mattg31
Posts: 61
Joined: Fri Jan 05, 2018 9:55 pm

Vibration Analysis

Fri Apr 13, 2018 10:46 pm

Wondering if anyone can provide any advice or experience on choosing an appropriate sensor for measuring small vibrations on a machine. The vibration I am trying to measure is essentially a single "thud" or pulse every 5 minutes or so. I can feel this pulse with my hand when I press my hand on the machine, but it is fairly faint.

I have purchased this peizo:
https://www.digikey.ca/product-detail/e ... ND/3911239

I have connected this peizo to this ADC:
https://www.digikey.ca/product-detail/e ... ND/5761229

Using a raspberry pi 3 Model B, and the items above, I can sense large vibrations (banging on the machine with my hand), but it is not sensitive enough to pick up my target vibrations. Even with the gain on the highest setting on the ADS1115.

Is an accelerometer a better choice perhaps?
Or am I on the right track but need to amplify before the ADS1115?

Any feedback/advice is appreciated.

Thanks!

Idahowalker
Posts: 445
Joined: Wed Jan 03, 2018 5:43 pm

Re: Vibration Analysis

Fri Apr 13, 2018 11:59 pm

A high impedance amplifier will be needed for the sensor.

I figure the pulse you are trying to detect is very quick so you may need a sample and hold circuit.
Without knowing why you are deleting my postings, I will not know how...

mattg31
Posts: 61
Joined: Fri Jan 05, 2018 9:55 pm

Re: Vibration Analysis

Sat Apr 14, 2018 12:53 pm

Thanks for the reply @Idahowalker.

So if I was to use this circuit:https://imagebin.ca/v/3yM5Db0FFIa5
reference: http://www.ti.com/lit/ds/symlink/lf198-n.pdf

Would it be a good start to:
Connect the peizo to the Vin pin (with other peizo pin connected to ground)
Resistor R1 would have ground on the bottom pin, and +5v on the top pin

Assuming R1 is adjusted correctly, when the vibration occurs, the Vout pin will briefly go high?
Am I understanding this correctly?

Thanks for the help!

Idahowalker
Posts: 445
Joined: Wed Jan 03, 2018 5:43 pm

Re: Vibration Analysis

Sat Apr 14, 2018 4:53 pm

After thinking more about this matter, I would attach the vibration sensor to the unit and using an o-scope, monitor for the incoming signal in question. You would be able to determine the characteristics of the signal you are looking to capture, such as amplitude and periodic occurrence. It may take hours or days to capture enough transient signal before its electrical characteristics can be worked out. Also, you will catch other signals that you might need to discriminate against.
Without knowing why you are deleting my postings, I will not know how...

mattg31
Posts: 61
Joined: Fri Jan 05, 2018 9:55 pm

Re: Vibration Analysis

Sat Apr 14, 2018 7:22 pm

Ok, Will do. I will report back when I hopefully figure this out!
thanks again

Idahowalker
Posts: 445
Joined: Wed Jan 03, 2018 5:43 pm

Re: Vibration Analysis

Sun Apr 15, 2018 1:35 am

mattg31 wrote:
Sat Apr 14, 2018 7:22 pm
Ok, Will do. I will report back when I hopefully figure this out!
thanks again
As a note, good choice for the sample and hold component. I'd not get a sample n hold component yet; wait for the signal analysis results.
Without knowing why you are deleting my postings, I will not know how...

User avatar
Gavinmc42
Posts: 3151
Joined: Wed Aug 28, 2013 3:31 am

Re: Vibration Analysis

Sun Apr 15, 2018 2:13 am

Film piezos are generally not as sensitive as ceramic ones but they are not as brittle.
You can capture ADC data in a circular buffer but also have an amplifier/comparator to set a trigger threshold.
That way you can capture the pretrigger waveform as well.

