Hi,
we are currently integrating the BMA400 accelerometer in a sensor product and have some questions regarding the FIFO watermark.
Is it safe to set a new FIFO watermark without a FIFO flush using a multi write operation while the sensor is in normal mode i.e. measuring and filling up the FIFO? Will the watermark be evaluated during a multi write operation to the two fifo watermark registers FIFO_CONFIG1 and FIFO_CONFIG2? The main concern is that BMA400 evaluates the watermark while only one byte of the new fifo watermark e.g. FIFO_CONFIG1 is written and an undesired watermark interrupt is triggered with some inconsistent watermark value.
Datasheet p.35 is also confusing:
Watermark interrupt condition is also updated after the end of the serial interface (burst write)
transaction which wrote into the registers fifo_watermark<10:8> or fifo_watermark<7:0>.
The behavior of the FIFO watermark is shown in the figures below.
because according to datasheet p.106 there is no "burst write" operation:
Multiple write operations are possible by keeping CSB low and continuing the data transfer. Every data
must be preceded by R/W flag and address, there is no address auto-increment like in burst read
mode.
So I don't think there is a "burst write" operation e.g. for setting the 11bit watermark or is there a write operation that does auto address increment?
