O2 Sensor Open/Close Loop

[Scot A. Sealander]

kleenair at ix.netcom.com wrote:

 [Big snip]

> Regarding what OEMs
> do on late model vehicles, I had a 1996 3.8L Buick on the dyno, which
> maintained its closed loop operation at WOT.

 How do you know that it maintained closed loop?  Scan tool? Don't believe   
 it!  In the GM code I have seen, they leave the closed loop flag on, even 
 when it is in PE.... and that is most definitely not closed loop.


> However, their motive may
> have nothing to do with peak power.  That may just be a way of protecting
> the catalyst by reducing excess fuel in the exhaust during extended high
> engine load periods.   
  Maybe, but I personally doubt it.

> Also, just because the system stays in closed loop does not necessarily
> mean actual A/F is 14.7:1.

  Well, in the normal GM scheme of things, that is untrue.  I have the       
  feeling that open/closed loop operation is not well understood....
  though I have not looked at any newer code.  My experience stops at about 
  the 1993 model year or so.....

> If the system is flexible enough, you can
> make the Rich to Lean transient slower than Lean to Rich transient, in
> effect raising the average O2 sensor voltage. 

  Is this in reference to the normal dithering of the O2 volts in closed     
  loop?  Making the controller asymetrical?  Some GM systems are that 
  flexible.  The Buicks I have seen are not so flexible.


> Most systems can also use
> a different target O2 voltage based on MAP and RPM.

  Now I was under the impression this was because of the O2 sensor       
  temperature (and the temp effect on O2 sensor volts), not because of a     
  different desired AFR.... The sensor is not very useful very far from 
  stoich, so I don't think trying to stay in closed loop would buy you 
  anything.

 
  I have been meaning to make a decent scan tool to watch the O2 controller  
  at work in the ECM under actual driving conditions.  I have completely   
  worked out GM ALDL protocol, but am not really a programmer. :-(  This 
  makes it hard to build a scan tool. 


  Scot Sealander    Sealand at clarityconnect.com