# A couple of questions on accelerometer specs

#### Braeden Hamson

Feb 18, 2016
237
So I've joined my schools Formula SAE racing team. Pretty neat. They've stuck me on the task of building an impact data logger for their car's impact attenuator.

The impact should not exceed 18G max. Taking into account the length of what is in essence a crumple zone, I've calculated the main impact to take place over the course of 1 ms. Or to put it another way 1 KHz.

Here's the datasheet for the accelerometer they gave me.
https://www.nxp.com/docs/en/data-sheet/MMA3201KEG.pdf

On page 3 under output signal you can see "Bandwidth Response" is 400 Hz typical. Also of note I'm following their schematic on page 5, which includes an RC filter. (But that filter has a 15KHz cut off)

I don't fully understand what bandwidth response means, in this sense.

Lastly, since 0 G is 2.5 V and +20 G is 5V can I assume that the scale is 0.125V/G?

#### Harald Kapp

##### Moderator
Moderator
Nov 17, 2011
13,098
Bandwidth Response" is 400 Hz typical
It also states f-3dB in the "symbol2 column which means that the signal from the sensor will roll off above 400 Hz with a 4th order low pass characteristic (see the description on the header of the datasheet). Basically this means you can get a useable signal for events with a frequency less than 400 Hz but you will see events with a frequency greater than 400 Hz only heavily attenuated.
their schematic on page 5, which includes an RC filter. (But that filter has a 15KHz cut off)
This filter is the anti-aliasing filter for the subsequent ADC. It also reduces clock noise, see note 4 on page 5.
I've calculated the main impact to take place over the course of 1 ms. Or to put it another way 1 KHz.
You cannot simply convert the duration of a single event (impact) into a frequency or vice versa. Precisely said a single event has not a single frequency or band of frequencies but an infinite amount of frequency components. That is not useful for your application.
What you will get as output from the sensor is probably a pulse with more or less a square form as the car will decelerate during impact. This pulse probably will show a falling slope as the car decelerates slower and slower due to the impact energy being dissipated within the structure. The pulse will certainly be overlaid by lots of noise. I guess (really only a guess without scientific background) the pulse will be similar to this waveform:

Expect also some undershoot as the elasticity of the structure will let the car rebounce.

Lastly, since 0 G is 2.5 V and +20 G is 5V can I assume that the scale is 0.125V/G?
Yes you can since the description states that the chip has a linear output.

#### moneytide

Apr 20, 2014
5
There was a Bloomberg article about a team that used similar data that you are going to harvest with your racing team impact measurements.

They used the information you are gathering and put it with AI to 3D print the optimal frame

Last edited by a moderator:

#### Braeden Hamson

Feb 18, 2016
237
Thank you very much. My professor gave us one commandment "thou shalt always read the notes." Still true I suppose. Your post will really help me fully digest this particular datasheet. And I think your graph is accurate. We tested the sensor with an oscilloscope (unfortunately we can export data so can't use it for data logging here) but when I hit the board just on the bench the resultant graph looked like yours. We could even see the stick-slip undulations in the graph from the broadboards rubber feet. Which was really cool.

One more question. If you were doing these tests (bear in mind these tests are just so we know what we're in for when we take them to a real testing lab, preliminary testing I think you could call it) would you be happy with this sensor?

#### Harald Kapp

##### Moderator
Moderator
Nov 17, 2011
13,098
would you be happy with this sensor?
Yes, at least for a start. You may find out during use that one or the other parameter may not be the best for your application, but trial and error (at least a bit of) is part of the fun, isn't it?

#### Braeden Hamson

Feb 18, 2016
237
But trial and error (at least a bit of) is part of the fun, isn't it?

Haha that's very true, thanks again

#### davenn

Moderator
Sep 5, 2009
14,150
There was a Bloomberg article about a team that used similar data that you are going to harvest with your racing team impact measurements.

They used the information you are gathering and put it with AI to 3D print the optimal frame

#### davenn

Moderator
Sep 5, 2009
14,150
There was a Bloomberg article about a team that used similar data that you are going to harvest with your racing team impact measurements........................

They used the information you are gathering and put it with AI to 3D print the optimal frame

Have deleted you last post .... was empty other than the quote of my previous comment

you video was still not showing ...... try showing a url link as your embedding isn't working

Dave

Replies
3
Views
646
T
Replies
14
Views
1K
tehtehteh
T
Replies
1
Views
686
B
Replies
2
Views
726
josephkk
J
B
Replies
15
Views
2K
Jasen Betts
J