Electronic – LSM303DLHC low-power and high precision modes

First of all, the comment regarding your intended application is very important. For this answer, I'm going to assume the you want the highest data rate possible (i.e. Wii-type controller that sends near-instantaneous data). I give you two options below. In both cases, the receiver is simple: receiver board (either another XBee or the sister board of the transmitter you are using) connected to a USB-serial converter connected to your computer.

Option 1: Really easy to build, not-so-low power

An XBee 1mw Chip Antenna can be configured to transmit at whatever datarate you need, has a max. range of 100m, and, best of all, has six on-board ADCs. You can configure the XBee to automatically sample these and transmit their values. X, Y and Z from the accelerometer can go directly into AI0, AI1 and AI2 pins on the XBee and they can both be fed with 3.3V.

• Good: You only need a battery, the accelerometer and the XBee to transmit, and another Xbee to receive.
• Bad: The XBee has 45mA peak current when transmitting.

Option 2: Not so easy to build, low power

The transmitter board you mentioned can be powered by 3.3V (according to the datasheet). Alternatively, you could use this 2.4GHz transciever. In both cases, you will need a microcontroller (easy) or an ADC with serial output (harder, will need a PCB, something along the lines of the MAX1245) to convert the analog signals from the accelerometer to digital signals for the transmitter. I recommend you use a microcontroller (Arduino Pro Mini 3.3V for prototyping, then just an ATMega328 when you're done with the design).

• Good: Low power (7-10mA).
• Bad: Will cost a lot more design time and burnt fingers.

Option 3: Stereo FM transmitters modulating 3 channels of data

This is in response to "Can I do it without converting the signals?".

1. Use resistors and op-amps to bring the accelerometer's readings down to 90mVrms.
2. Use two of these stereo FM transmitters, hook up X to Rin, Y to Lin, and Z to Rin of the other transmitter.
3. Configure the transmitters to Tx on different frequencies.

On the receiver side:

1. Aquire the FM signals and demodulate back to right and left audio. You could use something like this.
2. Amplify and offset the three channels (right and left channel on one, and only right on the other) to measurable voltages (op-amps would work, but I have a feeling we just added a ton of noise to the whole thing).
3. Plug these signals to any microcontroller (Arduino, if you're feeling lazy) with ADCs and send them out as serial data to your computer.

NOTE: Most FM tuners will have band-pass filters that will screw this whole idea up.

In order to compensate for temperature drift, you need to place your accelerometer in a temperature chamber (e.g. these) and concurrently record both temperature and raw acceleration values while slowly changing the chamber temperature over the range of interest (ideally about ~1°/minute). If your accelerometer is well placed, you know that the true value you should be measuring is [0, 0, -9.81 m/s] and any deviation has to be due to temperature drift. Based on this, you can apply some kind of fit to the measured data to build a calibration that you subsequently use to predict the drift based on the measured temperature.

My experience is the following:

• A polygonal fit of order 3 is enough to implement the calibration. Depending on the temperature performance of your particular sensor (I haven't tested it), a linear fit may be sufficient.
• The sensor gain (the factor that convert from raw value to G) is relatively immune to temperature, but the sensor offset (additive value) is temperature dependant.
• For sensor types that are prone to temperature drift, there is a high degree of sample-to-sample variation. You should thus consider calibrating every single parts.
• More recent parts, such as the LIS3DH, exhibit much lower temperature drift as older parts. This particular one could go without temperature calibration for most applications but the most demanding.