looking for miniature circuit to detect orientation in 3D

[circuitspecialists]

I have used an accelerometer to measure tilt to 1 degree. The units that output a digital valua are easier to interface using an RF link. There are many good application notes and examples available by the manufacturers. Some that I have used are MEMSIC and FREESCALE.

 

[Mr RB]

Why do you keep saying you need a "six pointed object"? A 3 axis tilt detector (as used in cameras) or tiny 3 axis accelerometer will both work to measure the orientation of your "dice" cube. Both are not much bigger than a grain of rice.

 

[en2oh]

6 points represents the faces of the cube. My problem is that I need to know which face is up, at any given point in time.
Nitrous

 

[Mr RB]

That's what a 3-axis accelerometer chip does! It has 3 analogue output pins, and will tell you which axis is experiencing 1g or -1g (from gravity).

 

[en2oh]

So, the accelerometer will experience (at rest), the force of gravity at all times. So, depending on which is experiencing + or - 1G compared to the one that is experiencing 0 G, which is the axis perpendicular to the force of gravity, you know precisely which face is up?
nitrous

 

[en2oh]

Mr Black,
I noticed on your website that you have an entry that looks at the OSRAM optical Tilt Sensor.
An example you give is actually using a cube (you mention dice) with three sensors at a 30 deg incline to address sensitivity (hysteresis) issues.

The device I was proposing would have used a 4 position IR tilt sensor PLUS a dual direction ball tilt switch at right angles to the 4 position IR tilt sensor. Would that not (electromechanically) achieve the desired result?

Nitrous
(missed your CV - when will it be updated?)

 

[en2oh]

Does a PIC have the power to work with ADXL335 three axis acceleration board?
I see arduino code for this device used to determine orientation there actually is a bildr tutorial on the subject.
Nitrous

 

[Mr RB]

Check out this tiny little accelerometer board from Pololu for $12;

**broken link removed**
https://www.pololu.com/catalog/product/1246

You power it with 3.3v, then there are 3 output pins XYZ, and they output an analog voltage that detects gravity of 1G (when the device is not in motion). All you need are 3 ADC input pins on your arduino, and some very simple code.

 

[dr pepper]

I wonder if the o/p got mixed up with a chip that has '6 point' measurement, meaning something like a 6050 that has 3 axis accelerometers and gyro's in the same package.

 

[en2oh]

Mixed Up... for sure
But that wasn't actually what I was trying to say.
If the orientation detection was based on a purely mechanical sensor, ie a ball that makes electrical contact with one of 6 "switch points" that represent the 6 faces of the cube, I had envisioned a wire mess cube, with the centres of each face "pulled out" into points. At each point of the cube faces, there would be a contact point that the electrically conductive ball would be free to contact (between the point contact and the wire mesh of the cube.

Using this approach, gravity would bring the contact ball to the relevant dependent position. The "up face" would always be the face opposite to the switch that was now closed due to the metal ball.

It sounds like the accelerometer solution (especially if X, Y and Z are discrete outputs) might be sufficient although I understand that a 6 DOF sensor would be more stable. (also adding to the complexity of the software required to make sense of the data.

Does that make sense?
Nitrous

 

[dr pepper]

If I think really hard yes, but then it goes again, glad its not my project.

 

[en2oh]

as the saying goes... the craziest thinking makes perfect sense to the fool thinking it, at the time
tnx

 

[jpanhalt]

Let me go back a little to my earlier post (#7) about "absolute spacial orientation."

Is your cube moving (e.g., relative to the earth) when you want to determine which face is up?

I don't recall it being said that the cube is stationary and affected only by the acceleration of gravity. That distinction might make a difference in how you detect which face is up relative to you, earth, or the floor. For example, what if the cube were on a stick being swung around a pivot in the stick and parallel to the floor? Now, the "6" might be up relative to you, earth, or the floor, but the acceleration detected might be almost perpendicular to that reference.

BTW, my earlier comment was not so much about the spelling as about the meaning of "absolute." In the example I just gave, one might define that "up" as an absolute orientation relative to earth.

John

 

[en2oh]

Thanks for the clarification. The cube would likely be stationary (ie sitting on a desk).
Alternatively, the cube would be held. As a side question, any advantage or disadvantage of a sensor with three discrete outputs for X, Y and Z compared to I2C or SPI.
Again the hope is that a PICAXE would be used to bridge the sensor to the RF data transmitter.
Thanks
Doug

 

[jpanhalt]

I2C and SPI are communication protocols. I don't see what that would have to do directly with the ±x,y,z coordinates.

This is just a guess, but considering the size of accelerometers and MCU's compared to the much larger RF links for high data rates (like XBee), I would do everything pre-RF and then just transmit a simple code for which side is up. That is, I would not try to transmit the data and do the analysis at the receiver end. Again, I have played with XBee and Linx. I am not familiar with the really small stuff.

John

 

[en2oh]

The communications protocol question really only dealt with choice of sensors. Some seem to have I2C and others are X, Y, Z outputs. I've read that I2C can be tricky to work with.
Nitrous

 

[en2oh]

There is an accelerometer module from Freescale that offers embedded sensor orientation detection.
Anyone have any experience with this type of feature? The link to the notes on the feature is below.

https://cache.freescale.com/files/sensors/doc/app_note/AN4068.pdf?fsrch=1&sr=1#page4

Any thoughts on ease of this feature compared to say an analogue X, Y, Z sensor that compares neutral axis to the one that is either plus or minus?

Thanks
Doug

 

[crutschow]

There is an accelerometer module from Freescale that offers embedded sensor orientation detection.
Anyone have any experience with this type of feature? The link to the notes on the feature is below.

https://cache.freescale.com/files/sensors/doc/app_note/AN4068.pdf?fsrch=1&sr=1#page4

Any thoughts on ease of this feature compared to say an analogue X, Y, Z sensor that compares neutral axis to the one that is either plus or minus?

The digital output of the Freescale device may allow a simpler interface to indicate the orientation of the device. But it does have an I[SUP]2[/SUP]C digital I/O which probably is easiest with a microprocessor that has a built-in I[SUP]2[/SUP]C interface.

 

[en2oh]

I was thinking about using a PICAXE chip, which does have [LATEX]I2C[/LATEX] support.
I was hoping for something that relatively easily would give specific orientation of an at-rest sensor.

Thanks for your comments
Nitrous

 

[crutschow]

I was thinking about using a PICAXE chip, which does have [LATEX]I2C[/LATEX] support.
I was hoping for something that relatively easily would give specific orientation of an at-rest sensor.

I'm not familiar with the PICAXE but I would expect it should not be too difficult to generate a program to decode the output of the chip and make a decision, based upon that output, as to which side is up.