Quantcast

Down-Home Vibrometer

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
130
Location
Altus SW Oklahoma
For several years, I owned a C-150 formerly used by Spartan - Tulsa for training.
The overhaul engine in it had chrome cylinders and was remarked on for its smoothness.
I supposed that Spartan would have used vibration analysis in the course of training A&Ps.

I needed to smooth minor prop chips once or twice over the period of ownership.
I noticed that Vibrometers go for $150 - $2000 so I looked for a homebrew.
As it turns out, 3 axis accelerometers with analog outputs are now available at attractive prices.
ADXL335 ~ $15
ADXL335.jpg

So far so good. Next was the question, how to display the vibrations in X, Y, Z ?

I settled on a chinezium toy oscilloscope a DSO150 ~ $28
2020-11-06 14_25_21-Window.jpg

This device displayed the orthogonal accelerations in units of g as desired but lacked a most important facility - a second channel to display a trigger pulse, which I envisioned as a laser pointer and a phototransistor with a ruby colored plastic filter. Although I did not consider the extra cost of a two channel display was warranted for this test, I see that triggering on a sensor trained on a prop blade and adjusting the display speed suitably would indicate the crank angle for maximal amplitude. (I omit the details of connecting the accelerometer to an ethernet socket, so that a regular enet cable can be arranged to remote the sensor into the cockpit, etc.)
 

TFF

Well-Known Member
Joined
Apr 28, 2010
Messages
13,833
Location
Memphis, TN
All balancing I have done is either done by clock or degrees. 0 or 12 o’clock is set as point of timing trigger. So essentially a trigger to start the clock and vib peak timed from that point Then convert it to measurement to know where to add weight.
 

TFF

Well-Known Member
Joined
Apr 28, 2010
Messages
13,833
Location
Memphis, TN
It’s pretty interative with a position sensor. If can identify the heavy side, it’s all good. I watched a guy do wheels on the car by just feeling it with his hand. It was amazing. There is also a mathematical way that watch balances use. Blade to blade is pretty easy on a two bladed prop. Cord balance where the weight needs to be perpendicular to the blade makes it much harder.
 

cowlove

Member
HBA Supporter
Joined
Nov 20, 2014
Messages
21
Location
Seattle, WA
I liked this very practical contribution. Sugru, cellphone, app.
Not the best way to check out an aero engine - way too close to that whirling widow-maker. Otherwise; cheap cheerful and practical. $5 for the Vibration app.
There's a slightly more refined method that does some math on three trial weight vibration measurements and tells you where to add weight.

Fairly brief discussion: Spreadsheet for Balancing

Spreadsheet to do the math: http://www.conradhoffman.com/Balancing.xls
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
130
Location
Altus SW Oklahoma
There's a slightly more refined method that does some math on three trial weight vibration measurements and tells you where to add weight.

Fairly brief discussion: Spreadsheet for Balancing

Spreadsheet to do the math: http://www.conradhoffman.com/Balancing.xls
I am torn between strewing flower petals before cowlove's path, and looking askance for his damning (with faint praise) of Conrad Hoffman's EXCELLENT procedure.
You have certainly done a great service in making more accessible, Hoffman's paper.
Thank you!
 

karmarepair

Well-Known Member
HBA Supporter
Joined
Jan 13, 2011
Messages
283
Location
United States
There's a slightly more refined method that does some math on three trial weight vibration measurements and tells you where to add weight.

Fairly brief discussion: Spreadsheet for Balancing

Spreadsheet to do the math: http://www.conradhoffman.com/Balancing.xls
Reading further down the thread, it looks like one Mr. Carter turned Mr. Hoffman's spreadsheet into a web calculator! http://www.carter-engineering.com/balancecalc.html

It's mentioned in the Practical Machining forum, but you can use a mini speaker as an accelerometer thusly (stolen from a different online forum):
"How about using a small loudspeaker? Load the dome up with a little bit of weight like a passive radiator. Then attach the frame to the <test article> (hot melt or sticky wax). While bigger than normal accelerometer it is configured like a seismometer. The output would be an ac signal that could be amplified with a preamp/phone preamp/or power amp depending on the gain and voltage range required."
 

