Dual-Axis Accelerometer

[naor52]

I got the Acceleration sensor working (if people are still having problems with this, PM me, i'd be happy to help) but the values i'm getting are odd....

when sitting still the value returned by Get_Analog_Value(rc_ana_in09) is around 716-717, and can get to about 915 or so when shaked.....

is this normal?

how can i transofrm these to normal 0-255 values? should i?

Thanks!

[Noah Kleinberg]

In my experience with it, that's normal.

Just like with the gyroscope, you'll need to first calculate the bias and then subtract that from all of your readings to get accurate numbers.

[sciguy125]

when sitting still the value returned by Get_Analog_Value(rc_ana_in09) is around 716-717, and can get to about 915 or so when shaked.....

is this normal?

how can i transofrm these to normal 0-255 values? should i?

Thanks!

I'm sure that you need to figure out the bias. However, you also might want to make sure that it's level. Remember that gravity will register on an accelerometer.

As for converting to 8bit, just shift it two bits to the right

8_bit = 10_bit >> 2;

[KenWittlief]

shaking it doenst provide any usefull information.

use gravity as your test acceleration - first tip it one way (9.8M/S^2 down) then rotate 90 degrees (0 M/S^2) then tip it up the other way (9.8M/S^2 up).

Use those readings as your reference / zero points.

[Kevin Watson]

I'm sure that you need to figure out the bias. However, you also might want to make sure that it's level. Remember that gravity will register on an accelerometer.

Here's my first-cut at code that will give an angle relative to the gravity vector, which points down. I'm working on a calibration routine for the bias,
but for now, just assume it's at the 2.5v mid-point. This code will work with the ADC code on my website. I imagine this code would work great for knowing what angle your ball launcher was at.

-Kevin


int Accel_Angle(void)
{
int x_axis;
int y_axis;
int angle;

x_axis = (int)Get_ADC_Result(X_AXIS_CHANNEL) - X_AXIS_BIAS;

y_axis = (int)Get_ADC_Result(Y_AXIS_CHANNEL) - Y_AXIS_BIAS;

angle = (int)(ANGULAR_CONVERSION * atan2((float)y_axis, (float)x_axis));

return(angle);
}




Here are the constants:


// Analog channel on the robot controller that's hooked-up to
// the x-axis accelerometer
#define X_AXIS_CHANNEL 1


// Analog channel on the robot controller that's hooked-up to
// the y-axis accelerometer
#define Y_AXIS_CHANNEL 2

// 800Hz sample rate & 16 samples averaged per update (see adc.h)

#define X_AXIS_BIAS 2048

#define Y_AXIS_BIAS 2048


// Pick the angular unit by removing the // from one of these two lines.
// #define MILLIRADIANS
// #define TENTHS_OF_A_DEGREE
#define DEGREES


#ifdef MILLIRADIANS
#define ANGULAR_CONVERSION 1000.0// milliradians per radian
#endif

#ifdef TENTHS_OF_A_DEGREE
#define ANGULAR_CONVERSION 573.0// tenths of a degree per radian
#endif

#ifdef DEGREES
#define ANGULAR_CONVERSION 57.3// degrees per radian
#endif

// function prototypes
int Accel_Angle(void);

[Biff]

The info is in the manual for the controller and in other places. Here goes anyway, the A/D converters on the robot control have 10 bit resolution, the code snippet above shows how to get rid of the lest significant bits. To get 0 to 254 numbers. The A/D converters on the operator interface (OI) are 8 bit, no need to bit shift them to get to 0 to 254. If you are comparing the numbers from say from a pot on the IO and one on something attached to the robot it's good to keep in mind that the a/d converters have this difference.

[Joe Hershberger]

Here's my first-cut at code that will give an angle relative to the gravity vector, which points down. I'm working on a calibration routine for the bias,
but for now, just assume it's at the 2.5v mid-point. This code will work with the ADC code on my website. I imagine this code would work great for knowing what angle your ball launcher was at.

-Kevin


If you guys decide to use the accelerometers to measure angles on your bot, keep in mind that the accelerometer will sense all accelerations in its axis. If you are moving or getting bumped or if your machine shakes violently, then you will pick up those accelerations as well as gravity, and the angle you compute will be less precise due to that noise. This will almost definitely keep you from shooting well on the run or under duress.

Good Luck!
-Joe

[KenWittlief]

what Joe said^

and remember the magnitude of gravity doesnt change - so if your sensors are reading more that 9.8M/S^2 (vector reading) that tells you the bot is accelerating, not sitting still

[Goldeye]

I got the Acceleration sensor working (if people are still having problems with this, PM me, i'd be happy to help) but the values i'm getting are odd....

when sitting still the value returned by Get_Analog_Value(rc_ana_in09) is around 716-717, and can get to about 915 or so when shaked.....

is this normal?

how can i transofrm these to normal 0-255 values? should i?

