Vibration Monitor (Hyde, Wanttaja?)

In article ,
RST Engineering wrote:
Proximity probes? Or accelerometers?

The proverbial "it depends". _first_ you have do decide what your
'frame of reference' for vibration is. If you're looking for
excursions in the drive shaft, relative to the engine block, then
a pair of proximity sensors in quadrature to measure excursions from
the shaft axis, plus a strain gauge at a fore-to-aft thrust bearing
gives you complete data. Convert from XYZ to polar co-ordinates,
Fourier transform on the magnitude component, check each passband
against loading-corrected reference values, and illuminate corresponding
indicators.

If you're looking for things that might affect all parts of the engine
equally, then, can you use the airframe for reference, or not? If "yes",
then the same approach works, just on a different scale. If, "no", then
you're pretty much stuck with three _inertial_ accelerometers, mounted at
mutual right angles.

I understand about the analog circuitry and actually plan on making a five
or six channel filter at each of the possible resonance points relative to
the fundamental ... and then strobing the filters to light a "normal",
"low", "high" lamp for each channel.


Sounds like you're talking about using a group of narrow bandpass filters
and sampling the short-term average signal level out of each filter,
"Looking for" a marked increase vs a base-line reference (at that frequency),
as an indicator of problems.


I see a couple of difficulties -- not necessarily insurmountable, but
there, nonetheless. (1) Most, if not all, the 'frequencies of interest'
are dependent on engine speed. thus the filter 'center frequencies'
will have to track engine RPM. (2) "Normal" vibration _magnitude_ will
vary with the load on the engine. The trigger threshold would seemingly
have to adjust to compensate. 3,000 RPM in a 2,000 FPM climb is a
different engine environment than 3,000 RPM in level cruise. I don't
know enough engine mechanics to guess whether just the throttle setting
is going to be sufficient to determine the required adjustments. At
a guess, torque relative to the airframe would be a better indicator.

Lastly, I'd suggest one channel that is comparatively broadband. specifically
to catch unanticipated "out of band" goings-on.

Jim




As is tradition, we tend to get off topic ;-)
I think you can accomplish this using the methods discussed.
Mount 2 proximity probes 90 degrees apart.
Calibrate the readings for normal low vibrations (be sure to account for
any nonlinearities in the probe).
Design the circuit to trip the buzzer/lamp when the vibration exceeds the
normal level.
You may need some analog circuitry to help (gain, etc).
But you don't need to get much fancier than that.

Adam

In article ,
mindenpilot wrote:

"Robert Bonomi" wrote in message
...
In article ,
Morgans wrote:

"RST Engineering" wrote in message
...
Yes, all of the below.

Sometimes you drill a hole in a small magnet and use a small screw to
hold
it to the propeller backing plate. Sometimes you use a strip of
reflective
tape on the prop itself. Sometimes you use a pulse from the #1 magneto
lead.

Jim

This guy was talking about a notch and proximity detector for the phase
detection. I know about all of the things you mentioned, but the
proximity
detector's workings are new to me.

Think of a proximity detector as 'ultra short-range radar'. grin
It may use reflected RF energy, or 'optical'.

where the 'excursions' you're trying to measure are smaller than the
wavelength of the measuring 'beam', you can use simple phase-shift
between outgoing and returning signal, to determine distance.

where the distance is much larger than the wavelength, you have to
impress a carrier on the beam, and measure phase-shift in the carrier
frequency. this gets an 'approximate' distance, that can be further
refined by phase angle measurements of the beam itself.

Capacitance tracking is also a possible approach.
and/or "Hall effect".

These can get 'messy', due to inherent non-linearity in the technology,
that has to be compensated for, in 'reading' the signals.

Capacitance tracking works best where there are *very*small* vibrations
involved, and a very _smooth_ surface to measure against. The technique
is capable of mapping individual atoms/molecules in a crystal lattice.
Scientific American had a write-up -- at least 15 years ago -- about
a new 'super microscope' (successor generation to the scanning electron
microscope) that worked in that manner. a _very_fine_ 'needle' was
carefully moved, raster-style, across the object being 'scanned', and
the capacitance changes between the needle and the object were mapped.



Actually, the proximity probes I am familiar with work on the "eddy current"
principle.
That is, RF energy is directed at the shaft through a coil. Some of the
energy is lost into the shaft in the eddy currents.
The energy that is coupled back into the coil is the measured signal.
Since the amount of energy dissipated in the shaft is proportional to the
distance between the shaft and probe, the result is a displacement
measurment.

No doubt about it, there are *LOTS* of ways to measure proximity.

Eddy current detectors have somewhat limited application.
1) the material of the item measured must be uniform in composition where
it is 'presented' to the detector.
2) the material has to be electrically conductive to a degree. If the item
under test is a good insulator, you don't get eddy currents, If the item
is _too_good_ a conductor, you don't get eddy currents.

For engine shafts, etc. those items are a "non-issue", agreed.

They don't work worth a d*mn if you're trying to measure motion in a rubber/-
polyethylene/vinyl/whatever drive belt, though. grin

"Robert wrote

No doubt about it, there are *LOTS* of ways to measure proximity.

and some others contributed lots

And I think I'm almost sorry I asked! g?
--
Jim in NC

Well, a Spectral Dynamics SD2001 FFT would be well received if you have one
laying around unused {;-)


Jim



"mindenpilot" wrote in message
...
At any rate, what you intend on doing is definitely do-able.
I recommend playing with probes a bit to see what you get.
Sounds like a fun project.
Let me know if there's anything I can offer...

