MOSS – Motion smart senzor

Block SymbolLicensing group: ADVANCED

Function Description
The MOSS block implements an advanced filter for incremental (quadrature) position sensors. The block requires special hardware, as correct operation necessitates knowledge not only of the current value from the sensor but also the timestamp of the last pulse, the direction of movement at the last pulse, and the timestamp of the reference moment (from the same source as the pulse timestamp). The output of the block is not only the filtered position but also the velocity and acceleration. For proper operation, it is necessary to appropriately select the alpha parameter. A smaller value reduces noise but increases the signal delay.

If no pulse is received in the stalled time interval, sensor is considered stopped and the outputs (pos, vel, acc) are set to 0. If pos is greater then maxpos, the internal position processed by the Kalman filter is decremented by an integer multiple of maxpos and incremented back for output. This causes the filter algorithm to calculate small enough numbers and not reduce accuracy due to rounding errors. The default value should not normally be changed. If no pulse is received for a long time, the predictor output will drift. To overcome this drift, if no pulse is detected for longer then mindivert time, the output position is clamped to $\pm$ 1 pulse from input (measured) position.

Note 1: It may seem impossible to determine the position more accurately than the quantization error ( $\pm$ 1 pulse) of the measurement, but knowing the velocity allows for a more accurate estimation of the position. Furthermore, it may appear that velocity cannot be determined more accurately than as a ratio of the number of pulses and the difference in timestamps, but by plotting both signals, it is evident that the MOSS signal improves. Determining acceleration through differentiation leads to utterly unusable values.

Note 2: Internally, the block implements a Kalman filter for a system with 2 integrators (i.e., input acceleration, output position). The filter is discretized anew in each period, where the discretization period is the current difference in timestamps. To derive the filter, it is necessary to know the input and output disturbances. If we consider white (Gaussian) noise, it suffices to know their ratio, which is the alpha parameter.

This block does not propagate the signal quality. More information can be found in the 1.4 section.

Input

tsPulse	Time stamp of last pulse	DWord (U32)
cntPulse	Last state of pulse counter	DWord (U32)
tsSync	Time stamp of synchronization pulse	DWord (U32)
cntCorr	Last pulse correction	Double (F64)
flags	Input status flags (1: POS, 2: NEG, 4: RUN)	DWord (U32)
R1	Block reset	DWord (U32)

Parameter

freq	Frequency of source time stamp [Hz] $↓$ 0.0 $⊙$ 100000000.0	Double (F64)
stall	Time for activation of halt state [s] $↓$ 0.0 $⊙$ 0.08	Double (F64)
alpha	Design parameter of Kalman filter $↓$ 0.0 $↑$ 200.0 $⊙$ 26.0	Double (F64)
maxpos	Rounding error optimization of Kalman filter $↓$ 0.0 $⊙$ 1e+10	Double (F64)
mindivert	Time for activation of predictor diversion limiter [s] $↓$ 0.0 $⊙$ 0.003	Double (F64)

Output

pos	Filtered position	Double (F64)
vel	Filtered velocity	Double (F64)
acc	Filtered acceleration	Double (F64)
status	Output status flags (1: POS, 2: NEG, 4: RUN, 8: INIT, 16: PULSE, 32: STALLED, 64: DIVERT)	Long (I32)
iE	Error code	Error

[Previous] [Back to top] [Up] [Next]