05/26/2004

Knock Detection

Just for giggles.

Each knock sensor is a microphone.  Use it as such.

Take a calibrated noise source - even sine wave will do.  Put it into a
speaker.

Plot the output frequency response vs the input frequency.

Only range that matters is 5 to 10 kilocycles. That's where knock
occurs.

Now, lets enter the way back machine and revisited history.  You want to
look for a dude named Bodine.  A.G. Bodine Jr.  Not a big name.  Unless
you are into ultrasound and oil.

Mr Bodine was THE MAN when it came to sound in the oil fields.  Like
well over 800 patents relating to sound in the field - back when patents
actually meant something.  

A minor example is slant drilling.  Now to turn the bit, you need a way
of driving it that does not get crimped or hossed up during that ninety
degree turn outward, and a bit thirty five miles long with a 90 bend is
just a tad hard to conceive.  Unless you drive the bit with ultrasound
waves hydrodynamically coupled to the bit.  Heck, you could drill in
pretzel patterns if you wanted to.

Now this dude has some patents on automotive knock and how it works.
Having invented most of the ultrasound mechanisms in the oil field word,
you might get a clue he may have some clue about sound and knocking.
But, after all, Nobel Prize winning bio-chemist Linus Pauling was full
of sheeite over Vitamin C because he did not have an AMA Union Card
Doktors license.  So I guess a Grad Student at Phuque U. probably knows
as much.

His work shows that knock frequency is determined essentially by bore
diameter - as knock is a "Regenative Acoustical Wave".  The temperature
has an effect, but as combustion temperatures for hydrocarbons are
determined by chon, and most engines run stoic, the knock frequency
follows bore diameter.

Filtering in another device may reduce the activity of valves etc from
false knocks - but back to basics - its going to happen at a certain
frequency.  Lower the combustion temperature ( light load etc ), the
frequency drops radically.

Get picky. A 454 chevy or a 502 cadillac has the about the same size
bore as a 460.  Anything from a 4.25 to 4.50 bore would be anal in
frequency.

Clock the sensor to the cam and I quarentee dang near any device will
work.

If you wanted to play, he describes the nodes and anti-nodes in a
cylinder and how to find them.  Someday, if I get around to it, I'll
forward the patent data to Dave.  Unless Chef Chef's interfere and you
lose.

If you wanted to get an approximation of knocking frequency, you could
apply a sound source to the chamber, and run it at levels until you see
regeneration.  A sweep from 5 to 10 kilocycles should do it. ( Hertz -
that's what my arss feels like after having a nice habaneras and corn
dinner and doing a democrat. )

Separate speaker and mike.  Run it every few degrees from top dead
center till about 90 past tdc.  

Then correct it from 498 rankine to a band from about 4000 rankine to
5500 rankine.

4000 rankine to 5500 rankine is the normal combustion temperature which
will determine the normal speed of sound in the chamber and the volume
will be from clearance volume till 3 to 4 clearances with a factor for
the bore.

Somebody plot that out and tell me again how special and unique the
sensor has to be to detect knock.