Firmware design

AN3288

12/27

Doc ID 18011 Rev 2

if( Sample_Vector[k] > max ) max = Sample_Vector[k];

}

/* calculation of the moving average value */

mov_avg = ((subtotal - min - max)/8);

Figure 9 shows the representation of the acceleration in a straight track, and subsequent

deceleration during the braking phase. Obviously, these characteristics depend on the type

of vehicle used for modeling the system.

As shown, the noise corrupts the acceleration/deceleration signal captured by the

AIS326DQ and causes fluctuations from the original values; therefore the noise can be

removed by averaging several adjacent samples.

Figure 9.

Acceleration raw data and moving average filtered data

The filtering algorithm dramatically reduces the presence of noise, even though it slightly

reduces the amplitude of output signal; to overcome this drawback, the deceleration

thresholds defining the braking effort have been set opportunely.

Concerning latency, the filter introduces a delay in the order of some milliseconds; this delay

affects the sampling rate of the entire system even if it is not so critical for the feasibility of

the application.