# Testing Accelerometers

1. Dec 3, 2013

### Branny12000

Hi there I want to come up with an experiment to test two accelerometers to see how much they vary. I was thinking about dropping them each from a height but I am unsure how I could then work out what values I should be getting. Also does it matter what height I drop them from as I realise it would take some time to reach 9.8m/s2. Also would it be easier perhaps to drop them down a slide. I am very new to this area and so I am unsure about the calculations that would be needed and how they would relate to the readings which are in counts. ie 1g, 2g etc

Any help is much appreciated as I am a bit lost
Bran

2. Dec 3, 2013

### voko

The best way would be by attaching them to a horizontal wheel, which you can spin with an accurately controlled angular speed. Then using the equation for centripetal acceleration you can check how far off they are at various speeds.

I would not recommend dropping, because it is too difficult to control, plus that can damage them permanently.

3. Dec 4, 2013

### Branny12000

Hi there,

Well I was given a protractor and a piece of string so I suppose I could always do a simple experiment of fastening the accelerometer to the piece of string and allowing it to swing from different angles. So then for example if I allowed it to drop from 45 the maximum reading I should get it 1g in the horizontal direction (note I am testing tri-axial accelerometers). Is that correct?

4. Dec 4, 2013

### voko

Accelerometers measure proper accelerations, i.e., the accelerations that deviate them from free fall. In a pendulum-like set-up, those would be the accelerations due to the tension of the string; so they would be directed along the string, not horizontally (unless the angle of swing is greater than 90).

If your purpose is to compare two accelerometers, then you are not interested in absolute values of indicated accelerations. You just need to check whether they are the consistently close.

5. Dec 4, 2013

### mrspeedybob

Just sitting still, the accelerometer should read gravitational acceleration of 9.8 m/s2.

If they are directional you can tilt them to get any reading between 0 and 9.8 m/s2

6. Dec 5, 2013

### Branny12000

Thanks so much for your help. I see what you are saying about the string. The trouble is I don't know where I can get a horizonal wheel that I can alter the speed of at will.

7. Dec 5, 2013

### Branny12000

Ps so in the string experiment the accelerometers should be measuring the tension in the string counteracting free fall hence if it is let go from 45 degree it should be gx(cos45)?

8. Dec 5, 2013

### voko

I do not know what your resources are. In principle, an electric motor with some frequency control would do it.

If you are very constrained, then the pendulum setup is probably the best you can have. You just need to make sure that you attach the accelerometers in the exact same way during all the trials, and record their output. You won't be able to calibrate them absolutely, but at least you can compare them to one another.

9. Dec 5, 2013

### voko

That will depend on the initial and the current position. A lot will depend on how much air resistance there is, that will also be felt as something pushing them from free fall.

10. Dec 5, 2013

### Branny12000

The other option I was thinking about was putting the accelerometer on a toy car and rolling it down a slide however, again I'm not sure how to work out the acceleration that would be experienced

11. Dec 5, 2013

### voko

Why do you need to know the acceleration up front? If you only need to compare, why bother?

12. Dec 5, 2013

### Branny12000

Well we would also like to see how far off the actual value it is. It is not a definite consideration but osmething that we would ideally like to know also.

13. Dec 5, 2013

### voko

That is complicated because you would need to take into account many things that you cannot easily control: air resistance, friction, varying gravity, and even centrifugal and Coriolis forces! That is why the controlled wheel is much preferred: it gives you a constant speed despite all those things, meaning a constant acceleration.

14. Dec 6, 2013

### davidm1732

I'm assuming you're using accelerometers that can measure down to dc? If they're piezo based sensors, they're designed for ac signals, so won't respond well to the relatively constant type accelerations you're discussing here. You'd need an oscillating acceleration..... rotating imbalance, a sine wave played through a speaker or shaker, a vibrator etc....The followup is how are you measuring the output from the accelerometer...do you have a fixed readout, or are you putting a voltage signal into an oscilloscope or FFT analyser? If the latter, then you're looking at an AC signal.

Do you have the make/model # of the accelerometer?