bjornhi 
Hi fellow freaks,
I need for an application a simple sensor that tells the orientation of the device. Simply, if the user rotates the product 180 degrees the unit should detect it and turn the LCD graphics accordingly.
What is the simplest way of doing this? Are there any cheap gravity sensors that can digitally tell were the earth is?
Any input welcome!
Kind regards
Bjorn
bjornhi Realizing now that a normal acceleration sensor could be enough? Any tips on ICs to use?
/Bjorn
Brutte If that is one device then use any acceleration sensor you like, possibly with +-2g range. And if it is >1000 pieces then use a light sensor.
mckeemj 
You don't define what "cheap" is, that can change things. But, I'm a fan of the Analog Devices ADXl345, ADXL346 and the new ADXL362. They are all designed for just the sort of thing you're looking at doing. Plus, they have lots of other useful features built-in.
Martin Jay McKee
In addition to digital accelerometers, if the device is truely only in one of two directions, and money is an issue, then you could also consider a caged ball sensor. This has a small metal ball inside a box, (cage), with electrodes to tell whether how the box is oriented. It needs manual debouncing, and is mechanical, but in large quantities it probably costs less than a MEMS digital accelerometer.
JC
bjornhi Thanks for all the quick replies.
Brutte: How would the light sensor logic work? Assume that light direction is up?
mckeemj: Well, the cheaper the better :-) Analogs devices look nice but costs much.
DocJC: Can you recommend any caged ball sensors?
Best bid so far is Freescale MMA7660FC. http://cache.freescale.com/files/sensors/doc/data_sheet/MMA7660FC.pdf
/Bjorn
Farnell/Element14 have the freescale kl25z frdm board with xtrinisic sensors for about $25. You get two accelerometers, baro sensor and magnetometer to play with. And it works with mbed. Job done.
If you just want one, I've got a bag full of these:
http://www.ebay.co.uk/itm/6DOF-MPU-6050-3-Axis-Gyroscope-And-Accelerometer-Module-3-3V-5V-For-Arduino-New-/290940385072?pt=UK_ToysGames_RadioControlled_JN&hash=item43bd65f730
I could throw one in an envelope for you.
John
bjornhi John, thanks for the kind offer. But I'll pass as I would need 1k-2k to cover my needs.
/Bjorn
bobgardner A 2 axis sensor gives tilt +-45 deg in pitch and roll using arcsin(); A 3 axis sensor gives full 360 deg in pitch and roll using atan2(); Does your gizmo pitch and roll a complete circle?
Right. So I'm at least 976 short...
John
Brutte Assume that light direction is up?
if the user rotates the product 180 degrees
Bjorn,
I was looking for a source for a caged ball sensor and can't find one at the moment. I have, however, actually seen them. Perhaps they were a custom part for that particular device?
In any event, if you go to Mouser's site and search on caged ball switch there are a number of Mountain Switch devices.
These appear to all be single axis "rolling ball" switches, so two would be needed for X/Y quadrature encoding, three for adding the Z axis.
Note that they seem rather expensive, especially if several are needed. Hence making the MEMs accelerometer chip look better and better.
JC
Brutte What about making a round, beefy metalized via in a PCB (hole with inner metalization)? Then both PCB sides openings can be covered with SMD parts (uC in QFN, resistor ladder etc) to make the ball stay inside. A stainless ball cannot be soldered during process and it does not corrode in typical environment conditions. Gold flash is also a standard PCB process. I am not sure about the influence of the PCB cleaning process and the dust during usage..
So the inner metalization sides could be divided into N equal parts/contacts (N=3 for example) and the grounded QFN bottom pad and resistor ladder contacts could be used as the remaining two contacts for top and bottom..
Then a dumb scanning procedure would easily find out which contacts were shorted..
Would a bullseye level with some segmented metalization on one surface and full metalization on the other do the trick? The air bubble dielectric constant would change the capacitance between the metalization as it moves. Refraction could also be measured as could inductive coupling if the bubble were a metal ball. Of course costs could quickly get out control.
Bjorn,I was looking for a source for a caged ball sensor and can't find one at the moment. I have, however, actually seen them. Perhaps they were a custom part for that particular device?
In any event, if you go to Mouser's site and search on caged ball switch there are a number of Mountain Switch devices.
These appear to all be single axis "rolling ball" switches, so two would be needed for X/Y quadrature encoding, three for adding the Z axis.
Note that they seem rather expensive, especially if several are needed. Hence making the MEMs accelerometer chip look better and better.
JC
I do not know of any American distributors for them, however.
valusoft 
I have used a tilt switch from www.signalquest.com for starting a musical toy. Arrange a few appropriately and it might work as a gravity sensor.
bobgardner 100s of millions of phones have 3 axis accelerometers in them. Have we rejected this best practice for some reason?
100s of millions of phones have 3 axis accelerometers in them. Have we rejected this best practice for some reason?
I have used a tilt switch from www.signalquest.com for starting a musical toy.
valusoft No mercury. Rolling ball.
stevech Hi fellow freaks,I need for an application a simple sensor that tells the orientation of the device. Simply, if the user rotates the product 180 degrees the unit should detect it and turn the LCD graphics accordingly.
What is the simplest way of doing this? Are there any cheap gravity sensors that can digitally tell were the earth is?
Any input welcome!
Kind regards
Bjorn
