BMA400 Watermark Issue

[Mari]

Hi to all,

I have issue with my BMA400, the sensor is managed by a PIC18 on I2C bus. I configured the sensor by following the official example code on GitHub. 

The configuration of the sensor is:

- ODR=25Hz => 25frame/s => 175bytes/secondo @12-bit di risoluzione
- range=2g
- normal power mode set
- interrupt active low, push-pull on INT1 pin

- active FIFO watermark on x,y, z axis, watermark=217bytes ( => 31(frame)x7), NO sensortime, teaming mode enabled.
- set latching mode (interrupt cleaned after status interrupt read out)
- FIFO_WM_INT is enabled and mapped on INT1 pin

The init phase follow the diagram above:

One flush cmd is sent before to enable the interrupt.

Every time the FIFO wm trigger the interrupt, a dedicated routine was called, the flow chat of is:

the operations are more or less the same on example code done in the example file (https://github.com/boschsensortec/BMA400_SensorAPI ).

And now the problem that I see:

- if I use ODR>25Hz, after the first FIFO wm complete acquisition, the other are incomplete, by increasing the ODR, the lost frame are increasing too, but with the oscilloscope I can see the change of the pin state;

- so, i fixed the ODR to 25Hz, but I  notice that the accelerations are very similar for every fifo wm, for example in picture below, in left side the sensor is upside-down compared with the position (and acceleration) on right side

- after the extraction and print of the accelerations, a flush cmd is sent and after that I must repeat the FiFO configuration, otherwise I can't see a new interrupt.

I also used as refernce the documents "How to use BMA400 FIFO.pdf".

I'm sure that the problem is on FIFO configuration, but I don't know/understund what I'm missing or what I'doing wrong. I hope, someone in the community could help me....

To exclude an hardware problem, I also simulate a FIFO wm by using "data ready" with interrupt and in my opinion, it is working, I can modify the configurations and I read correclty the acceleration data when the sensor is moved in all the directions!

Thanks to all,

Mari

[BSTRobin]

Hi Mari,

I setup BMA400 FIFO watermark example code for your reference, hope it is helpful for you.

It runs well on my side:

[Mari]

Hi BSTRobin,

thanks for your sample code. I used it to compare the mine and understand the problem, at the end I have find out the problem in my code: in the library of the sensor, I modified one routine because, as it is implemented, it fail on compiling phase.

The routine is "bma400_get_regs()" and the problem is the declaration of the local array "uint8_t temp_buff[(len + lun_dummy)] ", in which the dimension  is fixed by using a variable, this is not allowed with XC8 compiler and I fixed it to 100bytes. So, it is not enought for all data, with a bigger array, data are correct and it are alligned with the position of the accelerometer. 

Marilena