a question about oxygen sensors

if complete combustion happens at 14.7:1 and the motor is running at stoich, where is the oxygen coming from that the O2 sensor is using to produce any voltage/reading? at stoich shouldnt all the available oxygen and fuel be burnt? hence complete combustion? same for rich conditions, wouldnt the motor burn up a 14.7:1 ratio and exhaust leftover fuel?

G'day Collin.

Wideband O2 sensors are pretty complicated in how they work. Essentially, they are a narrow-band sensor, and a solid state electronic oxygen pump in the same package. The Wideband sensor samples the exhaust gas, and then uses the oxygen pump to either remove or add oxygen until the exhaust mixture reads as it would for stoich combustion. This is all done by the wideband O2 sensor controller using a PID control loop in real time. The amount, and direction of current being applied to the oxygen pump to get the integrated narrow-band sensor to read stoich is directly proportional to the AFR the engine is currently running... The solid state oxygen pump typically draws in air from the wiring harness connection.

Hello Zac,

Thank you for your knowledge. Only one thing I am wondering... at 14.7:1 when they say complete combustion happens, what exactly does that mean? complete compbustion.. does all the fuel and oxygen break down during combustion? I guess there's some oxygen molecules left in tact after the combustion process?

The stoichiometric fuel ratio, is by definition where you have exactly the right number of fuel molecules to oxygen molecules, so that when all molecules are broken down into their constituent atoms (Hydrogen, Oxygen, Carbon) there are enough fuel atoms to combine with exactly the right number of oxygen atoms with none left over.

For example, we can look at the simple combustion of Hydrogen in pure Oxygen. This will make the chemistry somewhat simpler than fuel and air, which contain a whole bunch of other chemical elements). If you have two Hydrogen atoms and one Oxygen atom, and you combust them, you end up with a single water molecule. At a slightly higher level, if you have 2 Hydrogen *molecules*, each consisting of two Hydrogen *atoms*, and you have one Oxygen molecule, consisting of two Oxygen atoms, and you burn this mix, then you will end up two water molecules with nothing left over. This is a stoichiometric ratio. 2H + 1O = 1H20.

If you have more Oxygen or more Hydrogen, then you end up with left over fuel or Oxygen, and this means you are correspondingly rich or lean.

This chart shows the relationship between AFR and oxygen concentration.

I think the question here, is how can you measure engine charge 'rich-ness', if there is no oxygen in the exhaust to measure, as that's what an oxygen sensor measures? I mean, you can't measure how much oxygen _isn't_ there, after its all been converter to H20, C0, NOx and CO2?

I suspect the answer is something to do with how the zirconium dioxide element in the oxygen sensors works... I'm no chemist, but I suspect there much be some concentration effect going on, ions moving across the element trying to reach some sort of equilibrium that generates the voltage output of the narrowband element?

I hope there is someone out there with more intimate knowledge that can chime in, I'd love to know more about what is actually going on. Till then, I'll just trust all the anecdotal evidence and trust that when my wideband tells me I'm at Lambda 0.8, I really am :-).

For our purposes there's always some left over oxygen because combustion is never fully complete. We can use an emissions bench, a Bosch type wideband O2, and a scatterplot to show correlation between oxygen concentration. See attached files. One is from the technical product information of the Bosch LSU 4.9 and shows the correlation between O2% and the signal current.

The AFR vs O2% is a scatterplot I made based on engine dyno results using E10 fuel. The bottom plot has emission calculation of AFR sampled from both banks of a 2 bank engine ("AFC") and the O2 concentration also from the emission bench. The top plot has the average of 2 Bosch LSU 4.2 sensors' AFR calculations from each bank.

Great data Ray, cheers for that, in particular this golden nugget:

The upper graph shows that there is no o2 left in the exhaust gas if the charge mixture was stoich, or richer. The Ip current is zero when there is a zero percent concentration of o2 in the exhaust gas sample. The lower graph shows an Ip current of zero corresponds to a lambda value of 1.0 for the intake charge, stoich, so combustion was complete. This doesn't answer our question of how can we measure how much o2 _isnt_ there, to figure out how rich our intake charge was?

I've resorted to wikipedia:"

The sensor does not actually measure oxygen concentration, but rather the difference between the amount of oxygen in the exhaust gas and the amount of oxygen in air. Rich mixture causes an oxygen demand. This demand causes a voltage to build up, due to transportation of oxygen ions through the sensor layer. Lean mixture causes low voltage, since there is an oxygen excess."

Is this why oxygen sensors have a pretty high minimum operating temperature maybe? The oxygen ions are drawn across the sensor element, reacting with an exhaust gas sample resulting from a rich intake charge, and being consumed? The richer the intake charge was, the more demand for oxygen, the more ions that move across the sensor element, so the higher the sensor voltage? Hell, it might not even be unburned fuel in the exhaust gas sample that consumes the oxygen ions, it might be carbon dioxide (CO2) forming from carbon monoxide (CO) or something...

The current changes directions when the sensor is rich of stoich. That's why it's negative current. The sensor either pumps oxygen in or out depending om whether it is rich or lean of stoich. The amount of pumping and the direction indicates the AFR.

Also, the other scatterplot shows that there is never 0 percent oxygen in the exhaust as an overall average.

We agree that a wideband sensor doesn't actually output any current though, right?

The wideband sensor controller drives current through the oxygen pump to achieve a 450mV output from the narrowband reference cell within the wideband sensor. The direction, and level of current it needs to supply to pump enough oxygen into, or out of the exhaust gas sample to get a 450mV output can be directly related to the AFR of the intake charge. I've used a couple of ECU's with integrated wideband controllers where you could adjust the gain settings of the control loop to suit different sensors and applications. I never actually experimented with changing the default settings though.

For your scatterplot, how did you measure the O2 concentration in the exhaust?

Love the discussion dude, its good to geek out once and a while :-).

The o2 concentration in the scatterplot is measured by a lab grade emissions analyzer.

Interesting. I find it odd that if wideband sensors rely solely on the O2 concentration in the exhaust, their accuracy doesn't drop off hugely once they're reading gas resulting from an intake charge richer than around 12.5:1, as according to your scatter plot, there is relatively little difference in O2 concentration richer than that...

We've just got a 5 gas analyser hooked up on the dyno, so once this wiring course is locked down, I can hopefully have a play and learn something :-).

The newer wideband sensors like the LSU 4.9 and the LSU ADV are more accurate in the rich range.

Great info here guys, I had always thought that wideband sensors directly measured the O2 in the exh gas stream - and that logic broke down under rich conditions and this has really helped explain things - thanks!

It was great discussion guys, one of my headache solved.

thank you