Help on Experiment using Rotary Encoder

Hi everyone

Im currently an student a University in the mist of doing my Final Year Project. This project requires me to make use of a rotary encoder to measure the Torsional Vibration of a rotating shaft. As I'm very very new to such products, I have no idea how to go about doing it.

The gist of the project is to obtain the signal of the rotating shaft with and without the torsional vibration. By taking the difference of the two signal, I would be able to generate the signal for the torsianl vibration at that particular speed. However, I do not nkow how to go about choosing one encoder that is suitable. I would need one that is able to cater to very high rotating speed, say 30,000RPM. But how to go about choosing one?

Lets say for example, I need to obtain the signal for a shaft rotating at 10,000RPM, how do you guys go about choosing an encoder?

Kindly guide me along..

thanks and regards.

dylan

I would start HERE

Maybe look here HERE ALSO

PROPELLER DRIVE SYSTEMS AND TORSIONAL VIBRATION

Thank rsdoran

I have been to these sites, but they did not relate to me on how to go about choosing a suitable encoder and the steps and procedures in choosing one. Would u have any ideas in how to go abou;t choosing a suitable one for a given rotating speed??

High resolution is the key!

The more counts/revolution the better.
You will need something other than a PLC to this job, but once you get the right materials it will be very easy.

BTW, 10000 RPM? 1/6 of a rev/millisecond. Ff you had a 6000 count/rev encoder would 1 count/millisecond resolution be good enough.
I think I would try for 10 times that resolution.

Now how many counts/sec is that? I don't think you will be doing that using a PLC or a high speed counter card.

encoder

Dylan
First of all you should specify exactly what is your operating speed(RPM) cos one of the important parameters which the encoder manufacture declare is "MAX RPM" this parameter tell you what is the max speed which the encoder can obe operated in without any damage for the encoders specially the bearing and shaft.
Seconed: according to the accuracy you need you should select the no. of pulses per revoulation(PPR) you need to achieve this accuracy.

BTW, 10000 RPM? 1/6 of a rev/millisecond. Ff you had a 6000 count/rev encoder would 1 count/millisecond resolution be good enough.

Click to expand...

Peter "its now 1000 count/millisecond"
any how as peter said the more counts/ rev is better but take care you should also select accurately which device will deal with this signals ( plc with high speed counter module , counter, ...etc)
because each manufacture for such products also declare " MAX FREQUANCY " for the high speed inputs which you can't exceed in any way . in peter example the frequancy is 1 MHz.
I hope that this will help you.
good luck

I was getting rushed. I am atl work.

I was trying to say that the resolution would be 1 count out of a 1000 per millisecond or .1%

I routinely ask my customers to provide 40,000+ counts per revolution to get good speed estimation. I have never had a problem where the customer had too high a count/rate. Usually the problem is just the opposite.

I recommend using a real motion controller. After all you still will need to control and measure speed/positions. Another must is that the two encoders must be sampled at exactly the same time.

Hesham. 1 MHZ count rate is easy for a good motion controller.
The encoder would need to be a high quality one with differential outputs to handle the high speeds.

It would take only a few minutes to read the data into a MathCad spread sheet and subtract the two sets of data and do a FFT on the difference. The FFT yields the frequency of the vibration. The motion controller that I have access to saves the counts from multi axes in buffers that can be read into excel spread sheets or better yet MathCad.

Dylan hasn't said how many degrees of torsion he expects or at what frequency expects the shaft to vibrate at.

Dylan, are you trying to model a system or do quality control?

I thank hesham and peter for your valuable inputs.

Peter, Im trying to model a system whereby it enables me to measure the torsional vibration of a shaft of various speed. Given a shaft roating at a given speed, the system enables me to determine if torsional vibration exist and if it does, what is the signal of this vibration.

Degree of torsion - what is it?

he frequency of the rotation of the shaft - has it got to do with the motor being used?

kindly advice..

thanks alot

dylan_fc said:

I thank hesham and peter for your valuable inputs.

Peter, Im trying to model a system whereby it enables me to measure the torsional vibration of a shaft of various speed. Given a shaft roating at a given speed, the system enables me to determine if torsional vibration exist and if it does, what is the signal of this vibration.

Degree of torsion - what is it?

Click to expand...

How much the shaft twists in degrees.

he frequency of the rotation of the shaft - has it got to do with the motor being used?

kindly advice..

thanks alot

Click to expand...

It is you that mentioned 10000 RPM! If you what to measure vibrations as the shaft spins a 10000 PRM then a motor is need that will turn that fast. Simple.

Peter Nachtwey - thank a lot yah. Im really at a lost of wat to do for my expt. anyway, the max RPM that i presume my professor wants the shaft to rotate at would be 30,000RPM. but i doubt he can provide me with such a motor.

so lets say I can use and encoder which is able to have 40,000+ count PPR, I would be able to cater for the motor that rotates at 30,000RPM. Is that correct?

As for the degreee in the torsion, I would say, the smaller the degree that i'm able to record the better. Does that means that to record a smaller degree of torsion vibration, a hiher resolution is needed?

Peter Nachtwey said:

How much the shaft twists in degrees.



It is you that mentioned 10000 RPM! If you what to measure vibrations as the shaft spins a 10000 PRM then a motor is need that will turn that fast. Simple.

Click to expand...

peter? u there? need more of ur advice.. can u lend me a hand?? thanks

To be brutally honest.

I don't think your professor or you knows what he is asking for.
Provide specications and preliminary calculations.

BTW, your questions are important too. I can tell from the questions whether you have a clue about how to proceed.

Convince me that I won't have to just tell you the answer every step along the way.

You should read the thread about linearization and tell me how you can use the least squares in your project. You had better consult with your professor before answering. Hint, the least squares can be used for other things than linearization. What else?

This project or topic is way beyond the scope of PLCs. It is, however, one of the few worthy projects that a college student has posted on the forum.