Thanks!
No one here gave a particularly helpful explanation unless you have a good understanding of the inner workings of the code, so here ya go. :confused:
Get_Analog_Value returns an int value from 0-1024, rather than the unsigned chars. This is based on the voltage returned along the signal wire from the accelerometer.
The readings from the accelerometer pointing straight up with it resting on a nonmoving surface should be equivalent to an acceleration of zero.
When you turn the accelerometer onto its side, one of the values it returns should change. The difference between this value and the original one represents 1g of acceleration, and can be used to scale all the readings.
Precisely how to do that in code is for you to figure out.

By the way, using a timer and low level access to the ADC (Analog-digital converter), it is possible to sample the device more than once per loop, and closer to if not more frequently than the rate than device updates itself. The adc code on http://kevin.org/frc can help you do this. Use the gyro code as an example, or wait until kevin uploads his new accelerometer code. (Soon, please ;) )

On a related note, the specs for the accelerometer have it working at 3V. Is there a way to change the power supply to that?

[Tatsu]

No one here gave a particularly helpful explanation unless you have a good understanding of the inner workings of the code, so here ya go. :confused:
Get_Analog_Value returns an int value from 0-1024, rather than the unsigned chars. This is based on the voltage returned along the signal wire from the accelerometer.
The readings from the accelerometer pointing straight up with it resting on a nonmoving surface should be equivalent to an acceleration of zero.
When you turn the accelerometer onto its side, one of the values it returns should change. The difference between this value and the original one represents 1g of acceleration, and can be used to scale all the readings.
Precisely how to do that in code is for you to figure out.

By the way, using a timer and low level access to the ADC (Analog-digital converter), it is possible to sample the device more than once per loop, and closer to if not more frequently than the rate than device updates itself. The adc code on http://kevin.org/frc can help you do this. Use the gyro code as an example, or wait until kevin uploads his new accelerometer code. (Soon, please ;) )

On a related note, the specs for the accelerometer have it working at 3V. Is there a way to change the power supply to that?


I dont think you need to "change" it to v3.3.. thats what U1 on the board is for i think, cant find the datasheet for it its a MCP something something (microchip power IC probably) anyway.. if you want to "change" the 5v output to 3.3, or 12 to 3.3 just use a voltage divider.

[Joe Hershberger]

On a related note, the specs for the accelerometer have it working at 3V. Is there a way to change the power supply to that?
The accelerometer's data sheet is on the US FIRST website here (http://www2.usfirst.org/2006comp/Specs/Analog_Devices_Ultracompact_Dual-Axis_Accelerometer.pdf) and is the best source of information on the device. Page 3 tells you that the power supply can range from 2.4V to 5.25V. The datasheet does list some example specs when you assume that the supply is 3V. If you look at the bottom of page 9, there is a section entitled USING THE ADXL311 WITH OPERATING VOLTAGES OTHER THAN 3 V that should shed some light on the effects of a different supply voltage.

Cheers!
-Joe

[Joe Hershberger]

I dont think you need to "change" it to v3.3.. thats what U1 on the board is for i think, cant find the datasheet for it its a MCP something something (microchip power IC probably) anyway.. if you want to "change" the 5v output to 3.3, or 12 to 3.3 just use a voltage divider.
In general, it's not a good idea to use a voltage divider in a power supply for a device. If you use large resistors, then any current you draw into the device will cause a larger voltage drop across the upper resistor and lower your supply voltage. If you use relatively small resistors, then you can reduce that drop, but you then have a low total resistance across your battery. This will drain your battery and make your resistors very hot. You should use a voltage regulator designed specifically for power supply applications. For relatively low currents < 500mA, you can just use a monolithic voltage regulator like an LM7805 (for a 5V regulator). LM7812 is the 12V version. You get the idea. :D A search on Digi-Key should yield plenty of results.

Cheers!
-Joe

[railerobotics]

Here's my first-cut at code that will give an angle relative to the gravity vector, which points down. I'm working on a calibration routine for the bias,
but for now, just assume it's at the 2.5v mid-point. This code will work with the ADC code on my website. I imagine this code would work great for knowing what angle your ball launcher was at.

-Kevin


int Accel_Angle(void)
{
int x_axis;
int y_axis;
int angle;

x_axis = (int)Get_ADC_Result(X_AXIS_CHANNEL) - X_AXIS_BIAS;

y_axis = (int)Get_ADC_Result(Y_AXIS_CHANNEL) - Y_AXIS_BIAS;

angle = (int)(ANGULAR_CONVERSION * atan2((float)y_axis, (float)x_axis));

return(angle);
}




Here are the constants:


// Analog channel on the robot controller that's hooked-up to
// the x-axis accelerometer
#define X_AXIS_CHANNEL 1


// Analog channel on the robot controller that's hooked-up to
// the y-axis accelerometer
#define Y_AXIS_CHANNEL 2

// 800Hz sample rate & 16 samples averaged per update (see adc.h)

#define X_AXIS_BIAS 2048

#define Y_AXIS_BIAS 2048


// Pick the angular unit by removing the // from one of these two lines.
// #define MILLIRADIANS
// #define TENTHS_OF_A_DEGREE
#define DEGREES


#ifdef MILLIRADIANS
#define ANGULAR_CONVERSION 1000.0// milliradians per radian
#endif

#ifdef TENTHS_OF_A_DEGREE
#define ANGULAR_CONVERSION 573.0// tenths of a degree per radian
#endif

#ifdef DEGREES
#define ANGULAR_CONVERSION 57.3// degrees per radian
#endif

// function prototypes
int Accel_Angle(void);


In this code what is "atan2"? When I compile this code it gives me a prototyping error. I deleted the line with "atan2" and it compiled just fine. I can't seem to find "atan2" anywhere else in the code.

[Kevin Watson]

In this code what is "atan2"? When I compile this code it gives me a prototyping error. I deleted the line with "atan2" and it compiled just fine. I can't seem to find "atan2" anywhere else in the code.Um, before posting, perhaps you could just google "atan2()".

-Kevin

[railerobotics]

Sorry, Thanks. I still can't figure out why it is giving me a protoype error on the line that has atan2.

[doubleslash]

Sorry, Thanks. I still can't figure out why it is giving me a protoype error on the line that has atan2.

#include <math.h>

As a rule, any sort of math function has a good chance of being in "math.h," again, nothing a quick Google search can't confirm (http://www.opengroup.org/onlinepubs/007908799/xsh/math.h.html).

[Team 614]

Um, before posting, perhaps you could just google "atan2()".

-Kevin
PWNED>!!!!!11!!lll1!l!11!one!!11!

[railerobotics]

Here's my first-cut at code that will give an angle relative to the gravity vector, which points down. I'm working on a calibration routine for the bias,
but for now, just assume it's at the 2.5v mid-point. This code will work with the ADC code on my website. I imagine this code would work great for knowing what angle your ball launcher was at.

-Kevin


int Accel_Angle(void)
{
int x_axis;
int y_axis;
int angle;

x_axis = (int)Get_ADC_Result(X_AXIS_CHANNEL) - X_AXIS_BIAS;

y_axis = (int)Get_ADC_Result(Y_AXIS_CHANNEL) - Y_AXIS_BIAS;

angle = (int)(ANGULAR_CONVERSION * atan2((float)y_axis, (float)x_axis));

return(angle);
}




Here are the constants:


// Analog channel on the robot controller that's hooked-up to
// the x-axis accelerometer
#define X_AXIS_CHANNEL 1


// Analog channel on the robot controller that's hooked-up to
// the y-axis accelerometer
#define Y_AXIS_CHANNEL 2

// 800Hz sample rate & 16 samples averaged per update (see adc.h)

#define X_AXIS_BIAS 2048

#define Y_AXIS_BIAS 2048


// Pick the angular unit by removing the // from one of these two lines.
// #define MILLIRADIANS
// #define TENTHS_OF_A_DEGREE
#define DEGREES


#ifdef MILLIRADIANS
#define ANGULAR_CONVERSION 1000.0// milliradians per radian
#endif

#ifdef TENTHS_OF_A_DEGREE
#define ANGULAR_CONVERSION 573.0// tenths of a degree per radian
#endif

#ifdef DEGREES
#define ANGULAR_CONVERSION 57.3// degrees per radian
#endif

// function prototypes
int Accel_Angle(void);


When I build this code I get a syntax error and it points me towards the bracket under int Accel_Angle(void). What's wrong?

[TubaMorg]

LOL Ok yall....quit taking code snippets and trying to compile them. Take the snippets with grace and appreciation for the fact that there's people here that can figure this stuff out. It is YOUR responsibility to adhere to proper C syntax to get it to compile. ORRRR you can wait for Kevin Watson to finish tweaking it and supply the entire source on his site. We should ALL thank him for his work. Without him half the teams would be lost! Hip hip....hoorayyy! :)

Oh in case it wasn't clear from my post, you need to add in the stuff that makes it a compilable C program (i.e. Math.h and appropriate brackets). Look at the rest of your working code for guidance. You could put the constants in Accel.h then the function blocks in Accel.c or something.