looking for miniature circuit to detect orientation in 3D

[Mr RB]

There's a new low-cost accelerometer PCB you can get on ebay, I think it was bout $7. The output is digital I2C using 2 wires, and the 3-axis resolutions were quite low, (only 5bits for each axis if I remember right) so it's simple to use. The chips was specifically designed for phones and cameras to read an orientation.

Pololu and Sparkfun have a more complex I2C accelerometer product that has 12bits for each axis. That will give finer angle sensing but will be a bit more work to interface.

 

[en2oh]

Thanks Mr RB.
Will have a look. Is there an easy answer to why lower resolution is easier to work with?
Thanks again for your thoughts
Nitrous

 

[en2oh]

Took me a while to find it, but I think this might be the module you were thinking of.
The resolution is 6 bit and it has a "Tilt" register to identify up/down, left/right and front/back.
Will have to look at this a little closer. Great size, too!
Thanks again
Nitrous

**broken link removed**

 

[crutschow]

......................
Is there an easy answer to why lower resolution is easier to work with?

Just that if it's more than 8-bits then you need to use double precision (two registers) for each word if you are using an 8-bit microprocessor.

 

[en2oh]

Thanks Crutschow. I had forgotten that from way, way back in my computer sciences days.
Nitrous

 

[Mr RB]

Thanks Mr RB.
Will have a look. Is there an easy answer to why lower resolution is easier to work with?
Thanks again for your thoughts
Nitrous

Yep, Crutschow said part of it. The other part is the data comms speed, I2C sends one bit at a time so reading three axes at 5bits per axis is much faster comms to get a full sample than reading 3 axes at 12bits per axis.

And you save time on the data conversion too as 5bits fits into a byte with no processing needed, 12bits needs to be arranged and formatted and then stored in 2 bytes, which is many times slower.

 

[MrAl]

Hi,

If i understand this project correctly it is supposed to be a 6 sided die that is able to transmit it's orientation at any time out to a host receiver.

There's going to be some dynamic characteristics to think about as well as the static ones that have already been mentioned. If the sensor has to be able to determine the orientation of the cube at any time while it is moving, then probably the best way to go is to simply transmit the information directly from the accelerometer to the host and let the host worry about the processing. The signals are going to take a bit of processing to determine what the cube is doing at all times while it is moving, unless we only need the information after it has stopped moving in which case it's just a matter of measuring the three outputs including their polarities.

But that's just the electrical dynamics too. The mechanical ones include the change in the center of mass as the thing is moving. We cant have that or else the probability of the cube coming to rest on one side will increase for one or more sides and decrease for the rest. So whatever is mounted inside the cube has to have a distribution of mass equal to that of a cube or a sphere. Anything else favors either one or more sides probability of appearing 'up' once the cube stops moving. The addition of mass also has an effect, where ideally the mass added should exactly match that of the mass removed (inside to make space for the electronics). The addition of mass of course may be less of an issue except in the most demanding case where the weight has to be matched to the weight of a standard reference cube. The most important though is the distribution of the mass because we would not want to favor some sides coming up more often than others.
The probabilities are affected more when the unequal mass is comparable to the mass of the rest of the cube. Thus making the rest of the cube heavy relative to the added mass means the probabilities will change less favoring one or more sides less than if the cube was lighter.
For example, if the added mass was shaped like a semiconductor chip 0.1 inch thick and 0.4 inches wide and 0.4 inches long, the best bet would be to mount four of them side by side forming a cube and mount that cube right in the center of the die. Of course there are other possibilities too, such as mounting six chips each one parallel to one face of the die and at an equal distance from the center.

 

[en2oh]

Thanks Mr RB, Mr Al.

Just to clarify, the die will not be subject to rolling motions. The most movement it will experience is in the spectators hands, while they are choosing which number/colored face to place on top. Once they've made that decision, the die will be either on a table surface, covered from view or, remain in the hands, such that the orientation does not change.

I had forgotten the IIC single bit transfers. Although time is not critical in this application, detecting and reporting the orientation within a few seconds is fine.

As an aside, I've found a very small transmitter for this project. The two choices seem to be around AM transmitter with a super regenerative receiver. The other is ASK modulated, both at a data rate of ~3k. Is there any advantage of AM vs ASK in this case? In particular, I'm worried about noise (if I recall correctly, super regen recovers are pretty sensitive but largely at e expense of selectivity). I may be totally off base here so would appreciate your comments.

Thank
Nitrous