2012-09-12 07:22 AM
Hi all,
I am using accelerometer & gyrometer in LSM330DL to build my own IMU. I want to ask about scale factor that is given in LSM330DL datasheet, it is really strange. When I use accelerometer, FS is 01 --> scale factor is 2mg/digit. However, when sensor stands horizontally, the acceleration in z-axis is 16g !!! I realize that if value on each axis is divided by 16, the result is correct. So, how about gyrometer? FS is 00 --> scale factor is 8.75mdps/digit. Do I need modify this scale factor (like accelerometer)? Thank you. #lsm330dl2012-09-12 11:11 AM
When I use accelerometer, FS is 01 --> scale factor is 2mg/digit.
How do you calculate that ? According to the datasheet, FS 01 means +-4g, so signed short (16 Bit) at 8g range means 0.12mg / LSB.
2012-09-12 07:09 PM
Thank fm,
The scale for accelerometer is 0.12mg/digit, but in datasheet it is 2mg/digit !!! For gyrometer, the ST support team replies that 8.75mdps/digit is correct. I do not know what the meaning of 2mg/digit in datasheet.2012-09-12 11:23 PM
I did not find a statement saying ''2mg/digit'', I only used the spec. data of fullscale range
(+-2g/+-4g/+-8g/+-16g) and data size (16 bit two-complement). This resulted in the stated 0.12 mg resolution per LSB. If I remember correctly, there was a large ''Preliminary'' printed across the datasheet...2012-09-13 03:13 AM
Dear Customer,
Thanks for your interest in STMicroelectronics MEMS products.
The problem for the accelerometer is that the data is on 12 bit left aligned. So, to take the data in mg it's necessary to shift the data on the right by 4 (or simply divide by 16) and than multiply by the sensitivity.
Instead, for the gyroscope, the data is on 16 bit.
Kind Regards,
ST MEMS Support Team.
2012-09-13 03:53 AM
So it's clear finally.
I can't remember having seen any notice of 12 bit data for the accelerometer. The register map just declared H and L byte, without any further comment. Admittedly, I did not read through the whole spec. in all detail. Perhaps the left-alignment makes it easier to pump up the 12 bit two's-complement value in hardware. I usually prefer right-aligned data in this case.