reztronics

Home - Products - News - Services - Music - About Us

Spartan Bandpass

Senior Project for BSEE
Reza Ghassemi
San Jose State University

Abstract

The goal of this project was to create an easy, efficient and free software to introduce entry-level engineering students to the concept of the frequency domain. It was decided to design an efficient digital filtering software to be used for this educational tool. To this end, students should be able to use the software to emulate passing an audio-range signal through mediums with various bandwidth limitations including, but not limited to, the telephone system.

Introduction

Digital filters have gained respect in the technical community for their efficiency and for having low amounts of noise and accurate cut-off frequency characteristics. In addition, a system implementing a digital filter can easily be reconfigured to solve a wide array of problems without any hardware changes. Analog filters are extremely cheap to make, consisting of op-amps and passive components but lack the advantages that their digital counterparts excel in. As an example reconfiguring a simple analog filter requires replacing several components, making it impractical in a changing environment.

However, digital filtering hasn’t been cheap historically. The programmable chips used by companies such as Texas Instruments, Motorola, and Analog Devices are not extremely expensive, but their development environments are. This makes their use as an educational tool unlikely and costly. In the software field, filtering programs are made in every shape and size. In the audio field in particular, sound editing programs such as Sound Forge, Wavelab, and Cool Edit are extremely popular but also cost up to $500.

For this reason, this software project was started with the aim of developing a free software that educators could use without licensing or permission.

The software had several requirements. First, it had to run off any PC running Microsoft Windows. Second, it needed to work with any sound card. Finally, it needed to be easy to use.

Product Description

The software first records the analog input signal from the computer’s sound card using a Microsoft Media Control Interface (MCI) command. The file can now be played back for the user to hear. It then parses the samples in the resulting wav file and stores them into an array. A spectrum analysis of the signal is performed, and the resulting time and frequency domain signals are displayed to the user. The user then inputs a low and high cutoff frequency and selects the order of the filter. The program then generates the resulting filter coefficients and difference equation. The samples are then passed through the difference equation. Another spectrum analysis is performed and the resulting time and frequency domain signals of the output are displayed. Finally, the output file is stored in wav format and the user is able to play back the file and hear the result of the filtering.

The program was designed using the C language. It’s graphical interface was constructed using the National Instrument’s LabWindows integrated development enviroment.

The Wav Format

The Wav format is a subset of Microsoft’s RIFF file format. It is used for digital audio and supports both uncompressed Pulse Code Modulation (PCM) and compressed data schemes. It is essentially a chunk of data with separate smaller chunks of data inside of it that represent file information and samples. The Wav file is separated into three main areas: a RIFF header, the format chunk, and the data chunk. The full file structure is shown below.

  Offset  Length   Contents

  (Riff Header)

  0       4 bytes  'RIFF'

  4       4 bytes  <file length - 8>

  8       4 bytes  'WAVE'

(Format Chunk)

  12      4 bytes  'fmt '

  16      4 bytes  0x00000010     // Length of the fmt data (16 bytes)

  20      2 bytes  0x0001         // Format tag: 1 = PCM

  22      2 bytes  <channels>     // Channels: 1 = mono, 2 = stereo

  24      4 bytes  <sample rate>  // Samples per second: e.g., 44100

  28      4 bytes  <bytes/second> // sample rate * block align

  32      2 bytes  <block align>  // channels * bits/sample / 8

  34      2 bytes  <bits/sample>  // 8 or 16

(Data Chunk)

  36      4 bytes  'data'

  40      4 bytes  <length of the data block>

(Begin Audio Samples)

44      4 bytes  <sample data>

48              4 bytes  <sample data>

...

Windows MCI Commands

The Windows Media Control Interface is a high-level suite of Windows commands that let the software talk directly to the media devices in the computer. The advantage of the MCI suite is that it communicates with the devices that the operating system itself has already designated for that specific purpose. In our case, the sound card has a user-defined default recording input. This could be either a line input or microphone input. The MCI command assumes that the default is the device to use, and hence it eliminates the need for the elaborate code used to access each individual input device. The MCI controls are used in this program to record and save the wav files and to stop and start their playback.

Filter Coefficients and Difference Equations

The coefficients for a digital filter can be found from the following parameters:

analog cutoff frequency

L = damping factor

sampling frequency

Which gives us the following coefficients:

This allows us to filter the signal using the difference equation of the form:

While testing our filter, we came across a simple LabWindows function that completes both the filter coefficient and difference equation problems in one step. The Butterworth Bandpass function takes the following parameters and outputs the filtered samples. Since executing the difference itself took a large amount of time and provided less than accurate results, we chose to use the Butterworth Bandpass function to solve the filter coefficients and execute the difference equation.

Finances

Since the project was aimed at being free, the goal was to prevent its development from becoming costly. Thankfully, a group member owned a copy of the development environment. In addition, all research materials were either borrowed from the library or found on the internet. Therefore, the goal of keeping the project free was achieved.

Competition

Competitors in our market are those who develop professional audio editing software. Since there are other features in these programs other than filtering, their costs are quite high. Below are the costs of the top three audio editing software:

Cool Edit $150

Wavelab $300

Sound Forge $500

These products are a huge overkill for educators wanting to explain the concept of filtering to their students. This projects wins over these products because it provides a single function to the educators at no cost at all.

Conclusion

The frequency domain is complicated concept for engineering students to grasp. Educators have a need to explain this concept to students in an informative and easy way. The audio range is the frequency range most commonly known to all students. Because they hear as a part of everyday life, teaching them the frequency domain in a way that will let them hear the results is the most effective way of teaching them. This project sought to solve this problem by providing a free and easy software for educators to use for explaining the concept of filtering and the frequency domain to their students and allowing them to simulate passing a signal through channels such as the telephone system. The project was a success and was completed on time with no budget.

Products

Home