I'm trying to learn arduino programming but it can be frustrating.
I want to measure tilt angle in °'s. One or two axis tilt. to 1/10 degree.
I've bought freescale accelerometers, an adxl335 & 345, mpu6010, mpu9010, memsic2125, a 10dof board and some others. I get readings but not reliable angle measurements. I'm not even sure all the "G" readings are accurate.
There's so much to do to log the data, monitor on an LCD or OLED screen, send control signals.
But I'm spending 99% of my time trying to get reliable angular readings.
My question: Is there an easy way to measure tilt, in degrees, with a small sensor.
I've figured out how to work my OLED screen, reading and writing SD Cards, but I'm getting nowhere with the angles. I only need 1 reading every 2 or 3 seconds.
Maybe I am missing something. But I sure could use some advice.
First, you have to calibrate the accelerometer so that the "g" readings are correct. In general there is an offset and scale factor for each axis.
Second, if you look at the measurement noise figures for those devices and work out the consequences, the very best you can do is about +/- 1 degree accuracy in a single measurement.
Averaging hundreds to thousands of readings should allow you to do better, as long as the sensor has been calibrated as carefully and accurately as possible.
I want to measure tilt angle in °'s. One or two axis tilt. to 1/10 degree.
Imagine a perfectly rigid bar 100 metres long.
Now lift one end up by 17.5cm. (about seven inches, for the metrically-challenged)
That's the kind of precision you're asking for.
Hi,
I don't think people realize the reason all these inclinometers and accelerometers have come on the market so cheap is because of the wii, xbox, playstation etc needing fast attitude sensors for their hand-controls.
They are cheap because;
They make them in the 10s (100s) of millions
The games manufacturers want a cheap sensor.
Crude because they don't have to be accurate.
The games manufacturers want a cheap sensor.
I have played with a mpu6050 model and it is good, but I'd only use them to measure g forces.
In my line of work we use industrial inclinometers to level platforms and they are not cheap even for basic 0.5Deg accuracy.
Tom.... 
If big, slow and limited incline range is not a problem one can think at some kind of pendulum-based mechanical "amplifier" of the incline.
Or a pendulum + load cell. Think of a miligram-scale weight sensor that measures horizontal push of an (almost) vertical heavy pendulum.
Ive been measuring angles for over 20 years, With a tool developed by Sperry in the 80's. A one axis 2" dia. disc 7/8" thick that measures capacitance differences in a gas filled chamber. It has ±0.05° accuracy over a ±60.0° range with 0.1° resolution. But they're so big. I use the Cline Labs version. It's thinner.
At level Development I can get a $60.00 2 axis MEMS sensor that measures ±0.05° accuracy over a ±90.0° range at 0.002° resolution. Way better than I've been using. But they put it in a 2" x 2" x 5/8" box with a pigtail. Measurement Specialists has similar. But they cost so much and they put them in such large boxes!
I was just hoping that I am missing something. I'm not even getting the accuracy out of my Memsic 2125 that the Sears Craftsman Digital Level gets using the same sensor. I expect that what I am looking for is out there but I cant find it. Maybe I should be using something other than Arduino? Or an outboard ADC or something. I just don't know where to go next. Just looking for ideas.
I'm not even getting the accuracy out of my Memsic 2125 that the Sears Craftsman Digital Level gets using the same sensor.
I suspect you have not tried careful calibration and lots of averaging.