s. Then the fuel economy loss of running E85 won't be nearly as bad as it is right now. Running E85 in a naturally aspirated gasoline engine at 8.5-9:1 is like running gasoline in a 7:1 compression engine...and I back all that with absolutely no fact. Just my opinion...

[fidstang]

I thought that the narrow band O2's see voltage not actually read what's there. They are set from the factory to output a specific voltage at stoichiometric. So my theory is that the ecu will error and log a code for the rich condition. Mean while trying to compensate for the rich condition it is seeing.
Just my thoughts don't know if it's correct or not.

[cgrey8]

You are absolutely correct with your understanding of HEGOs switching at stoic during CL operation.

What I was getting at is running the engine leaner via the HEGO BIAS. My stock X3Z tune already has negative values for HEGO bias in the loaded RPM/Load conditions. I've lowered my HEGO BIAS a tad more to keep me on the rich side of the HEGO switch more than the lean side. Don't get me wrong, my HEGOs still switch quite regularly to maintain close to stoic, but there is a slightly more obvious preference or bias to staying rich more than lean during loaded conditions vs sitting and idling. As well, the EEC more-aggressively responds to lean conditions (i.e. lean tip-ins) so I almost don't feel them and as a passenger, you probably wouldn't feel them.

With E85, as long as you don't run into drivability issues (again those lean tip-ins), I'd think you could get away with not being biased to rich conditions even under load thus helping you to stay well below the CO and HC emission standards (CO and HC are the two emissions that go up as the AFR gets rich).

I didn't explain myself very well on that one...my bad.