The RMS noise of LIS3MDL and LSM303D

[looeyc]

Hi,

We had been developing some geomagnetic measuring products with LSM303D. Since the LSM303D had been discontinued, we had to switch to an alternative part. The LIS3MDL seemed to be a good choice. However, during prototype verificating, we found the LIS3MDL has a much higher noise level. As the following pic shows, the RMS noise of each axis is about tow times higher than LSM303D.

According to the datasheet, LSM303D and LIS3MDL should have the same RMS noise performance, which should be around 5 mgauss. Why there is so much difference? Is there any suggestions if we want to get a same or even better noise performance with LSM303D?

By the time, the configurations of these two parts are almost the same: 3.3V power, FS=12gauss.

With Regards

[looeyc]

Is there anyone can help me?:grinning_face:

[Eleon BORLINI]

Hi looeyc, they should perform similar, you're right... Were you capable to test other parts after the one you shared the results?

Btw, I would suggest to use LIS2MDL (instead of the LIS3MDL): LIS2MDL has a 3-mgauss-rms noise in high resolution mode, wider FS and is more stable temperature-wise.

regards

[looeyc]

Hi, we have tested a MKI137V1 evaluation board from ST, which is the evaluation board for LIS3MDL . The RMS noise was still rather high. So it should not be a board design problem.

At the same time, we have tested a MKI185V1 evaluation board, which is for the IIS2MDC. The IIS2MDC has a 1.5 mgauss/LSB sensitivity and is a new part in ST's longevity program, while the LIS3MDL is 0.438 mgauss/LSB with a full scales of 12 Gauss. However, we were a little disappointed, since the result shows the IIS2MDC has almost the same RMS noise level with LIS3MDL after sensitivity alignment. By contrast, the LSM303D was always much better.

​The attached excel file is our test data. The gray parts are the orignal data read out from sensors. You can find out the difference if compare the red part data.

We wil have to find another sensor IC, ​If this problem could not be resolved.:worried_face: