• Protip: Profile posts are public! Use Conversations to message other members privately. Everyone can see the content of a profile post.

Air-Fuel Ratio

Yellow Rose

Suspended
Joined
22 November 2001
Messages
2,256
Oxygen sensors are in the exhaust pipes; therefore, how is it that they monitor the volumetric ratio of air-to-gasoline which is in the intake side of the engine? If by inference, this implies that the sensor controller has an algorithm which assumes a combustion efficiency of X%, no? I guess that the sensors are actually looking for a combustion byproduct, and by stoichiometry, deduce the theoretical air-fuel ratio? If so, what parameter and equation is employed?
 
Andy

Good question, I am sure there is an expert out there. I have made some assumptions about the 02 sensors and would be interested in a more detailed explanation of the monitoring and control of the fuel/air ratio in an internal combustion engine.
 
Great question!!
I do not know the answer, but I do a lot of air/fuel ratio testing with FI motorcycles.
I use a Dynojet 250 dyno that can be used with an eddy current absorber or as an inertial dyno.
I do make custom maps using an O2 sensor. I use the Dynojet Tuning link software, that almost automatically makes the map.
Is there anything like it in the automotive world??
I find that both different engines and different fuels make their best HP at different fuel ratios. they are usally between 12.5 to 13.2 to 1.
 
According to the shop manual the O2 sensors measure the Oxygen level in the exhaust and outputs a voltage dependant on how rich or lean the air/fuel ratio is.

I looked up:

stoichiometric
adjective
(Chem)
1 concerned with, involving, or having the exact proportions for a particular chemical reaction
example: a stoichiometric mixture

2 (of a compound) having its component elements present in the exact proportions indicated by its formula

3 of or concerned with stoichiometry
[ETYMOLOGY: 19th Century: see stoichiometry]


The manual does not say what the stoichiometric ratio is but I think it is something around 14 to 1. So when you burn 14 parts air with 1 part fuel the amount of O2 that is left would yield a output by the sensor that would be at a voltage level that indicated an stoichiometric air/fuel ratio.
 
The engine controller is given the inputs of amount of air (MAF or MAP sensor) and amount of fuel (based on precisely controlled injectors). Therefore, it "knows" (this part I'm speculating that some sort of simplifiying combustion equation is used) how much oxygen to expect to be read by the O2 sensor. If this amount of oxygen expected is off, the fuel delivered is modified and learns over time how much fuel to deliver (fuel trim). However, another interesting thing is that this "expected oxygen" amount is not constant over time, and not just based on throttle position (rich at WOT, etc.) but the engine controller is constantly varying the combustion from rich to lean, rich to lean, I believe it usually cycles about 5 times/second and this is done for cataylst efficiency (sorry don't know why the cataylst is more efficient this way).
 
If you remember your high school chemistry, it will help a little bit. When gasoline vapor which are basically HC (hydrocarbon) molecules and oxygen react in a combustion process, they combine to form mostly CO2 and water. But from chemistry, we know that molecules like to combine in a certain ratio. What's left over is just ejected untouched (though high temperature and presence of other molecules make other nasty molecules). Now it just happens that it takes about 14.7 part of normal air (which contains a fairly consistent percentage of oxygen) and 1 part of gasoline to make a perfect reaction with no extra oxygen or extra HC molecules left. Having a perfect reaction is good because it makes your catalytic converter work better and also your exhaust fume cleaner. It's also important for other factors such as the amount of power your engine makes.

So now to the sensor. What an oxygen sensor does is that it's has chemical elements (i.e Zirconia, Titania) and catalysts (i.e. Platinum) that reacts with both oxygen and HC. And the elements are designed in such a way (and yes, there are several designs out there) that they affect the way ions (molecules that carries an electrical charge) travel in the chemical elements. Now when extra oxygen or HC is present in the gas, the sensor will generate these ions due to chemical reactions. If you put 2 wires across the direction of travel of these ions, you can actually measure a change in either voltage or current because the ions are carrying a charge. And that change is sensed by the ECU to determine if it has injected too much gas or too little gas previously into the engine for the given amount of air. It will then follow to make the correction in the future.

If you notice, the computer really doesn't control how much air is going into the engine. Your right foot does the controlling (unless TCS decides otherwise :eek:). The computer just measure the amount of air either by using a MAF (measures flow) or a MAP(measures pressure) sensor. And given the fixed size of the throttle body, the MAF and MAP will give the computer a signal proportional to the amount of air going in. Then it decides how much gas to put in via the injectors. The oxygen sensors are primarily used to correct any variation in the air and gas mixture in a closed feedback loop.

Hope that helps. If you really want the details, search in the US Patent Office web site, there are tons of patents on oxygen sensor technologies.

Eddy
 
For 87 octane gasoline, the most efficient A/F ratio is 14.7:1. For most power at this octane rating, the ratio is 12.5:1 (richer mixture). Higher octane fuels have different numbers for these "first derivative = zero" points on efficiency and power graphs.

Basically, the A/F meter is measuring the amount of molecular oxygen in the exhaust stream, and the ECU compares that reading to the amount of molecular O2 coming in the intake. The intake O2 is measured as a standard 20.9% of the incoming air read by the MAF/MAP. If the amount of O2 measured by the A/F sensor (in ionized form) is approaching the amount measured by the MAF/MAP, then the engine is running lean. If the reading is below the ratio allowed by the ECU (the ratio determined by the expected fuel octane), then the engine is running rich.

This is my take on the A/F operating mechanism.
 
Thanks !!

Eddy - I knew you were a pretty sharp guy, based in your deep conversations regarding Motec with sjs. Thanks for the reply. Actually, thanks to all for the good feedback.
 
NSXT - from where, is the source of graph number two? Graph number one I have seen several times, as it relates to a naturally aspirated engine.....is #2 specific to forced induction?
 
Andy, I have seen you post that a SC engine needs a richer mixture, then even 12.5. Why does a supercharged or Turbocharged engine require a richer mixture?? Is it just to cool the pistons?
 
I thought they were from two Bosch fuel injection and emission control manuals, but I cannot seem to locate them. I do have PDF versions of other Bosch manuals which contain similar Lambda sensor data, if you are interested.

These graphs are not specific to FI or NA engines. The O2 output does not care how that air or fuel got there, only the ratio. The engine power output graph is also just chemistry, and not induction specific. FI or NA, maximum power comes at 12.6 A/F, maximum efficiency happens at 15.4. Keeping away detonation is a different issue.
 

Attachments

  • page8a.jpg
    page8a.jpg
    63.9 KB · Views: 530
Andy, I have seen you post that a SC engine needs a richer mixture, then even 12.5. Why does a supercharged or turbocharged engine require a richer mixture?? Is it just to cool the pistons?

This will vary from engine-to-engine and from person-to-person, but from my readings and discussions, under boost the AFR should be between 11.8:1 to 12.2:1. I personally adopt an upper limit of 12.5:1 because I have found it to be too difficult to maintain a 0.4:1 variance.

A more rich AFR will result in cooler piston tops; however, too rich and the engine will experience long term cylinder wear because the excess gasoline is diluting the lube oil.
 
Although not directly related to our situation it is worthg noting that if you change the fuel, such as running methanol, then all the ratios for stoichiomentric, max power, etc. also change. I'm not certain how much difference today's typical 10% ethenol mix makes but probably not much.

I'm not sure that direct "piston cooling" as such is a result of a richer mixture. The idea it to prevent pre-ignition and/or detonation. A lean mixture does generate more heat which in turn heats everything in the combustion including piston tops, valves, spark plug etc. so in that sense more fuel means those things run cooler and are less inclined to trigger ignition at the wrong time. But perhaps that's the long-winded version of what you meant. I believe that a leaner mixture is also inherently easier to ignite and therefore at greater risk.
 
Oxygen sensors are in the exhaust pipes; therefore, how is it that they monitor the volumetric ratio of air-to-gasoline which is in the intake side of the engine? If by inference, this implies that the sensor controller has an algorithm which assumes a combustion efficiency of X%, no? I guess that the sensors are actually looking for a combustion byproduct, and by stoichiometry, deduce the theoretical air-fuel ratio? If so, what parameter and equation is employed?

I have an answer to your question. This is how Dynojet does it on the Motorcycle dynos.

The O2 sensor puts out a voltage, 1 volt equals 14.7 -1 AF ratio
.9 volts is equal to 14.7 times .9 or 13.36-1 AF ratio.
1.1 volts is 14.7 times1.1 or 16.33 -1 AF ratio
Yes different engines have different efficiencies, the number really is not that important. They tune for highest HP, not always looking for the AF number is is just a baseline or starting point.
 
Last edited:
Arata said:
I have an answer to your question. This is how Dynojet does it on the Motorcycle dynos.

The O2 sensor puts out a voltage, 1 volt equals 14.7 -1 AF ratio
.9 volts is equal to 14.7 divided by .9 or 16.33-1 AF ratio.
1.1 volts is 14.7 divided by1.1 or 13.36 -1 AF ratio
Yes different engines have different efficiencies, the number really is not that important. They tune for highest HP, not always looking for the AF number is is just a baseline or starting point.

I don’t' think that really answers the question. It also sounds like you are describing a standard "narrow band" O2 sensor such as found in 99.99% of all cars and cycles. That sensor is essentially an on-off switch centered at 14.7 and flips between 0.1 and 0.9 volts on either side with very little useful resolution beyond that and virtually none in the prime power tuning range from 11.8 - 12.8. (As shown in the first graph above.)

Wideband sensors do not put out a simple voltage like their inferior cousins, which is why they require a special controller to interpret the "signal" and translate it to a linear voltage output. But that output and be anything at any point depending on how it will be used. So the controller may be programmed to put out a 0-5v signal with 2.5v at 12.5:1 so hat it is "understood" by a particular dyno, or set for a 0-2v range for a specific ECU.

But in the final analysis in all cases yes, a voltage is converted into an A/F ratio. The difference is how they get there and what resolution the provide.

Of course, this didn't really answer the question either. :) I've got a nice long paper on the subject if you're really interested but a quick search of the web will yield the same thing.
 
Somehow with our inferior O2 sensor we can make maps and hold AF ratios well within a range of 12.5 to 12.8.
We can make the maps quickly make them at 250RPM ranges, of course if we only reved to 8000RPM we would make it every 100RPM, make them without step tests, without holding loads forperiods of time, without causing large engine temp changes.
Are you familar with the Tuning Link software??
I do not believe there is anything in the automotive world like it, there should be, and I believe there will be in the not to distant future.
Here is what a finished map looks like, looking at the numbers.
The map goes all the way to 16,000RPM, this shows a portion.

CBR600RRmap.jpg


Info on Tuning Link Tuning Link
 
Arata,

I must be missing something. I went to the link and reviewed some of the info. In a sample Screen shot of the User Interface they show a fuel mixture of 13.2 - 1. My question is how do they read the mixture?? Are you saying a standard 02 sensor is giving this data.

Sorry,, I am just struggling with this idea. As Sjs states the standard 02 sensor is really only a switch, you need a wideband to get accurate data like that.

Please fill us in. Great discussion.

Thanks,
LarryB
 
I don’t' think that really answers the question. It also sounds like you are describing a standard "narrow band" O2 sensor such as found in 99.99% of all cars and cycles. That sensor is essentially an on-off switch centered at 14.7 and flips between 0.1 and 0.9 volts on either side with very little useful resolution beyond that and virtually none in the prime power tuning range from 11.8 - 12.8. (As shown in the first graph above.)

The Dynojet O2 sensor is a wide band sensor, I am told that MOST late model cars also use wide band sensors, they do vary in quality. Most Motorcycles do not use a closed loop system, and do not use any O2 sensor, there are some that do, some Buells, and some Calif model bikes.

When using thge Tuning link Software the fuel rato target is can be pre set for any AF ratio, it can be set for a number of different AF ratio, this is usally dependent on throttle position and RPM.
usally low throttle, low RPM settings are a little leaner then high RPM large throttle openings.

I do not do any SC or Turbo FI work, again I am told the set up the AF ratio on the rich side, 12.0 to 1, as a safety measure as it does not make as much power being that rich, but when you have a turbo say on a Hayabusa 1300 and it is already making over 350RWHP, if you need more you can increase boost pressure.
 
Arata said:
The Dynojet O2 sensor is a wide band sensor, I am told that MOST late model cars also use wide band sensors, they do vary in quality.

