"Rosine, Steve" wrote:

I have an idea to propose as an alternative to the audio amp method of knock
detection for those with cockpit PC's.  I intend to use this as diagnostic
tool to determine bearing condition on my original design PSRU (photos and
drawings to follow in a few days).

THe knock sensor output is essentially the same as a microphone, and can be
connected to the microphone input on a PC sound card.  The growth of digital
music means that there are many programs available that can provide a
real-time FFT display of an audio signal.  I have tested a program called
Spectrogram (www.monumental.com/rshorne/gram.html), that seems quite
versatile (and it's FREE!).  It handles stereo sound, so that means two
knock sensors displayed at once.    I have tested this program with a $25
knock sensor for a Caprice (don't know the year) which has a 1/4" NPT
fitting, and the window air conditioner above my computer.  I was impressed
with the clarity of the display and noticable changes in vibration profile
when I changed A/C fan settings.

Attached is an example from the web page of a warbler call....Imagine the
spectrum of the call of the Wankel.

Steve Rosine

Mustang II / 13B

This is starting to approach the methods used by "real" (large budget) engineers
to check out a prop/engine combination for problem vibrations.  If you mount
your
sensor on a prop (metal) and whack the prop with a plastic hammer (at
various locations
down the length of the prop) and look at the FFT of the resulting vibrations
you
can locate the bending natural freqs of the prop.  Probably want to record the
first three nat freqs at least, for or five is a bit better.  I've never
seen a real system have
a severe problem with anything higher than a third, but a suppose safe is
better than
sorry.  Usually 4thand 5th have such low amplitude that you  can ignore them.

Now - with the sensor located on the engine (or gearbox) close to the prop run
the engine/prop at various RPM and loads sets.  RPM variation alone should
work,
but you can't get to cruise or climb RPM on the ground with a fixed pitch
prop.
If you see any significant  response (spike in the FFT) at or near one of
the known
(from previous static prop tests) bending natural frequencies of the prop -
STOP.
Get another  prop and start over, or placard the aircraft against operating
anywhere
 near that "bad" RPM.  Some Mooneys are placarded since they don't have the
pendulum 6th order dampers on their cranks.  Can't run from 1900 to 2300 IIRC.

  With PCs, we are now able to do things at home that cost from tens to
hundreds of thousands of dollars a decade or two ago.   Cool!

Bill

By the way - FFT stands for Fast Fourier Transform, which is math gobldygook
for a cool way of extracting the individual vibration components (sine waves)
which make up the overall vibration (which looks like "noise") signal.  Each
part of
the whole engine/prop assembly is vibrating at a natural or forced frequency and
they all add up into a complex mess.  FFT can "put Humpty Dumpty back
together again"
and sort out both the freqency and relative magnitude of all the
contributory vibration
inputs (sine waves).  A very cool application of some really complicated
higher math.
Long EZ builder & pilot
mechanical engineer

 
The AirCraft Rotary Engine NewsLetter.              Powered by Linux.

http://home.earthlink.net/~rotaryeng/            http://www.linux.org