MikroElektronika Forum
https://forum.mikroe.com/

Acelerometers
https://forum.mikroe.com/viewtopic.php?f=13&t=14413
Page 1 of 2

Author:  rspadim [ 07 Apr 2008 09:12 ]
Post subject:  Acelerometers

hello guys, i'm tring to understand the acelerometer
i will use it in a car
what's my problem?

i'm driving from 0km/h to 50km/h with 4 seconds, ok i get 12.5km/h/second = 13.8m/s/4= 3.472 m/s/s
if i drive in a horizontal line that's ok, the earth gravity will not change my horizontal aceleration,
but if i get a montain terrain where the horizontal and vertical line changes the force of earth gravity, and this change the acelerometer axis aceleration

the problem in less words:
how could i cancel the earth gravity on acelerometer inside a car?

one insue, i can travel from 0meters to 1000meters in a montain, so the gravity can change from 9.8 to 9.82 or less i don't know, but the earth gravity can change on travel.

any idea? i can't use internet to know the gravity of the gps position :( :P hehe

Author:  rspadim [ 07 Apr 2008 09:27 ]
Post subject: 

i was thinking
if i get the diference of aceleration?
for example
today my acell is 9
acelerating i get 9.1
my "real" aceleration is .1
after some time i get 9.1 again and my "reak" aceleration is 0 :(
could i implementing something that work with this?
any idea?

Author:  rspadim [ 07 Apr 2008 09:43 ]
Post subject: 

i think that my problem is workarounded by wii remote control, so i think that we can have a solution

Author:  drdoug [ 07 Apr 2008 18:51 ]
Post subject: 

I am working on a project using angle displacement not acceleration but I think similar principles apply.

I believe you would be best served by getting a 2(x-z) or 3(x-y-z) axis accelerometer. First you determine your middle point on flat ground with 0 acceleration (except gravity)(~ 30 samples averaged should work)(x_zero and z_zero).

In order to compute your velocity on a hill, you need to compute your angle at a constant velocity (0 acceleration). To get an accurate reading, you need the z and x axis.
x = x_zero-x_sample
z = z_zero-z_sample

x_angle = atan2(x,z)*pi_rad;

When you are accelerating you then take your measurement(s) (~30 again) and compute the resultant vector. You then just take the vector difference and that should give you your acceleration(multipli.

I don't have time now to work out the rest of the solution for you but I hope that gets you started.

Does this make sense? Anybody?

Author:  rspadim [ 07 Apr 2008 20:08 ]
Post subject: 

i was thinking something like
start engine, get the "0" time aceleration informations and start measuring the forces without the "0" time acelerations

could it work?
any idea?

Author:  drdoug [ 08 Apr 2008 01:11 ]
Post subject:  Re: Acelerometers

rspadim wrote:

the problem in less words:
how could i cancel the earth gravity on acelerometer inside a car?



Ans: Put the car in space.

I misread your question. I thought you wanted to know straight acceleration on an angle, not fluctuating gravity.

I think it is just a matter of reference(for angle or a changing g).
Pick your reference point(sea level) and then subtract the difference. I don't know if there is enough sensitivity to measure this since they seem to be affected by voltage and temperature so I don't think you'll be able to tell a .01 difference reliably.

Author:  rspadim [ 08 Apr 2008 07:48 ]
Post subject: 

i was thinking, about the problem...
if i get an inclinometer, put all sensors near wheel (the suspension oscilation will not act on sensors, just disc temperature), after this i get aceleration and inclination of the sensors, and subtract a value acquired when car was stopped or a situation of no brake being applied and no agressive aceleration on speed (using wheel speed sensors)?

think that the sensors have a good temperature correction

could we implement this?
the aceleration of car with a acelerometer and some others sensors??

