O2 sensor question

[Garfield Willis]

On Wed, 22 Mar 2000 13:18:18 -0700, "Chad" <chad at dynojet.com> wrote:

> The new model Cadillac Catera uses a wide band O2 sensor.  This particular
> sensor needs a fairly sophisticated control loop to make it work though (I
> haven't worked with many others yet).  I've seen it done analog, a bosch
> unit we had in used opamps, but we decided to go digital.  We used a
> microcontroller to control the heater (it has to be the right temperature to be accurate),
> check the O2 level, and communicate with the A/F display (in our case, a PC).

I've just been ruminating about this, and I wonder if this isn't another
*variant* of the older LSM-11 style showing up once again. I know that
Bosch does have a current-pump sensor (the "LSU"), but didn't think it's
been deployed in cars yet. It's used in Bosch's LA-3 Lambda Meter, tho.

BTW, these older types of sensors are referred to by Bosch and others as
"Nernst" type (sometimes also described as "switching" type if a generic
ordinary, because of their high non-linearity and an ECU's normal use of
them to merely find the stoich "switch-crossing"), to distinguish them
from the current-pump styles like the LSU and the NTK's, which have both
a Nernst cell and a pumping cell; hence they are referred to by their
possession of the additional O2 pump).

So, do you remember Chad how many wires the sensor you mention above
had, or if it had an O2 pump? If just the 4, that could explain why the
careful control of the heater, because ...

One thing I don't remember anyone pointing out/observing, is that this
sophistication for the heater *control* is needed because the Bosch
heaters all seem to have negative tempco's, meaning the resistance of
the heater core goes DOWN as the temp goes up, so things can get carried
away without tight controls over the heater current. The NTK-derived
sensors (at least the current-pump styles) all have heaters with
positive tempco's (resistance goes UP as temp goes up), so they are
inherently self-regulating to a large degree. This is why the NTK papers
show that even with a +BATT swing from 10V to 15V along one axis, a
600-950 degC tip temp swing along the other axis, still the change is
within +-0.15 AFR. By way of contrast, the same schmoo of variables on
the Bosch 4-wire "wider-band" sensors (non-current-pump style) would
give you at least a couple WHOLE AFRs of variation. The heater
resistance alone goes down by over an order of magnitude between 650-900
degC on the Bosch's, this according to their own specs on the LSM-11.
Without factoring in this temperature dependance during measurements,
they'd be useless on the rich side, where this effect is strongest with
the LSMs. Their specs show without this temp compensation, over the
range 650-900 degC, the temp effects alone would completely swamp the
full scale measurement range (from stoich to say 10AFR). So anyone
building meters or controllers around these sensors per force MUST
include both temp compensation and heater control. This had been
discussed a year or so by Frank Parker, it's just that I don't recall
the issue of "the tempco's being negative" being a factor in needing
both temp compensation AND tight control of the heater, ever being
mentioned.

Thankfully, neither of these steps is required for the NTK-derived
sensors. You CAN apparently cut down the +-0.15 AFR slop significantly
by regulating the heater voltage, and some AFR instrument companies do
this, but it seems like way overkill to attempt to achieve greater than
1.0-1.5% accuracy. Too many other factors, including the electronics,
begin to intrude. Even the LA-3 meter's specs claim no more than a
+-1.5% "measurement tolerance" as they style it.

It's somewhat surprising (if my surmisal above is correct) that both
OEMs and aftermarket meter makers may be trying to get the cost down, by
opting for using these older type sensors. But the way NTK had locked up
the meter market with their exclusive arrangement with Horiba, coupled
with Bosch being VERY late in developing their own current-pump style
sensor (Honda-NTK deployed theirs in autos in '90, Bosch didn't have the
LA-3/LSU ready until '97 IIRC), this could be more a business than a
technical decision. One other thing to mull over, tho. The Nernst-style
sensors, at least the LSM-11, would appear to maybe have a higher
tolerance/lifetime in the presence of lead. In  fact, remember Bosch
marketed them first as their "lead-tolerant" O2 sensor. Horiba, for
example with their Lambda Checker box (which uses an NTK sensor)
suggests a sensor lifetime of just 20hrs! in a "poisonous environment",
which they refer to as exposure to either lead or high oil consumption.
This may suggest a certain relative fragility, if you will, with the
current-pump type sensors. Only time and field experience will tell. I
know that Frank and his motorsport team have been using NTK boxes in the
presence of lead, and IIRC haven't seen any real early mortality in
their sensors, so maybe Horiba's spec is just intended as a cautionary
to scare off, and also cover their warranty's butt.

Just some thots (and maybe exhaust gas :) along the way.

Gar


----------------------------------------------------------------------------
To unsubscribe from diy_efi, send "unsubscribe diy_efi" (without the quotes)
in the body of a message (not the subject) to majordomo at lists.diy-efi.org