Machine vibrations can be in the audio range so you can use audio recording gear/methods.
Stick an electret microphone in tiny sealed box and use 24bit audio ADC?
You need to mount sensor firmly to equipment too.
Drill out bolt head for electret microphone?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

mattg31
Posts: 61
Joined: Fri Jan 05, 2018 9:55 pm

Re: Vibration Analysis

Mon Apr 16, 2018 1:30 pm

Gavinmc42 wrote:
Sun Apr 15, 2018 2:13 am
Machine vibrations can be in the audio range so you can use audio recording gear/methods.
Stick an electret microphone in tiny sealed box and use 24bit audio ADC?
You need to mount sensor firmly to equipment too.
Drill out bolt head for electret microphone?
Yes great idea also! I actually just purchased an electret microphone for this reason, I'm hoping I can pick up the audio instead. Otherwise I will go back to the peizo film idea.

On a somewhat unrelated note, and I'm sure this is posted in another forum someplace, but i'll ask anyway :D
I'm using this:https://www.sparkfun.com/products/12642
And it states it works better (more stable outputs) with a 5v VCC. However this means the analog output will be in the 5V range, which is a problem for my ADC.
My plan is to just use a 3.3v zener diode on the output. Does that sound reasonable?

Idahowalker
Posts: 445
Joined: Wed Jan 03, 2018 5:43 pm

Re: Vibration Analysis

Mon Apr 16, 2018 3:58 pm

mattg31 wrote:
Mon Apr 16, 2018 1:30 pm
I'm using this:https://www.sparkfun.com/products/12642
And it states it works better (more stable outputs) with a 5v VCC. However this means the analog output will be in the 5V range, which is a problem for my ADC.
My plan is to just use a 3.3v zener diode on the output. Does that sound reasonable?
Using a zener will 'work' but will cause clipping of all the signal information above 3.3V.

A voltage divider will give some signal loss but will be a lot simpler over an active circuit.

The formula you want to work with will be Vout= Vin(R1/(R1+R2). The ratio of R1 to R2 would be 1K (R1) to 2K(R2). R1 and R2 would be wired in series; with the top of R1 being the input analog signal, the connection point of R1/R2 (Vout) will be the input to the ADC and the bottom of R2 would go to the common ground.

Consider the ADC input impedance. Using too low of an R1/R2 value could fall outside of a impedance range of the ADC.

When you have selected the physical resistors (1%) ohm them out and use their actual metered measured values in the formula; to check that the R1/R2 junction will not exceed 3.3 volts with the supplied resistor values.

You might, also, consider a good quality small value cap attached from the R1/R2 junction point to the input of the ADC.
Last edited by Idahowalker on Mon Apr 16, 2018 5:37 pm, edited 1 time in total.
Without knowing why you are deleting my postings, I will not know how...

mattg31
Posts: 61
Joined: Fri Jan 05, 2018 9:55 pm

Re: Vibration Analysis

Mon Apr 16, 2018 4:24 pm

Ok will do, thanks for the detailed reply @idahowalker, much appreciated!

gambafrolla
Posts: 1
Joined: Fri Jun 21, 2019 3:23 pm

Re: Vibration Analysis

Fri Jun 21, 2019 3:28 pm

hello!

I'm very interested to this discussion!

I'm interested in doing a spectral analysis with a raspberry. The target is to monitor vibration anomalies in a machinery for maintenance purposes.

Any news on your project @mattg31?

Avimesa
Posts: 3
Joined: Thu Jun 27, 2019 9:34 pm

Re: Vibration Analysis

Thu Jun 27, 2019 9:59 pm

You could use one of these boards: https://avimesa.com/hardware/
with an industrial vibration sensor like: https://www.automation24.com/vibration- ... TAQAvD_BwE

That sensor would be plug & play with the board. Avimesa has a message broker built in so you could use a simple app to display the data and send alerts.

This tutorial is for a temp sensor, but would work for vibration too
https://avimesa.com/docs/scripts-and-ap ... oring-app/

Return to “Automation, sensing and robotics”