Return of the diy_EGOmeter (YES, it's finally ALIVE!!)

[Gus Cameron]

At 13:34 23/04/98 -0700, you wrote:
>On Thu, 23 Apr 1998 12:28:51 +0100, Gus Cameron <cameroa at sbu.ac.uk>
>wrote:
>
>>Hmm, well, no-one really picked this one up and ran with it.  So.... I took
>>a walk across to the library and Lo! Seek and you shall be rewarded etc.
>>etc.  
>>
>>If you check out 
>>http://www.bracken.co.uk/misc/
>>you will see a figure relating % O2, CO, H2, NOx etc to Lambda.  It helps
>>to explain why the outputs from lambda (O2) sensors are non-linear and also
>>partially explains the differing curves that Garfield was reporting either
>>side of equivalence.
>>
>>Sorry if this was common knowledge already.
>
>Geez, no apology needed on this end, that's stellar stuff you found
>there. Makes me think I might wanna get the book that chart came out of
>too! The Horiba box also has a calibration procedure you can follow if
>you have known concentration O2 (span/model/sample) gas mix (term you
>use apprarently depends on what industry you're working in) with a
>prepared concentration (4% is recommended, but you can manually
>calibrate the box using anything known, between 0-25%).
>
>And Gus, your conjecture on the shape of the O2 curve in that diagram
>explaining the diff. shape on either side of stoich for the sensor IS
>dead on target, cuz the shapes of them curves on either side, EXACTLY
>mimic what you see in the NTK published 5-wire sensor curves in their
>SAE articles. I'd say you've really cleared up a couple of voids in my
>understanding from just that one graph. Nice sluething, dude. We oughta
>send you out with the black bag more often. B)
>
>Gar
>


and John R Bucknell wrote:
>Just for giggles, anyone tell me what the definition of lambda is?  I
>read somewhere that lambda is the excess air factor, i.e. lambda of .98
>is just a tad (2%?) rich.

John, you actually hit the nail dead centre, except 0.98 is weak, not rich.
Lambda is derived from the stoichoimetric air:fuel ratio.  This means that
you have just enough air to burn all the fuel, with no air left over.  i.e.
the fuel and air combine to form CO2 and H2O.  But the amount of air
required to do this depends on the fuel you are burning.  
I can give you the stoichoimetric equation for glucose no probs:  C6H12O6
(glucose) + 6H2O ----> 6CO2 + 6H2O.  i.e. for every one molecule of glucose
you need six molecules of oxygen and you get 6 molecules each of carbon
dioxide and water produced.  Now, the same principle applies to petrol, for
every one molecule of petrol you need a certain number of molecules of
oxygen to burn it completely.  Air is always taken to contain 20.95% oxygen
in case you were wondering, but you must take the air pressure into account
if you are to correctly calculate the number of oxygen molecules per cubic
meter.

Unfortunately petrol (gas?) contains a real complex mix of different type
of hydrocarbon molecules, and not all petrols are the same, so the actual
stoichometric values are harder to calculate and differ from one another
anyway, but are normally in the range 14:1 ---> 15:1, with 14.7:1 most
often being quoted.  These values are based on mass (weight) of air and
fuel, i.e. for every 14.7kg of air you need 1kg of fuel, but the principle
is the same.  For actual calculations see the book referenced at
http://www.bracken.co.uk/misc.

Now Lambda is simply the stoichoimetric A:F ratio divided by the observed
A:F ratio.  i.e. if the stoichoimetric A:F ratio is 14.7:1 and you divide
this by the observed AR ratio of 14.7 you get a lambda of 1.0, sometimes
refered to as equivalence.  
If the observed A:F ratio is 18:1 the lambda is (14.7/18) = 0.82, and the
mixture is weak (too much air).  If the observed A:F ratio is 12:1 the
lambda is (14.7/12) = 1.23, and the mixture is rich (too much fuel).

Now, and this is why I asked my original question, if the definition of
lambda = 1.0 is that all the oxygen is burnt, how can an oxygen sensor tell
you what the lambda is at values greater than 1.0?  You can't get negative
concentrations of oxygen!  The answer is in the graph, lambda is not a true
equivalence ratio: there is always a small amount of oxygen left over.
However, as you approach lambda for the weak side the O2 concn. drops
rapidly, after which it takes on an almost steady value, meaning that the
O2 sensor output is going to be much less sensititive on the rich side than
on the weak side.  Really what you want is a O2 meter to tell you if you
are running weak and a CO meter to tell you how rich you are running (again
see graph).
Interestingly hydrocarbon (HC) concn. does not vary much with lambda, as I
assumed it would.

Gar - the Horiba calibration routine that you refer to is what we do with
the paramagnetic O2 sensors.  In fact I could use a paramagnetic sensor to
calibate the DIY_EGO zirconium O2 sensors.   But do tell, is the curve
linear all the way up to 21% O2?  Can you give me a refernce on those SAE
papers?

byeee,
gus