spaschke

Well-Known Member
Joined
Oct 24, 2012
Messages
246
Location
Denver, CO
I was thinking of doing something very similar to detect flutter. embed one in a control surface, using an arduino with a lcd display, calculate and show the frequency of the vibration. This Arduino can send a signal to my Annunciator Arduino, which plays a warning mp3 on the intercom.
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
130
Location
Altus SW Oklahoma
I was thinking of doing something very similar to detect flutter. embed one in a control surface, using an arduino with a lcd display, calculate and show the frequency of the vibration. This Arduino can send a signal to my Annunciator Arduino, which plays a warning mp3 on the intercom.
I wonder if you have seen the trailer demo, with variable weight distribution?


What BJC and Bill Higden were hinting at so diplomatically, was this.
If your control surface is unbalanced AND subject to flutter AND the flutter is divergent, you may have just two or three seconds until the surface departs.

I am not glibly repeating what I have heard. I had occasion to tow a heavy big round alfalfa bale on a custom hay spike trailer. This weighed about 1500 lb, rather than the 1000 lb grass hay which I customarily bought There was no opportunity to distribute the weight forward, as any hay tow would want. And so, at the not excessive speed of about 35 mph, the trailer 'fluttered'. For perhaps 4 seconds. That's not quite the right term for a trailer oscillation, but the results were much the same. The truck stopped across the road - and the trailer had separated in 3 or 4 pieces.
(Even more unexpected, a truck driver drew up and asked if he could help - I explained - and he brought a flat bed to take the pieces home - where two or three days later, the trailer was rewelded, and looked rather stronger than before.) But this was Oklahoma, of course.
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
130
Location
Altus SW Oklahoma
There's a slightly more refined method that does some math on three trial weight vibration measurements and tells you where to add weight.

Fairly brief discussion: Spreadsheet for Balancing

Spreadsheet to do the math: http://www.conradhoffman.com/Balancing.xls
So I finally ran a trial, using a very heavy lathe chuck mounted on a servo motor, provided with a speed control. I duct taped the accelerometer to the motor body and noted the Z axis, after checking its signal moved appropriately from +1g to -1g when held upside down.
The visual waveform was unimpressive, but the digital data for peak to peak amplitude, went from 62 mV to 210 mV when I chucked a bent rod and spun it up. So I ordered some Sugru and some self-adhesive wheel leads, and felt well-pleased.
I was delighted that Conrad Hoffman's spread sheet was quite unstuffy about the units used - I input accelerations as millivolts, the unit reference weight in ounces and it offered up the required weight as a decimal fraction of my preferred unit at a specified arc in degrees from my baseline. Tickled pink!
 

BrianW

Well-Known Member
Joined
Jul 2, 2018
Messages
130
Location
Altus SW Oklahoma
A few days after this vibration thread completed, I got a private note which mentioned the familiar rough engine sound which occurs when embarking on a trip across any considerable body of water - and how it would be good to have a device listening for real as opposed to imagined auto-rough. Moreover, he included to a pointer to a machine-learning approach which could be trained to distinguish between "running sweet" at a variety of speeds and "running rough" from here:
from these folks:
1606244658838.png

This seemed to offer a hand-holding approach IF I purchased an Arduino Nano 34 IOT ($20) and a 3 axis accelerometer ($15) . So I did and began to carry out the recipe mentioned there.

After several days effort, I realized that an Arduino host like Arduino nano 33 iot operated from a Windows PC was NOT the natural vehicle they had in mind - and before deploying, one needs to pay for the use of their neural net implementation.
But by now, my interest was thoroughly aroused, so I looked around for some more accurate hand-holding written in native English, and I soon ran into this:

YES, I have some insight into using Arduino cards, with pre-digested libraries and working examples; and YES you can provide sound/vibration data samples before you hook ANY devices up to train a network to distinguish between good and rough vibrations/sounds and I noticed this approach uses "Edge" too - but with no pay later strings.
I commend it to you.
 
Top