How can we accurately measure the natural frequencies of our fixture?

[siddy88]

Hi all,

My fellow students and i are working on a university project. We are making a fixture and doing some vibration testing. We purchased a few disk motors and we are attaching those to our fixture. The set up is fine and we have been able to get the motors running and been able to vibrate our fixture however we have no idea how to actually test for the frequency of the system.

Basically I am getting at this- do we need labview, a DAQ etc? and if yes is there any tutorial on how this can be set up?

Any advice on how we should do this. We have an accelerometer that we are considering hooking up to the system but are not too sure if this is going to work. Any idea on where we can read up or does anyone have any good links that we might be able to follow up with?

thank you for your help

 

[Mech_Engineer]

Your forcing vibration will be directly related to the speed of your motors. Do you know what speed they are operating at?

 

[siddy88]

Hi Mech_Engineer,

The motor has a spec of 1G vibration and an operating voltage of 2.5-3.5V. Also the spec states that the weight within vibrates at a frequency of 10000RPM. Any of this information helpful?

 

[Mech_Engineer]

Hi Mech_Engineer,

The motor has a spec of 1G vibration and an operating voltage of 2.5-3.5V. Also the spec states that the weight within vibrates at a frequency of 10000RPM. Any of this information helpful?

The point is that your motor's speed spec -10,000 rpm- is the frequency of your vibration. 10,000 revolutions per minute, 166.67 revolutions per second, A.K.A. 166.67 Hz. Keep in mind though that if your motor is a DC motor, its speed changes with voltage.

For direct feedback of the forcing vibration, you can put a tachometer (speed sensing encoder) on the motor's shaft.

 

[siddy88]

Thank you. That was helpful . I have two more queries That I'd like your input on if convenient .
1) if I place 4 of these motors on my base plate would it create a series effect ie would my frequency then be 166*4 Hz?
2) how do I account for the direction of vibration effect ie would my test be valid if I just said that the device is vibrating at a frequency of 166 Hz if I placed one motor on the base plate? What about it's effect in all 3 axes? I am guessing a 3 axis accelerometer would give me some indication of that?

 

[Mech_Engineer]

1) if I place 4 of these motors on my base plate would it create a series effect ie would my frequency then be 166*4 Hz?

No, it would still be 166.67 Hz. The magnitude (acceleration) of the vibration could in theory be higher, but the 4 motors would probably all be out of phase and it's just as possible they would cancel each other out.

2) how do I account for the direction of vibration effect ie would my test be valid if I just said that the device is vibrating at a frequency of 166 Hz if I placed one motor on the base plate?

You have to describe what direction the vibration is in. My guess is your motor has a weighted flywheel and will be vibrating the system in a circular motion, so 2 axes. What exactly are you trying to "test"?

What about it's effect in all 3 axes? I am guessing a 3 axis accelerometer would give me some indication of that?

A 3-axis accelerometer might help, but it needs a fast enough data acquisition rate to see 166Hz, which means it would need to be taking data faster than that (say around 500Hz).

 

[siddy88]

Okay sounds very pertinent. I should have mentioned at the start that the engineering specification for this is that the fixture needs to have a natural frequency of greater than 100Hz. Considering all the advice you have given me, would I be wrong to say that I can just put one motor providing a 166.67 hz vibration and see if resonance occurs?

 

[AlephZero]

There are two basic ways to do vibration testing.

One way is to shake the structure at a fixed frequency and measure the response. To find the resonant frequences, you have to vary the shaking frequency and see when the response is biggest. From your posts, it isn't very clear if your motors will run at variable speeds or if you can control the speed easily.

You can measure the response with an accelerometer, or in simple cases you can just look at the structure and see at what frequencies you get a large (resonant) response.

The other way is to excite the structure with every frequency at once, measure how it responds against with time with an accelerometer, then do a Fourier transform of what you measure to find the resonant frequencies. Exciting the structure with every frequency at once is easy, you just tap it with a hammer.

If you want to produce a "proper" frequency response curve, you would also need to measure the hammer's force input against time using a load cell, but if you just want to know the resonant frequencies you can forget about that.

Another tip: computer sound cards make good digital-analog converters for "cheap and cheerful" vibration testing, if you are interested in frequencies between about 30 Hz and say 10 kHz. (The theoretical limits for a standard sound card would be 0 to 22 kHz, but the accuracy and sensitivity falls off at the ends of the range). Just connect your accelerometer to a suitable amplifier, feed it into an audio input to your computer, and record the "audio" as an uncompressed .wav or .aiff file. (Don't convert it into an mp3!)

 

[Mech_Engineer]

If I was going to find the natural ferquencies of a structure, I would use FEA. I have no experience measuring such a thing in "real-life," so I defer to Aleph's judgement... I just finished a project where we had to make sure the structure's natural modes were no lower than 20Hz (seismic concerns), we used FEA to confirm this.

I do know however that just exciting the structure at 166Hz will not tell you if your lowest natural frequency is above 100Hz. If I had to try, I would do what Aleph mentioned- whack it with a hammer and record the audio frequency response. Any spikes in the data below 100Hz will tell you whether you hit the mark or not.