Jump to content


Photo
- - - - -

Misfire Algorithm


  • Please log in to reply
6 replies to this topic

#1 Canuck

Canuck
  • Member

  • 2,388 posts
  • Joined: March 05

Posted 19 May 2014 - 04:48

I'm not much a mathematics smartie-pants unfortunately.  I mean, if I come across a formula and can translate the vocabulary, I can usually understand it but I'm not typically able to roll-my-own.

 

The stock Motronic 1.something in my old 5-series uses two VR sensors on the flywheel - one counts the 114 (116? Can't recall - it's an odd count) teeth and the other a single TDC(ish) pin.  I've adopted a 60-2 pulley and sensor for the pending MS2 install and have been looking at everything I could remove going forward.  However, it seems a shame to remove the makings of a perfectly good flywheel speed / accel / decel measuring system.  Combined with the cam-driven single-tooth sensor required on the MS2 for COP and something to process it all, these sensors should be able to identify a specific cylinder on misfire.  In a perfect world, it might be useful in combination with other sensors (IE per cylinder exhaust temp) to compare ignition and mixture specifics per cylinder (IE a faster or slower rate of acceleration from the point of ignition on that cylinder).

 

Does anyone have any insight into what the math might look like in the firmware?

 

 



Advertisement

#2 Greg Locock

Greg Locock
  • Member

  • 6,366 posts
  • Joined: March 03

Posted 19 May 2014 - 10:57

Have you checked on the megasquirt site?

 

 

misfire detection seems easy to me. Trace the instantaneous engine speed after each ignition event, if the crank fails to accelerate in a similar fashion to the previous one, tis a misfire. So, log the output of the toothed wheel, see if the TV is observable, then pull a lead, see if it makes a difference.

 

http://ilot.edu.pl/K... NO 1-2/R39.pdf

 

might be a good place to start


Edited by Greg Locock, 19 May 2014 - 10:59.


#3 Canuck

Canuck
  • Member

  • 2,388 posts
  • Joined: March 05

Posted 19 May 2014 - 15:11

Ah yes - fair enough.  I have a little Dataq data logger I can play with - will see what I can see.  Thanks Greg.



#4 Magoo

Magoo
  • Member

  • 3,718 posts
  • Joined: October 10

Posted 19 May 2014 - 22:04

around page 10 

 

 

http://www.motorcraf...df/OBDSM701.pdf



#5 Canuck

Canuck
  • Member

  • 2,388 posts
  • Joined: March 05

Posted 21 May 2014 - 12:34

That is awesome Magoo - thank you. Interesting to note that on heavy throttle application, some of the presented methods run into problems.

#6 Magoo

Magoo
  • Member

  • 3,718 posts
  • Joined: October 10

Posted 21 May 2014 - 18:06

That is awesome Magoo - thank you. Interesting to note that on heavy throttle application, some of the presented methods run into problems.

 

It's a snarl, but imagine not having it...which we didn't not that many years ago. 



#7 gruntguru

gruntguru
  • Member

  • 7,642 posts
  • Joined: January 09

Posted 22 May 2014 - 03:32

Similar technology can even be used as feedback to optimise things like ignition timing on the fly.


Edited by gruntguru, 22 May 2014 - 05:25.