Author:  Acetronics [ 08 Apr 2008 09:23 ]
Post subject: 

Hi,

You should see : :wink:

FREESCALE : AN 3461
MEMSIC : AN 00MX-012

http://www.dimensionengineering.com/app ... Gmeter.htm

as a good starting point

Alain

Author:  rspadim [ 18 Apr 2008 07:21 ]
Post subject: 

ok i checked, maybe only with an acelerometer can make the solution,

maybe when all wheel speed=0
i could get the acelerometer information and set the gravity aceleration

gravity = sqrt(x*x+y*y+z*z)

and after i can get aceleration of car with

car aceleration=sqrt(x*x+y*y+z*z)-gravity

so, i have another problem now
what's the aceleration direction? and gravity aceleration?

now a inclinometer should be used?
inclinometer give me the gravity position?!
so i could set the x,y,z gravity values,
removing it from acelerometer x,y,z information i could get car aceleration

i checked that the most powerfull car ellica have 0.8g of aceleration more than porche, and when we use good brakes we can get 0.5g of aceleration

so a 2g acelerometer or 3g should be used

i'm right? could anyone help?

Author:  drdoug [ 18 Apr 2008 14:54 ]
Post subject: 

rspadim wrote:

maybe when all wheel speed=0
i could get the acelerometer information and set the gravity aceleration
Yes, this is your center point.


rspadim wrote:
gravity = sqrt(x*x+y*y+z*z)

and after i can get aceleration of car with

car aceleration=sqrt(x*x+y*y+z*z)-gravity

You will really only need to worry about x direction for acceleration (postitive forward and negative braking). The y direction is if you are going around a curve and your butt slides on the seat.

rspadim wrote:
so, i have another problem now
what's the aceleration direction? and gravity aceleration?

now a inclinometer should be used?
inclinometer give me the gravity position?!
so i could set the x,y,z gravity values,
removing it from acelerometer x,y,z information i could get car aceleration
The accelerometer can be used as a inclinometer as long as there is 0 acceleration.


rspadim wrote:
i checked that the most powerfull car ellica have 0.8g of aceleration more than porche, and when we use good brakes we can get 0.5g of aceleration


so a 2g acelerometer or 3g should be used
The 2g will give the best resolution.

Try the demo that was listed earlier and many of your questions will be answered in the process. I think the "hill" effect will be negligible but you can work on that once you get a prototype running with just the x direction.

Author:  drdoug [ 18 Apr 2008 16:46 ]
Post subject: 

I have been working with a prototype board of the MMA7260 and you should be able to find them for <$20 on ebay. 3 axis!

Author:  Acetronics [ 19 Apr 2008 16:34 ]
Post subject: 

you should :!: have a look here ...

http://www.elektor.com/magazines/2007/a ... 1064.lynkx

Alain

Author:  rspadim [ 23 Apr 2008 00:14 ]
Post subject: 

the url is broken, another site is open without accelerometer informations, do you have another url?

Author:  Acetronics [ 23 Apr 2008 07:47 ]
Post subject: 

Sorry, URL is NOT broken ... I just have tried 5 sec ago.

just look at the FREE DOWNLOAD section ...

Alain

Author:  MarkoZ [ 23 Apr 2008 15:50 ]
Post subject: 

Accelerometers are usually uncalibrated and are not temperature compensated. This means that if you want 0.01g resolution/precision you have to put a lot of effort. You need to find a way to calibrate at two different temperatures and then correct the reading from it in real time based on it's current temperature. I have used KXP74 3-axis accelerometer from Kionix. With 1C temperature variation you can get 1-3mg of error reading. Also noise is about 5mg. Uncalibrated accelerometers can have anything from -0.15 to +0.15g error (even if data sheet tells they should be better than that). Also you have non-orthogonality problems, meaning that even if it has each axis separately calibrated and standing still, sum of accelerations on all three axes will not be 1g because axes are not perfectly aligned. Usually X and Y axis are normal to each other, but Z axis is off-plane, not normal to XY plane.

If you don't calibrate your self and provide some mechanism for temperature compensation and orthogonality correction your goal should be more like 0.05 to 0.10g precision. Try ti integrate that error during 1min, if you miss acceleration by 0.05g you will have a position error of almost 900 meters, and velocity error of 100km/h.

Put some numbers on paper and check if it's possible to do what you are trying with currently available accelerometers.

I didn't meant to scare you with this, just to give you some insight on this since I'm working with accelerometers for automotive applications for more than a year already.

Good luck!

Page 1 of 2 All times are UTC + 1 hour
Powered by phpBB® Forum Software © phpBB Group
https://www.phpbb.com/