Low pass filter software?

Question

I'm looking for digital low pass filter code/library/class for a .net windows forms project, preferably written in c, c++ or c#. I probably need to set the number of poles, coefficients, windowing, that sort of thing. I can't use any of the gpl'd code that's available, and don't know what else is out there. Any suggestions appreciated.

Answer 1

Here is a Butterworth Low Pass filter I wrote for a recent project.

It has some magic numbers as constants that was given to me. If you can figure out how to create the magic numbers with your poles, coefficients, etc, then this might be helpful.

using System;
using System.Collections.Generic;
using System.Text;

namespace Filter
{
public class ButterworthLowPassFilter
{

    //filter fc = 2hz, fs = 10hz

    private const int LowPassOrder = 4;

    private double[] inputValueModifier;
    private double[] outputValueModifier;
    private double[] inputValue;
    private double[] outputValue;
    private int valuePosition;

    public ButterworthLowPassFilter()
    {
        inputValueModifier = new double[LowPassOrder];
        inputValueModifier[0] = 0.098531160923927;
        inputValueModifier[1] = 0.295593482771781;
        inputValueModifier[2] = 0.295593482771781;
        inputValueModifier[3] = 0.098531160923927;

        outputValueModifier = new double[LowPassOrder];
        outputValueModifier[0] = 1.0;
        outputValueModifier[1] = -0.577240524806303;
        outputValueModifier[2] = 0.421787048689562;
        outputValueModifier[3] = -0.0562972364918427;
    }

    public double Filter(double inputValue)
    {
        if (this.inputValue == null && this.outputValue == null)
        {
            this.inputValue = new double[LowPassOrder];
            this.outputValue = new double[LowPassOrder];

            valuePosition = -1;

            for (int i=0; i < LowPassOrder; i++)
            {
                this.inputValue[i] = inputValue;
                this.outputValue[i] = inputValue;
            }

            return inputValue;
        }
        else if (this.inputValue != null && this.outputValue != null)
        {
            valuePosition = IncrementLowOrderPosition(valuePosition);

            this.inputValue[valuePosition] = inputValue;
            this.outputValue[valuePosition] = 0;

            int j = valuePosition;

            for (int i = 0; i < LowPassOrder; i++)
            {
                this.outputValue[valuePosition] += inputValueModifier[i] * this.inputValue[j] -
                    outputValueModifier[i] * this.outputValue[j];

                j = DecrementLowOrderPosition(j);
            }

            return this.outputValue[valuePosition];
        }
        else
        {
            throw new Exception("Both inputValue and outputValue should either be null or not null.  This should never be thrown.");
        }
    }

    private int DecrementLowOrderPosition(int j)
    {
        if (--j < 0)
        {
            j += LowPassOrder;
        }
        return j;
    }

    private int IncrementLowOrderPosition(int position)
    {
        return ((position + 1) % LowPassOrder);
    }

}
}

Keith

Answer 2

Ok, I found out how to get the coefficients you used. I downloaded Octave for windows and ran the butter command (as in MatLab) like this:

[b,a] = butter(3, .4, 'low')

Now I can use this code with other fs and fc parameters.