Implications of averaging Strain gauge data for real time applications

[harikkat12]

Hello,

I am novice WIth LabVIEW and cRio.

I am using cRio-9076 with 9205 analogue Input Module for measuring the bending strain from Strain gauge. The strain gauge is on a brake bracket of a quad bike, which is used to calculate the brake force. The strain gauge is connected to a amplifier with a gain of 561. For the purpose Calibrating my strain gauge i am hanging the weights to act as the force on the bracket. when i am measuring the strain output, i am getting lots of noise with data. So i am taking the average of 500 data points. I am using a while loop with 10ms wait for collecting the data to write as array to spread sheet.

I want to use the same strain gauge and data in real time dynamic situation. My question is that what would be the implication taking and using the data as explained above for for a real time situation where the i have to continuously use the data from the strain gauge to calculate the brake force??? I have attached my VI with this.

Also to reduce the noise i am planning to by implementing Low-pass filter. Would it be ok to use 'Butterworth Filtor'??

Any help would be really grateful.

Kind regards

[AndyClegg]

Hello,

Butterworth filter can be fine for the type of realtime filtering you are looking for .....

The suitability of a filter depends upon what you are using the data for and more importantly if the frequencies you want to measure (i.e. that pass through the filter) are sufficiently far from the frequencies of noise you want to filter out and their relative magnitudes. If they are close, then you have to do some clever filtering - e.g. high order, steep roll-off, possibly sacrifice a non-flat passband which will distort your signal a bit at expense of getting rid of as much noise as you can. If not then you can get away with something simple - e.g. butterworth filter.

Remeber design of filter is not just about the filter structure, but it has parameters (chose the wrong corner frequency and you could be filtering out parts of the signal you would like to retain) and dynamics (you may inject some phase shift into your remaining signal - which if used in feedback control could lead to instability). Also, remember your chosen sample time will have an impact on performance.

If you have the luxury of being able to try various things out - start with the simplest filter and only move onto something more complex (higher order) if you need more performance.