Oxygen sensors

[Dale Ulan]

> ok, so if its cold the output voltage stays between, say, .3V and .6V
> for an extended period of time. Once it exceeds these thresholds, we
> declare it's hot enough to use.

Something like that, anyways. There are a couple timers in there
as well.

> If the amplitude primarily indicates temperature, wouldn't a
> correction based on expected temperature have to be applied *before*
> a PID controller? I imagine that is, in part, what the tables
> do. Can you give us a better description of the tables?
> 

***************************************************************************
*
* Airflow dependant INTEGRATOR rate
*

KO2INTEG1       time12  180
		time12  110
		time12   79
		time12   55
		time12   40

 
KO2MEANTRIP
 db     $64,$60,$5E,$5E,$5E             ;Mean trip value

KO2RICHSTEP
 db     $73,$68,$5E,$5E,$5E             ;rich step trip

KO2NORMTRIP
 db     $5A,$5C,$5E,$5E,$5E             ;centre step trip

TO2AFFILT       ;airflow filter to slow ramp of O2 sensor
 db     $07,$0A,$10,$16,$1A

****************************************************************************
*
* Differentiator step size based on O2 osc size
*
*

TO2PK           fac256  47      ;
		fac256  47      ;
		fac256  47      ;
		fac256  47      ;
		fac256  51      ;
		fac256  63      ;
		fac256  63      ;
		fac256  71      ;
		fac256  86      ;
		fac256  110     ;
		fac256  125     ;
		fac256  141     ;
		fac256  157     ;
 

****************************************************************************
*
* Maximum allowed cross time. If the sensor is slower, than try to
* correct by using the INTEGRATOR. Uses O2 osc size as a scale.
*

TO2CROSSTIME
		time12  10
		time12  16
		time12  24
		time12  24
		time12  32
		time12  40
		time12  64
		time12  94
		time12  188
		time12  376
		time12  752
		time12  1876
 
****************************************************************************
*
* Oxygen Sensor expected cross-time adjustment for airflow
* 

TO2CROSSTIMEADJ ;airflow adjustable O2 cross-time
		time12  0       ; 0 Counts
		time12  0       ;10 Counts
		time12  10      ;20 Counts
		time12  10      ;30 Counts
		time12  0       ;40 Counts

****************************************************************************
*
* Decay rate of the D section of the PID controller (max: 1.000)
*
*

TO2IRATE        fac256  137     ; 0 Counts
		fac256  176     ;10 Counts
		fac256  273     ;20 Counts
		fac256  313     ;30 Counts
		fac256  375     ;40 Counts

****************************************************************************
*
* Integrator Time in ms.
*
* indexed by O2 osc size This is the integrator holdoff time.
*

TO2INTEG2       time12  2000
		time12  2000
		time12  2000
		time12  1750
		time12  1500
		time12  1250
		time12   876
		time12   750
		time12   750
		time12   876
		time12  1500
		time12  1500
 
****************************************************************************
*
*  O2 trip voltage adjustment
*

TO2ECTTBL       ;voltage effects vs. engine temperature
		db      $c0             ;header

		db      0               ;HOT            Engine Temp
		db      0               ;138.5 C
		db      0               ;108.5 C
		db      0               ;92.0 C
		db      4               ;80.5 C
		db      4               ;71.5 C
		db      4               ;63.5 C
		db      4               ;56.5 C
		db      4               ;49.7 C
		db      4               ;43.7 C
		db      4               ;37.3 C
		db      4               ;31.0 C
		db      0               ;24.0 C and colder



> What do you think of the approach of just hard-limiting the sensor
> output once it warms-up and integrating?

It'll work but won't be that good.

> Do the PID coefficients look like it's dominated by I?

Actually, the I section is kinda wierd, as the coefficient is 0 until
the D section has timed out. This time is in the order of 2 seconds,
as in the table above...

PID controllers become wierd when you start putting holdoff timers
all over the place. I can't simulate this very easily in Matlab, but
I certainly can run it on my HC11 board and figure out what's happening.

It looks as though the 'D' section is the dominating factor. As far
as I can tell from the source code, anyways. I don't think it's legal
for me to post source code in here, though.  The source I have is disassembled
from ROM, and any comments are those that I figured out...

-Dale