O2 Sensor type?

[robert dingli]

Hi Mark and other diy-efiers

> >
> >I ran some calibration runs last week against an ADS9000 four gas analyser.
> >There is some problem with the UEGO in that the voltage output wrt AFR
> >has a 1.7V discrepency at stoiciometric and about 1.0V at the extremes
> >of my tests (12:1 and 25:1).  I'll be faxing NGK later today to find out 
> >why the output is out of spec.
> >
> Robert,
>         Are yes, but is your ADS9000 correct! Four gas analysers use a
> "chemical cell" O2 sensor and they have a life of only about 12-18 months.
> Oxygen is used along with the other gases(3) to calculate the AFR so if this
> sensor is not reading correctly your AFR will be all over the place. The
> other problem is that most "cal gas bottles" do not contain any oxygen so it
> is hard to check.

This was my first thought.  I re-calibrated the ADS9000 and found that it
was almost spot on.  The ADS9000 is one of three that we have in the lab
along with a full EPA emissions bench and various individual sensors.  It
has 2-3 separate chemical sensors for the various gases and all of these
are showing the correct values at stoic.  A switching type EGO also backs
up these readings as do the torque readings.  I am inclined to believe the 
ADS9000 is correct.

> With the "firewall mounted" controller NTK only claim +/- 0.5 AFR as far as
> I can remember.

The NGK data in front of me claims :
temperature dependency             +- 0.03V
dispersion of amplifier in TC-6000 +- 0.02V

> I would be very suspicious of it being 1.7 volts out at stoiciometric that
> does not sound correct to me. After all it only gives a 0-5 volt output so
> an error of 1.7 volts is pretty bad!!!!

Sorry, I meant 0.17V.  It shows 2.830-2.832 V at lambda one.

> Make sure you have enough power for the heater circuit I have run into
> problems with cicarette lighter plugs ie. dirty contacts, etc.

It is connected directly to the battery which gives a very steady 14.2 volts
while running.

> The early controllers could drive a meter movement directly however the
> later type which you have cannot, so make sure you are not "loading" the
> output. High impedence input or build yourself a voltage follower op-amp to
> give it some grunt!

I looked into this also.  A moving coil voltmeter gives the same reading
as the high impedance DVM.

> Also it is very sensitive you can watch each cylinder's gases coming down
> the exhaust pipe if you hang a scope off it!

That's what I'm using it for.  It's a lots more useful than waiting for
a four gas analyser to settle.

> Also do you have the correct output table ie. volts to AFR. Also 25:1 is a
> bit on the lean side of things to be testing spans. Quoted scale shows 10:1
> - 22:1 so 25:1 is really pushing it.
> 14.7:1 AFR = 3 volts.
> 10:1   AFR = 1.7 V
> 22:1   AFR = 2.65 V

I think we may be talking about a different NGK controller here.  Firstly,
the one I am using cost AUD$750 ex tax rather than the $350 that you quoted.
Secondly, the span for the controllers (of which there are three types) are
10:1 - 30:1 AFR for the normal unit (TC-6000), 14.5:1 - 52:1 AFR for the 
TC-6000C and 10:1 - 14.5:1 for the TC-6000D.  All controllers use the same 
sensor.

The TC-6000 which should give 3V at 14.57:1, 1.85 V at 10:1 and 3.45V at 22:1.
Given that our engine can run smoothly over lambda 4 (but ideally up to lambda 
2) 22:1 is a bit rich for a lean limit.

> 
>   Hope this helps,

Thanks for your input.  I'll let you know how it goes.

Robert Dingli
-- 
----------------------------------------------------------------------
             Robert Dingli           r.dingli at ee.mu.oz.au

Power and Control Systems                 Thermodynamics Research Lab
Electrical Engineering                    Mechanical Engineering
   (+613) 9344 7966                          (+613) 9344 6728
  University of Melbourne, Parkville, 3052, Victoria, AUSTRALIA
----------------------------------------------------------------------