You were told wrong, and it has nothing to do with quality. Wideband and narrow band are two totally different animals and with the exception of a few limited production vehicles all cars use the narrow band type. I fully expect that to change in the future if cost and reliability can be controlled and emissions standards continue to tighten.

However, my prior remarks started by stating that your description "sounds like you are describing a standard "narrow band" O2 sensor ". The Dynoject almost certainly does use a proper wideband or the operator is ripping people off. The rest of my post was an attempt to clarify the difference between the types and that your statements of how they worked were not representative of widebands but that unlike the narrowbands they can be tailored for any output to match the desired equipment they feed.
 
Here is an example of a dyno run at 100% throttle showing RWHP (SAE) and AF ratio.

ZX636AFratio.jpg



SJS, the quality I was referring to was between wide band O@ sensors, not between Wide band and narrow band.

SJS, what type of O2 sensor does a 98 Cadilac Catara use??
Wide band or Narrow??
 
Catara uses a Bosch LSU4 sensor..

Regarding "quality" of wide band O2 sensors, it's hard to say since I have yet to see a paper available to public stating lab test results of O2 sensors from NTK or Bosch. And when someone claims accuracy over a particular sensor, I would be asking that they give a report on lab tests done using controlled gas tests on the sensor.

One of the reason why NTK UEGO and Bosch LSU4 sensors can be "off" sometimes is due to the fact that they require calibration. When new their calibration values are represented by the calibration resistor that comes with the sensor. Both Bosch and NTK does that. But as sensors age, their calibration changes. Given a typical installation in production cars, there is no way to recalibrate the sensors unless the sensor is removed and air calibrated. That's why all these sensors have a rated life of about 50,000km under normal conditions.

Recently, I was researching around and found that Denso/Toyota has came out with a new wide band sensor (they called it "Air Fuel Ratio Sensor") used in late model Toyota, Honda, Subarus, etc. Unlike the duel cell (one being the pump cell) design, it's a single cell design (which means it uses 4 wires instead of 5). And also inherent in the design is that it's a current limiting device and it has the property that it flows no (0)current when the gas mixture is at stoichimetric. This sensor eliminates the need for calibration all together. The only thing it will do when the sensor gets old is that your sensor response time becomes slow or the sensing dynamic range is narrower; just like a narrow band sensor. The catch with this new sensor right now is that it's not supported yet by any aftermarket sensor controllers or ECUs, though the electronics is fairly simple given that a microcontroller is used. However, it's being used widely by car manufacturers to meet the strict emission standards.

One good thing with this sensor is that it can be operational within 10 seconds of turning on the ignition switch versus 20-30 seconds for the Bosch and NTK sensor. This was one of the requirement needed to meet the strict emission limits when cold starting the car.

Eddy
 
I see that Eddy has already answered the question about the Catara sensor, not that it is in any way relevant. The vast majority of cars still use the narrowband type, which is a far cry from a flat statement that "MOST late model cars also use wide band sensors". I didn't mean to be indignant or flame you, I was just trying to point out that most "late model" cars do not have wide band sensors. From what Eddy says some are starting to adopt a new variant but those are brand new and hardly amount to "most". No big deal, just trying to clear it up before other people repeat the error and say "I read it on NSX Prime". ;)

Eddy was also correct about one of the reasons why the typical widebands are not widely used. And free air recalibration was a deciding factor in my purchase of the Motec PLM because no other affordable unit at the time offered it, including FJO. The new sensor design is interesting and sounds like the design I commented on in a thread last year, but it wasn’t even in production yet at that time so it is very new. It sounds geared to OEM uses where traditionally the only value that really counts is 1 Lambda, but so long as it starts passing current predictably as the mixture gets richer, then someone will build a controller to translate that into a linear voltage output. But as I also said earlier, I expect most cars to be so equipped eventually and to use them for more than just stoich to control low load emmissions. Perhaps this is the start.

BTW, I didn't see anything special in the Tuning Link screenshot. That would be typical of the software for any decent ECU such as the AEM I use, except I hope that isn't the primary fuel map because you wouldn't want it based on throttle position. Surely there is one based on MAP or MAF. Perhaps not for cycles. The software that comes with a Motec, AEM, etc. has countless fuel and timing maps and trims to take into account everything from barometric pressure to coolant temperature.
 
Back
Top