Knock sensor control (and other uses for ionization gaps)

[Matthew Lee Franklin]

P. Wales wrote:
>I assume if it pinged, the combustion ionisation will be different than if
>it fired normally. Exactly how, I don't know. Interesting subject though.

Yes, one of the things they looked for was an oscillation in the ionization 
signal.  After a hard knock, the pressure waves bounce around in the 
combustion chamber.  There are several modes of this oscillation so you'd 
have to consult an accoustitian for advice here.  The ionization current 
oscillates at the same frequency.  

My memory is cloudy here but I think there may have been a little problem 
with it.  It went something like if the spark plug is off to the side of 
the chamber the knock ion signal quality would be good, but typically the 
best location for combustion initiation is as near the center of the 
chamber as possible.  And if the plug were at the center the ion current 
signal quality would be poor.

I read that CAT uses a seperate ionization gap near the "end-gas" of one of 
their lean burn natural gas models to estimate flame arrival time.  The 
"end-gas" is just the part of the charge farthest from the spark plug which 
burns last.  They would adjust the fueling rate depending on the flame 
arrival time.  My memory is cloudy here, too, so call your CAT dealer for a 
brochure.  (Maybe it was Cummins?)

Another use for ionization gaps is in the exhaust port (yes I'm not 
kidding).  This can be used to sense incipient misfire.  When a hydrocarbon 
burns, ions are formed in the combustion products.  The concentrations 
decay with time.  If the port concentrations are low, it means that the 
combustion was completed quite a bit earlier.  If they are high, it means 
that the combustion just finished or is still occuring.  Basically, if the 
ion signal is high it means that the flame speed is slow and in danger of 
misfiring.  (Of course, 100% misfire would give zero signal.)  This 
probably adds too much expense and complication to a commercial 
application, but it is fun to tinker with.  A non-resistor spark plug and 
the right biasing circuit (100V is great, but 15V workes fine) can be put 
together for less than $10.  There have been a few papers on this.  If 
anyone is really dying to know, I can look them up.  

One problem that I had was that if the plug got too hot, its ceramic starts 
to get a bit too conductive.  Its conductivity approaches and may exceed 
that of the ionized gas, completely saturating the ionization current 
sensor.

Sorry for rambling so long, but I at home with a cold and have nothing else 
to do.


Later,
Matt