#### Equip your Genius

Welcome to the Parallax Discussion Forums, sign-up to participate.

# Accelerometer Question

Posts: 1,102
Hi All...

Going through my parts bin recently, I came across a half-wired up 3-axis accelerometer. That reminded me of a question from when I was tinkering with it some time back.
Does the chip generate any level of electronic noise. I mean if the chip were operating on the classic immoveable object, would there be any variation in the outputs?

Of course, when it was running on my desk there was always a small amount of motion displayed and, for my purposes, I just filtered it out. Just wondering if the innards of the accelerometer might be making a small contribution.

Thanks for any wisdom!!

Amanda

• Posts: 8,470
ajward wrote: »
Hi All...

Going through my parts bin recently, I came across a half-wired up 3-axis accelerometer. That reminded me of a question from when I was tinkering with it some time back.
Does the chip generate any level of electronic noise. I mean if the chip were operating on the classic immoveable object, would there be any variation in the outputs?

Of course, when it was running on my desk there was always a small amount of motion displayed and, for my purposes, I just filtered it out. Just wondering if the innards of the accelerometer might be making a small contribution.

Thanks for any wisdom!!

Amanda

Virtually all electronics will have some thermal or other noise, although it is usually very little in comparison to the expected signal.
• Posts: 6,419
edited 2017-12-15 - 00:57:05
You will have thermal noise with all electronics. However what appears to be random may actually contain useful data. If you take multiple samples, don't average, but instead just accumulate the value. Averaging truncates the data and introduces unnecessary error. By accumulating the values you can increase your signal to noise ratio by the square root of the number of samples you take, while any true random noise has a tendency to cancel itself out. Under ideal circumstances, the LSB will be on 50% of the time and off the other 50% of the time... if there is any bias in one direction or the other, then it's probably a valid part of the signal that most people ignore. In some cases it can be beneficial to pay attention to.
• Posts: 22,540
... By accumulating the values you can increase your signal to noise ratio by the square of the number of samples you take ...
I think you meant square-root, not square, right?

-Phil
• Posts: 6,419

I think you meant square-root, not square, right?

Well yes

-Phil[/quote]

• Posts: 1,102
Thanks guys!
I kinda thought that was the case. Just wanted to be sure.
Didn't do any averaging, just eyeballed to get an idea of the upper end and ignored anything below that.

@