E85Forum.com

pete · Joined: 11 Oct 2005 Posts: 2 Location: Phoenix AZ

Hello all.
I'm in the process of building a performance EFI system for the Ford 302 V8. My business partner brought up the idea of modifying a system to run E-85. My biggest question is with the O2 sensor. Our computer is fully user controlled but has the ability to use the O2 sensor for mixture correction. Is there any difference in the O2 content in the exhaust when burning gas at stoich and E-85 at stoich?

We also have the ability to use a Wide Band O2 but again I'm unclear what it will tell me when running E85.

The computer allows you to upload it's configuration to a laptop or palm and then download a different set of parameters. We would like to develop two sets of parameters for the computer. One for E85 and one for gas. This way the owner could simply download the gas configuration when he/she can't get E85.

hotrod · Joined: 19 Apr 2005 Posts: 872 Location: Colorado

No your O2 sensor is a "lambda sensor" and only cares about how much excess oxygen is present.

Stoich mixtures always return a lambda value of 1.

In closed loop my WRX has no issues at all with widely varying E85 percentages in the tank. It quickly sees the need to modify fuel trims and within a very short time you can watch the fuel trims change if you have an OBDII scanner hooked up.

The only caution is that when running the wideband if possible set it to report mixtures in lambda or equivalence ratio (which ever you prefer) so that you get valid numbers when mixing different types of fuels.

Larry

pete · Joined: 11 Oct 2005 Posts: 2 Location: Phoenix AZ

Thanks. That's what I wanted to know.

jachristner · Joined: 12 Oct 2005 Posts: 64 Location: Kokomo, IN

What are the differences between a normal and a wideband O2 sensor?

Are there advantages or disadvantages to either one in a car being converted to run on both full Gasoline and E85?

hotrod · Joined: 19 Apr 2005 Posts: 872 Location: Colorado

The common O2 sensor you find in most cars for emissions control and the cheap AFR gauges is only useful very close to a stochimetric mixture.

It switchs its output very drastically with very small changes in AFR. Pretty much tells you only you are rich of stoich or lean of stoich, but of very limited use to see small changes in AFR.

A wideband sensor is designed with a much more uniform response so it can see mixture changes from very rich (like 10:1 to very lean like 16:1)

Serious tuning demands use of the wide band sensor.

Larry

jachristner · Joined: 12 Oct 2005 Posts: 64 Location: Kokomo, IN

From what I could find on the internet explaining how the sensor works, I understood that it wasn't reporting the AFR of the engine but rather the percentage of gaseous O2 in the exhaust stream. If that's the case, the sensor would only know if the engine is running lean and maybe a rough estimate of how much, but not much else as a rich condition would most likely leave no or little O2.

B.A.Bowling · Joined: 15 Feb 2006 Posts: 6 Location: Baltimore, MD

I will add a bit to this thread....

A narrowband O2 sensor will show a point of stoichiometric the same for E85, gasoline, or any blend. In other words, a voltage of 0.45 volts from the sensor (Nernst cell) indicates a stoichiometric mixture.

However, the use of a wideband is a different story. The wideband utilizes an oxygen pump in feedback with a Nernst cell in order to maintain a stoich reading from the Nernst cell. However, the calibration curve, especially on the excess fuel (rich) side is significantly different for ethanol compared to gasoline. The reason for this is that the sensor reacts to a couple of the components that exist in the exhaust, and the partial pressures of these components will be different for different fuel blends. To use ethanol blends with a wideband requires a meter that allows you to enter the H/C (Hydrogen to Carbon) and O/C (Oxygen to Carbon) ratios. There is a chart I generated on the H/C and O/C ratios for different blends of ethanol to gas - see http://www.megasquirt.info/flexfuel.htm

If there is interest, I can go over in detail the chemical relationships (in plain english, I hope) to illustrate this, and how the wideband sensor operates specifically to ethanol. There is a lot of detail here: http://www.megasquirt.info/PWC/

- Bruce

hotrod · Joined: 19 Apr 2005 Posts: 872 Location: Colorado

Very good info there, thanks for the addition to the board.

Your system looks like a good alternative for a lot of folks that deserves some investigation by those looking for a conversion option.

B.A.Bowling · Joined: 15 Feb 2006 Posts: 6 Location: Baltimore, MD

hotrod · Joined: 19 Apr 2005 Posts: 872 Location: Colorado

Very interesting summary.

Thanks for all the work on that post.

Now the important points about wide band sensors I see are:

1. For practical purposes a stoich mixture reading can be assumed to be accurate for any common fuel mix.

2. With Gasoline / ethanol fuel blends on the lean side of stoich (the most important number due to lean combustion temperatures) your wide band errors will be small .

3. With Gasoline / ethanol fuel blends on the rich side of stoich the wideband will report mixtures which are too rich with the error increasing as you move farther from a stochiometric mixture.

4. The larger the ethanol fraction the higher the over-rich error will be.

In practical terms ---- while tuning gasoline ethanol blends ----- if your wide band reports a lean mixture, believe it.

If the wide band reports a rich mixture, you are probably not quite as rich as it looks. Do not lean out a rich mixture too much in one step, or you may end up leaner than is safe.

Larry

B.A.Bowling · Joined: 15 Feb 2006 Posts: 6 Location: Baltimore, MD

Yep, Larry you got it right!

Added info: practically all of the wideband meters out there use the "default" response curve that was published in the LSU-4.2 datasheet to generate lambda (then AFR from this). Here is the curve:

It was generated assuming a H/C ratio of 2 and perfect combustion.

It is possible to use the pump current obtained from the meter (most will give this to you) and then make up your own curve. An easy way would be with an exhaust gas analyzer - many shops have this and may allow you to run a few points. Lean is easy, and will be just a scale factor (from stoich). Rich is a little tricker, but use the CO readings from the 5-gas to calibrate with a few points. Unfortunately you will not get a H2 reading, but the CO reading will help scale things to make the curve much better. Once you make up a new table of lambda (or equivalence ratios, for academic types) you can then be confident on the tune.

Of course, if you have access to a 5-gas, then it may be better to use this for the tune Smile

. Just make absolutely sure that they set the H/C and O/C ratios in the meter (most will not even know these are there for setting), or the calculation (using Brettshneider) will be in error. In fact, 5-gas analyzers errors crop up all of the time if they assume regular pump gas but actually burning 10% ethanol (equiv) enriched...

- Bruce