I've become accustomed to using accelerometers as tilt sensors (i.e. measure component of acc. due to gravity) but then I came across a sensor category on RobotShop named "tilt sensors" separate from their accelerometers: http://www.robotshop.com/tilt-sensors-1.html
Are these sensors fundamentally different?
I couldn't find any info on their operation, so a quick explanation of the difference would suffice.
The "digital" tilt switches are essentially mercury switches or rolling-ball switches. Picture a glass tube with a ball bearing or blob of mercury, and two contacts at one end; tilt it in the right direction and the metal shorts the contact just like any other switch.
There are a couple of technologies for non-accelerometer analog tilt sensors, but seeing as how they're selling for about $80 (!), I see no reason to consider them in modern designs. Essentially, they are what was used before MEMS accelerometers became possible, common, and cheap.
I see. Surprising these things are still in production then.
Accelerometers haven't been cheap for very long.
The older tilt sensors were probably mostly used in relative expensive applications where users would get mad if they stopped being available very quickly; they probably don't trust the MEMS stuff.
A "digital" tilt sensor (on/off) has less functionality than an accelerometer but is a lot easier to interface to when you just want to know "tilt or not".
Now, an analog tilt sensor is generally designed to be very accurate and have known tolerances on tilt angle, when it's really important. The "$80 one", for example, guarantees 0.5 degree accuracy over a +/-45 degree range. I'm not sure an accelerometer-based solution can guarantee that (they're pretty noisy devices).
--
The Aussie Shield: breakout all 28 pins to quick-connect terminals
To clarify, i was asking about analog tilt sensors, not the binary tilt switches that happen to be on the same page i linked
In retrospect, those sensors look like they could simply be accelerometers with an added microprocessor and packaging so that they output a nicely filtered "tilt angle" rather than just g-forces. My impression was that the fluid based inclinometers I was referring to were generally larger...
westfw:
... those sensors look like they could simply be accelerometers with an added microprocessor ...
That was my first though exactly and this is no different than packaging a 2 or 3 axis magnetometer with a microcontroller and sell it as a tilt compensated compass.
Another example would be to take a $2 microcontroller, add a freeware floating point library and sell it as a $20 math co-processor.
Not that I blame anyone in any way – this is “stepping up the food chain” and if you only need a couple (boards or IC’s) it is hard to beat time/cost wise (assuming they put in the effort to actually match the specifications).
There are other methods of measuring tilt angles that have been in use for a long time. I recall one method published in Scientific American article years ago, that could detect tidal force changes, it was that sensitive! It used two small cups of mercury 3" in diameter, with a common liquid connection line between them about 2-3 feet apart. They applied a single crystal controlled 4Mhz oscillator signal to both cups and attached a small 1" disk 'Antenna" just above the surface of each cup. Then the two 'received' signals were rectified/filted and compared in a rf amplitude detection circuit and the generated DC difference voltage was amplified and sent to a slow moving DC voltage chart recorder. You could clearly see the two tide peaks and valley for each 24 hour period. That was an amazing demonstration of how sensitive an instrument even a hobbyist could build. It sounded like it takes some time to setup correctly, balanced, etc, and needed to be mounted on it's own stable earth foundation points. It mentioned passing truck traffic could interfere with the normal tidal signal!
Lefty
RuggedCircuits:
Now, an analog tilt sensor is generally designed to be very accurate and have known tolerances on tilt angle, when it's really important. The "$80 one", for example, guarantees 0.5 degree accuracy over a +/-45 degree range. I'm not sure an accelerometer-based solution can guarantee that (they're pretty noisy devices).
What tilt sensor can do this?