design of electronic level

[outer_space]

I want to make an electronic level based on an accelerometer. Is this possible to detect changes of 0.1 degree with an accelerometer?

The problem will be to condition the analog voltage of the axises. The accelerometer shows 0G voltage at 1.25V and 1G at 1.61V. To detect 0.1 degree it will need accuracy of x axis sin(0.1) = 0.0017. It must be able to detect 0.0017 * (1.61 - 1.25) = 0.000612V. With a 10 bit adc I would want to measure between 1.25V +/- 0.1V to achieve 0.1 degree resolution. What kind of problems could I run into? Could this be correctly amplified from 1.25V +/- 0.1V to 0-3.3V with an opamp or something else?

Any suggestions or comments are appreciated.

Best regards,
outerspace

 

[allanvv]

Why not use an accelerometer with digital outputs?

 

[outer_space]

Built-in digital outputs of accelerometer is 6 bits over the range of +/- 1.2G, this offers the resolution of less than arcsin(1/2^5) or about 2 degrees. If there was an 8-bit digital accelerometer, arcsin(1/2^7) is about 0.5 degrees. Both options seem insufficient for an electronic level. I am going to try a differential opamp to amplify the signal above and below 1.25V although I suspect there will be problems with noise, offset error, and gain error.

 

[allanvv]

Does your analog output accelerometer actually have the accuracy you need? Also, don't they make tilt sensors for applications like this?

Yeah a dfifferential opamp should work, and sensing 100uV shouldn't be too difficult to achieve after amplification. You'll have to calibrate out the offset (storing the perfectly leveled voltage) and gain error in your application anyway.

 

[outer_space]

Only way to find out is to build it. I don't see how to calibrate the 1.25V offset unless there is a DAC because the opamp needs this voltage reference. There may have to be a DAC or a trim pot for this.