Adam

"RST Engineering" wrote in message
...
Understood. What I started out to do (and still plan on doing) is to have
a device that will stay permanently mounted to the engine that can be
calibrated (adjusted, signed, pick a verb) when the engine is known to be
good and light a "your engine is about to come apart" lamp at the
appropriate time.


Excuse my ignorance, but couldn't you just feel the vibrations?

If you wait that long, you're probably more concerned if the chain that's
holding the motor to the firewall is going to break after the motor cuts
loose from the mounts. All too often, the vibrations start to pick up
seconds or miliseconds before a catastrophic failure.

To do such a health-monitoring function properly, you really want some
seeded fault data to characterize what a "bad" engine spectrum looks like.
How many engines do you want to sacrifice to get the data? You can approach
it from the "anything different from a healthy engine signature" standpoint,
but that will likely result in a ton of false positive fault indications.

"LCT Paintball" wrote in message
news:_Lq0e.14520$fn3.9681@attbi_s01...
"RST Engineering" wrote in message
...
Understood. What I started out to do (and still plan on doing) is to
have
a device that will stay permanently mounted to the engine that can be
calibrated (adjusted, signed, pick a verb) when the engine is known to
be
good and light a "your engine is about to come apart" lamp at the
appropriate time.


Excuse my ignorance, but couldn't you just feel the vibrations?

RST Engineering wrote:
Necessity, as they say, is a mother.

I am in the process of reinventing a square wheel called a vibration
monitor. The electronics is relatively trivial IF the input and output
parameters are known.

What we know is that the engine is going to have a fundamental frequency at
cruise RPM. Let's take the math-simple engine RPM of 2400. This gives us a
fundamental frequency of 2400/60 or 40 Hz.

But wait, he said. There are going to be other (sub) harmonics of that
frequency that will be of some interest. And, those harmonics will change
as a function of the engine being a two or four stroke, four or six
cylinder.

So, oh wise and noble gurus of engine stuffings, what (sub) harmonics are
going to be of most interest to us and what is their mathematical
relationship to the fundamental?

As an extra bonus question, my sensor is going to be an old phonograph
cartridge. Should I use the lightest weight "needle" that I can find? How
about a tiny little ball of lead at the tip of that needle? Would that help
the sensor? Or hinder it?

Lastly, once I get this sucker up and running with you all's good ideas, is
anybody game to bolt it onto their flying machine and report results? I can
do it for the 182, but I'd really like some other real-world reports.

Sometime ago someone published a program that runs on a PalmPilot. It
uses the microphone to listen to a plane engine and will tell how many
RPM it is turning, even from a half mile away (IIRC). Could you do your
engine vibration analysis through the microphone rather than
accelerometers and other sensors?

If you wait that long, you're probably more concerned if the chain that's
holding the motor to the firewall is going to break after the motor cuts
loose from the mounts. All too often, the vibrations start to pick up
seconds or miliseconds before a catastrophic failure.

To do such a health-monitoring function properly, you really want some
seeded fault data to characterize what a "bad" engine spectrum looks like.
How many engines do you want to sacrifice to get the data? You can
approach
it from the "anything different from a healthy engine signature"
standpoint,
but that will likely result in a ton of false positive fault indications.


Are you suggesting that a bad engine will give clues to it's demise enough
in advance that you could actually do something about it? Clues that a
monitor could pick up on, but an experienced pilot wouldn't?

Of course. It has been done. Depends on the failure mode, of course. There
are some failure modes that take a long time to develop that give early
indications, and some that don't. A ton of work has been done in this area
for military jet engines. Seeded fault test data is the key to this.
Unfortunately, that might mean wrecking a bunch of engines to get the data.
It's not a project for the average home-builder.

Personally, I wouldn't bother trying to get a vibration caution together for
a home-built. Doing right would be just way too expensive. It would be
cheaper to just buy something that's turbine-powered and get rid of the
hazards that way. Besides, a huge number of failure modes already show up in
CHT's, EGT's, RPMs, etc. You have to weigh the cost of covering additional
failure modes against the hazards. This is really a job for engine
manufacturers. Additionally, you have to take complexity and reliability of
the sensing and processing into account. A monitor that is always going
haywire on you would be worse than nothing at all.

I'm actually looking at some stuff like this for possible inclusion on a
future project right now for a different type of powerplant. If you can
reliably predict RUL (remaining usable life) for a critical component, it
could be possible to reduce the amount of redundancy in a complex system and
rely on health monitoring functions to let you know when it's time to
replace the part.

PHM (prognostics and health management) has been a big focus in the military
aircraft world in recent years. I'm hoping that some of this technology will
trickle down to us in the GA world. Hmm...maybe I should get with an engine
manufacturer and work something out... SO, how much would people pay for an
engine health monitoring system package as an option for a new engine (i.e.
one of the new generation...maybe a DeltaHawk)? My guess is that it would be
too expensive to ever sell.

Pete


"LCT Paintball" wrote in message
news:ipA0e.102105$Ze3.20828@attbi_s51...
Are you suggesting that a bad engine will give clues to it's demise enough
in advance that you could actually do something about it? Clues that a
monitor could pick up on, but an experienced pilot wouldn't?

Pete Schaefer wrote:

PHM (prognostics and health management) has been a big focus in the military
aircraft world in recent years. I'm hoping that some of this technology will
trickle down to us in the GA world.

One useful technology that keeps getting more real are self-powered
sensors that communicate via bluetooth or other wireless, so you could
just stick them on various places and not have to worry about cabling
and all those other points of failure.