So I want to make sensor to detect at where angle will the pipe crack if I give torsion force at it.
What sensor should I use?
Thanks.
Is the sensor supposed to detect the "pipe crack"?
a flex sensor?
at a previous job we used high speed camera's for those kind of tests and IIRC there was a background with reference lines so one could calculate the angle.
what kind of material is this pipe made of?
jremington:
Is the sensor supposed to detect the "pipe crack"?
Yup, when the pipe cracks, the sensor detects at where angle does the pipe crack.
robtillaart:
a flex sensor?
at a previous job we used high speed camera's for those kind of tests and IIRC there was a background with reference lines so one could calculate the angle.what kind of material is this pipe made of?
Read a bit abut flex sensor, does it only can detect bending moment?
This is what I want, twisting moment or torque => http://puu.sh/hNVhd.png
Is there any sensor that detect how far it moves with 2D (x,y) reference? So I don't need high speed camera like you did.
It's PVC.
Read a bit abut flex sensor, does it only can detect bending moment?
you must sample it with high frequency.
Analogread() can make approx. 8000 samples per second (standard mode) so you should be able to detect a sudden change in value at a millisecond precision.
//
// FILE: analogJumpDetection.ino
// AUTHOR: Rob Tillaart
// VERSION: 0.1.00
// PURPOSE:
// DATE: 2015-05-15
// URL:
//
// Released to the public domain
//
#define SIZE 200 // A MEGA can hold more.
#define THRESHOLD 50 // ==> threshold of cracking
uint32_t times[SIZE];
int bend[SIZE];
int index = 0;
void setup()
{
Serial.begin(115200);
Serial.println("Crack Detector 1.0");
// initial value
index = 0;
times[index] = micros();
bend[index] = analogRead(A0);
}
void loop()
{
// remember previous sample index
int prev = index;
// make new sample
index++;
if (index == SIZE) index = 0; // wrap the circular array
times[index] = micros();
bend[index] = analogRead(A0);
// detect the crack by a jump in flex (could also be an absolute value
if (abs(bend[index] - bend[prev]) > THRESHOLD )
{
// dump the last SIZE samples
Serial.println("\nTIME\tBEND");
for (int i = 0; i < SIZE; i++)
{
index++;
if (index == SIZE) index = 0;
Serial.print(times[index]);
Serial.print("\t");
Serial.println(bend[index]);
}
Serial.println("\ndone...");
while (1); // block the application
}
}
truncated output - (tested with a potmeter)
Crack Detector 1.0
TIME BEND
9407856 130
9407976 128
...
9431224 129
9431344 125
9431464 97
9431584 85
9431704 75
9431824 131
done...
attached the output in Excel graph
To measure torque you could use a similar program, but you should measure the torque applied AND monitor the position of the pipe. If it cracks it "rotates".
rotation detection can be done by drawing a white/black pattern and use a fast photo transistor to detect thepipe has moved. (LDR is too slow)
Thanks a lot for the explanation.
rotation detection can be done by drawing a white/black pattern and use a fast photo transistor to detect thepipe has moved.
Can you elaborate this?
put paint on the pipe an let a sensor monitor the reflection of light, if it changes it is probably rotated.
Have you considered strain gauge?
The question is to measure the angle, so I presume there is an axis somewhere being turned, a
rotary encoder is the normal way to measure angle.
What angular precision is needed?
Hi,
Can you post a diagram/picture of your test jig, so we can see what you are using?
Tom...... 
Just a side note.
PVC doesn't usually crack when bent. Most if the time it just kinks and subsequent pressure build up may then cause a burst of the weakened structure. Kink would cause severe flow restriction and pressure build on input side. Good place to look for sensory detection of failure.
This might be a silly suggestion, but since your pipe is PVC rather than metal, you could try a using a compass like the LSM303 attached to the pipe since presumably the earths magnetic field gives you a fixed reference to compare against the pipe rotation. Though I am not sure that a quarter of a degree is sensitive enough for this?
And since there is also an accelerometer you could set up tap sensing to trigger an interrupt to record the forces hitting the pipe - wether shaking or bending. Then you would get some idea of the progression of the failure event. I have heard of some people recording very fast accelerometer samples by loading rapid analog readings directly into Rugged Circuits QuadRAM boards.