[Diy_efi] Fw: wide-band O2 sensor comparison

[Garfield Willis]

I've posted this on efi332 also, because the pukes-that-be on diy may
block this from being posted over there.

On Fri, 11 Aug 1995 00:34:45 -0400, "Bruce" <nacelp at bright.net> wrote:

>> The readings at the two conditions agreed within less than half an
>> air/fuel ratio.
>> Idle in neutral:
>> ETAS WB-O2
>> 11.05 1.6 (10.69)
>> 12.05 1.95 (11.96)
>> 13.1 2.26 (13.3)
>> 13.5 2.35 (13.8)
>> 14.6 2.48
>> 16.1 2.63
>>
>> Brake torque to ~1200 RPM:
>> ETAS WB-O2
>> 11.05 1.58 (10.65)
>> 12.0 1.86 (11.5)
>> 13.0 2.16 (12.8)
>> 13.45 2.29 (13.57)
>> 14.55 2.46 (14.4)
>> 15.75 2.58

>Like I have said from the beginning, not a Lab grade devise, and for the
>average mechanic just wanting to get close, it works for me.

And like I have been saying from the dweebOmeter's beginning, you
apparently haven't even a clue what an iceberg lies beneath this one
little piece of data. You go try those tests on a large lot of sensors
under the same conditions, and that whopping 0.5AFR error you see above
is going to show up on the *OTHER side of actual* as well. What you're
seeing is about the norm of skew on ONE side of the data. Since you've
only tested with one sensor, you're seeing the error off of actual for
just one case; your particular sensor & circuit reads richer than actual
by 0.5AFR, another will do the same to you only read leaner than actual
by about that same amount. Over a large lot of sensors, you're going to
see some measuring 12.5 when the actual is 12.0, which means you'll see
a nominal error band of a WHOLE freaking AFR. Actual 12.0AFR: dweeb
range of readings (11.5-12.5). You bet your ass this isn't "lab grade".
It isn't "commercial grade", it isn't "instrument grade", it isn't even
"hobby grade", it's not ANY "grade"; all you can say is it's just
"horse-shoes close".

Now I hope people will consider what this means, when as I've been
trying to point out, you try to compare YOUR AFR readings with someone
elses, or on the dyno, or readings you took yourself with one sensor,
and are now using a replacement. In the most critical range around the
12s, when one person says they're reading 12.5, another may get 12.0
another 13! Switch sensors and you yourself can't even compare notes
with your own previous readings.

That's the point about accuracy. If you don't have it, you're pissin in
the dark. I've seen Bruce posting about how his motor "likes"
12.2-12.3AFR or some such, as if he could achieve a 0.1AFR accuracy.
=46act is, he doesn't actually know if that's 11.7AFR or 12.7AFR without
an ETAS to compare it with. Swap out the sensor, or use another dweeb,
and YMMV over a range of a whole stinking AFR.

Our own massive gas-bench testing on over 60 sensors with two different
dweebs also shows about a 1+AFR slop (error band), due partly to circuit
design flaws and partly to calibration errors. Do some more testing, and
I won't have to even spend the time formatting and graphing our own pile
data to prove the thing's a POS. You'll have proved it yourself. All you
guys had to do was do this testing with a decent lot size of sensors, to
see the error spread. Pretty late in the game to be waking up and
smelling the coffee.

Like I've said, you're NOT measuring AFRs with this device; you're
measuring a scale of arbitrary (and at least hopefully monotonic)
numbers that say "richer" or "leaner" from stoich, with some greater
accuracy & precision than a NB sensor to be sure, but NOT accurate
enough to actual call those numbers AFR measurements.

QED,
Gar


_______________________________________________
Diy_efi mailing list
Diy_efi at diy-efi.org
http://www.diy-efi.org/mailman/listinfo/diy_efi