Knock Detection

[John Dammeyer]

At 02:27 PM 06/11/1997 -0500, you wrote:
>To my knowledge, when a spark ignites a compressed fuel air mixture, a flame
>front moves across the chamber.  Behind the front is burned gases, In front
>of the front is unburned gases.  The pressure is the same throughout the
>chamber.  As the front progresses, the pressure rises.  If the flame front
>gets to the end, everything is OK and the chamber is filled with very high
>pressure gas which pushes the piston.  But if the temperature of the
>unburned gases rises above the autoignition temperature of the fuel the
>remaining unburned gases will spontaneously explode before the flame front
>gets there.  This is detonation.  The explosion causes the combustion
>chamber to "ring" like a piece of metal struck with a hammer.
>
>High octane fuel is hard to ignite and therefore resists this autoignition.
>Too much compression and/or too much spark advanve will cause detonation.  A
>poorly shaped combustion chamber which takes too long to burn or low
>tubulence or no quench area or other not easily adjustable factors also
>affect the propensity to detonate.
>

Good description and very similar to others that I have read.  However, in a
brief instant the description does gloss over _how_ the extra spark advance
will cause the detonation.

Assuming the engine is turning 4000 RPM when maximum advance is reached
then, as you stated the flame propogates nicely through the entire chamber.
Advance the timing by, say, two degrees, and detonation results.  This would
imply, by your description, that if the ignition starts 82 microseconds
earlier, (2 crankshaft degrees at 4000 RPM), the temperature rise would
somehow increase and we'd pass the autoignition temperature of the fuel.
Does this happen before TDC then or after TDC?

As I recall basic theory,  the entire reason for advancing ignition is to
make sure the entire charge is completely burned and supplying maximum
energy to the top of the piston before the piston reaches the bottom of the
stroke.  I assume that the charge is still burning when TDC is reached
because even with 30 degrees advance there is only 1.25ms of burn time
before TDC is reached.

Perhaps a better question is how long does the combustion process take?
(average)

>From another angle...

We know that detonation only occurs during the time between ignition and BDC
of the power stroke. A total of 210 degrees given 30 degrees total advance.
On a four cylinder engine the next piston in the firing order is on
compression stroke and its charge has already ignited when the power stroke
piston is at BDC.  This means in order to prevent detonation that next
cylinder detonation has to be detected, identified and applied to the
ignition advance of the cylinder under compression.  In other words,  if
detonation probably occurs within 180 crankshaft degrees from 30 degrees
BTDC to 30 degrees BBDC then there is time to reduce the advance for the
next cylinder afterdetonation is detected.  Question is when exactly does
detonation take place, or does the engine pay the price for two, (or more),
cylinders detonating before timing is cut back?

I am assuming of course that the engine knock detector is a high quality
magnetostrictive accelerometer tuned to 5-11KHz and 11-17KHz.  What's
interesting is of course the period for one cycle of a 5KHz knock pulse is
200 microseconds which is more than double the time of advancing the
ignition by two degrees at 4000 RM.

So the dilema is how to catch the knock in one cylinder and prevent it on
the following cylinder.  Assume each time knock is detected the advance is
retarded by 3 degrees.  We certainly don't want to detect it 5 times in one
power stroke and reduce advance by 15 degrees for the following pistons
power stroke. Once the knock is gone then over a number of cycles inch the
advance back 1/3 degree per cycle.  This is the way the SAAB 900 manual
states that they do it but they don't say when they detect the knock.

Assume firing order 1342

>From a software perspective we can set a flag that the knock occurred during
the interval from ignition to ignition; eg. Cyl #1.  Once set the flag is
cleared at the next ignition point (Cyl #3) which has been retarded by 3
degrees.  If that results in no knock then at the third ignition point (Cyl
#4), which is still retarded by three degrees the future retard is reduced
by 1/3 degree so Cyl #2 ignites with a retard of 2.667 degrees.

Make sense?  I've glossed over something here too.  If the ignition point is
set by a hardware timer determined before the knock is detected then in the
above example Cyl #3 won't be retarded which means it too may knock.  After
that though, Cyl #4 would be retarded.  If the knock occurred and was
detected before a hardware ignition timer was initialized then Cyl #3 would
be retarded.  So,  knowing when knock occurs can help reduce the incidence
of knock.  

Babble_mode := off.

Regards,

John

Automation Artisans Inc. is a company founded to provide unique, advanced
technology software and hardware solutions to the manufacturing industry so
that our customers can produce customized products at mass production costs.

Automation Artisans Inc.,   PO Box 20002 Sidney, BC CANADA V8L 5C9
Ph. 1-250-544-4950  Fax 1-250-544-4954