| Prev | Next | Start of Chapter | End of Chapter | Contents | Index | (13 out of 25)

First-Order Exponential Filter


The First Order Exponential Filter performs low-pass filtering to filter out high frequency noise. Its output value depends on the previous output value and the current input value. To specify how much the filter weights previous output values, set the attribute Filter Constant. The larger the Filter Constant, the more weight the previous output values have.

This filter does not require inputs that are equally spaced in time.

Filtering

The First-Order Exponential Filter uses this equation to compute its output value. The variables in the equation are explained in the table that follows.


This variable Is
outputn
The current output value
outputn-1
The previous output value
input
The current input value
time
The difference in time between the arrival of the last input value and the current input value

The coefficient is the amount of weight that the filter gives the current input, and the coefficient (1­) is the amount of weight that the filter gives the previous output value. The block computes with both the Filter Constant and the difference in time between the arrival of the last input and the arrival of the current input. The constant is computed such that values received long ago will not have as much effect on the current value as ones that were received a short time ago.

Specifying How to Round Output Values

To round the output values, set the field Quantization. The block rounds its output value to the unit you specify. For example, if Quantization is 0.1, the block rounds to the nearest tenth, and if Quantization is 1.0, the block rounds to the nearest integer.

This table shows some examples of rounding.

If Quantization is... The filter passes these values
none
0.5
1.43
1.77
0.1
0.5
1.4
1.8
0.5
0.5
1.5
2.0
1.0
1.0
1.0
2.0

This field is especially useful when you need to keep output values within known accuracy limits.

Configuring

This is the configuration panel for the First Order Exponential Filter.


Attribute Description
Quantization
A number that specifies the smallest increment that the block uses when rounding output values.
Filter Constant
The amount of weight that the filter gives the current input over the previous input.

Example

This figure shows the chart of a First-Order Exponential Filter with a Filter Constant of 5. The raw signal has more variation and the filtered signal is smoother.


See Also

For more information on how to use this block, see the sections below.

Click here for more information...
Basic Block Behavior

| Prev | Next | Start of Chapter | End of Chapter | Contents | Index | (13 out of 25)

Copyright © 1996, Gensym Corporation, Inc.