Arduino / Gumstix i2c Tutorial:
Accelerometer noise:
"A 1 μF tantalum capacitor (CS) at VS and a 0.1 μF ceramic capacitor
(CIO) at VDD I/O placed close to the ADXL345 supply pins is used
for testing and is recommended to adequately decouple the
accelerometer from noise on the power supply. If additional
decoupling is necessary, a resistor or ferrite bead, no larger than
100 Ω, in series with VS may be helpful. Additionally, increasing
the bypass capacitance on VS to a 10 μF tantalum capacitor in
parallel with a 0.1 μF ceramic capacitor may also improve noise.
Care should be taken to ensure that the connection from the
ADXL345 ground to the power supply ground has low impedance
because noise transmitted through ground has an effect similar
to noise transmitted through VS. It is recommended that VS and
VDD I/O be separate supplies to minimize digital clocking noise
on the VS supply. If this is not possible, additional filtering of
the supplies as previously mentioned may be necessary."
Noise on old version SEN-10321. In case any of you have the same noise problem, Ill try to save you some time. I have encountered about 30 counts of noise on the accelerometer. Spent a lot of time trying to “fix” my i2c I/F software which is fine. Turns out there is a nasty ripple on the 3.3v at the i2c clock freq (400khz in my case). Added a tantalum cap at the regulator (parallel the existing), and it knocked the noise way down to just 2 or 3 LSB.
Now to see why the magnetometer is wandering about.
What size cap did you use? I’m running into the same issue (noise levels of 5-12LSBs @ 15Hz ODR). The data sheet for the ADXL345 indicates that a 1uF tantalum cap should be used to decouple Vs from PS noise. Add 10uF tantalum cap in parallel if that wasn’t enough. I’m planning on putting in an order for some caps in that size range (0603 package, which is what the board uses) to try and clean up the measurements.
Also, fwiw, I’m not seeing significant noise on the magnetometer except on the Z axis (with the board mounted vertically).
i can fully confirm that. i also had quite some noise, especially on the accel. i added 3x 10uf 0805 ceramic caps parallel to the existing ones. improves every sensors reading.
I also added some caps. I added a 1uF cap in parallel with the existing 0.1uF caps used on each of the three major chips between Vdd and Ground. On the accelerometer, this dropped the self-noise from ~5.5 counts RMS down to ~1 count RMS. On the magnetometer, the self-noise dropped from ~3 counts RMS to ~1 count RMS. And on the gyro, there was no change (self-noise ~0.6 counts RMS). IMHO, Sparkfun should consider using larger Vdd caps in future revisions so that self-noise is minimized.