Problem Description

Updates:  (It's not so simple anymore!)

August, 2002 - Version 1: 

Here's an oscilloscope program that uses your sound card for input.  My hope is that it encourages a few  future scientists (and/or Delphi programmers) to experiment -- and have some fun in the process.   

Background & Techniques

Windows has defined data structures and functions which let programs  access the output of the analog-to-digital converter (ADC) on sound cards, or other cards that provide appropriate driver software.    This program uses this "Wavein"  Application Program Interface (API)  to display recorder input on your monitor.    There's not much practical use for the program except as a test bed for further development.  

 Kids however do love to "see" their voices and the waveforms of other weird sounds they can think of.   Anything that motivates them to learn more about the mysteries of the real world must be a good thing. 

Selecting input: To get some  input for your oscilloscope, you'll need to adjust the  "Sound Recording" audio properties of your sound card.  Specifics vary from version to version of Windows, but in general you can right click on the speaker icon on your task bar and select Volume Controls.   From the Volume Controls panel, click Options then Properties to bring up the  dialog that will allow you to  select devices and volume levels for  program inputs.  You can also get there from your Control Panel by selecting  Sounds & Multimedia then Audio then Recording Volume.    The microphone input is best for playing around, but you can also use "Wave devices" as input to view sounds that are being played, for example, via the Windows Media Player.   In fact, Media Player's visualization called "Scope" was my model for the display.

There's a "Save Image" button that allows you to save up to 10  screen snapshots with names OScope0.bmp through OScope9.bmp. 

As always, if you have any problems or questions, let me know.

Non-programmers are welcome to read on, but may want to skip to the bottom of this page to download an executable version of the program.

Notes for Programmers

The prototype input class for my TWaveIn class was a TWaveRecorder class written by John Mertus of Brown University.  A Google search on "John Mertus" reveals that John may be more famous for earthworms and some letters he wrote to Dave Barry than TWaveRecorder, but search on TWaveRecorder and you'll find the  article pertinent to this project.   

 I cleaned up a few things, eliminated some stuff that I didn't understand, and made a fairly major change in the way that  full buffers are handled.    Using the Windows Wavein API requires a multi-step protocol.

To start input -

When a buffer is full -

And, at close time -

The Callback problem: The WaveInOpen API function lets callers specify what actions to take when a buffer is full, or otherwise need processing.  The CALLBACK_FUNCTION option caused problems (program hangs) for me when the buffers were reset by calling WaveInReset.  A Google groups search turned up enough other users with the same problem  to convince me to switch to the CALLBACK_WINDOW technique.  (The callback function  problems by the way, were variously attributed to Windows versions or driver versions.)    Now, instead of calling the specified procedure when a buffer is full, the API issues a an MM_WIM_DATA message.  The procedure to be called must be defined as a message handler  for this message.   A callback procedure would be cleaner from the TWavein user's viewpoint, but we can't always get exactly what we want.  

On the main form, most of the code was for handling my experiments in processing the data.   I added  Gain and Sweep controls.  Gain multiplies  data by   2^(gain-3) , i.e. multiplies or divides data points by 2, 4, or 8, just as a quick way to provide some vertical scaling.  (Multiplication or division is handled by shift left (shl)  or shift right (shr) operations which shifts data bit by bit.   Each bit position shifted multiplies  or divides  the value by 2.  Sweep gives some horizontal scaling by incrementing the X-axis by Sweep pixels for each point plotted.  

Addendum November 16, 2003:    I had occasion to actually use the Oscilloscope program recently which led to Version 2, posted today.  There are three enhancements: 

Addendum December 5, 2004:  

Here is the next cool enhancement to our evolving Oscilloscope - Version 3  now also functions as a Spectrum Analyzer!.  Click the "Spectrum" button after a frame has been captured and see the amplitudes of the frequencies contained in the sample.  Fast Fourier Transform (FFT) code has been added to analyze the signal.  Results are displayed in a spectrum bar graph.   

Captured data in both the time and frequencies domains can be saved to a file for further analysis.  Also added user control of standard sampling rates  (11,025 or  22,050  or 44,100 samples per second).   

A buddy of mine is creating the next Stradivarius violin and needed the resonant frequencies of the instrument when he administers the the "thump test".    So here you go, Jim.  I also added time scale lines to the oscilloscope display and fixed a few other problems along the way.  If there are any bugs left, let me know.       

I'll publish a more extensive sample and write-up of the FFT process one of these days.   

Addendum February 28, 2006:    Version 4 posted today adds a dual trace function using stereo input signals.  The visual layout has been improved  to appear more "osclloscope-like".   Input signals for recordable devices  are  selectable via buttons and settings are saved from run to run.   Many thanks to "Krille", a very sharp Delphi programmer from Sweden for doing most of the work on these changes. .

 Addendum February 26, 2007:  It has been a year since the last update.  Version 4.2.2 was posted today correcting an error the spectrum analyzer which showed  extra overtones that weren't really there.  I saw it as a 3rd overtone, Angus G. who reported the error thinks it was the 7th.  In any event, I located the bug and fixed it  today.

March 29, 2014:  

Version 4.2.3 was implemented to correct the oscilloscope not running after Dual Trace" mode was set without a stereo formatted input source.   Some formatting  enhancements were also made.   There are lots of controls on the scope and it is still a work in progress, so feedback and bug reports are welcome. 
 

April 28, 2014:  Some changes were made in Version 4.2.3 today to improve diagnostics when the input device could not be started for display.   Problem seems to occur when sampling rate is modified in dual display mode.  The suggested solution is to set the sampling rate first,  then then turn on dual display mode.  If dual display is chosen for a monaural  sound format, display will contain two nearly duplicate displays since the program splits each pair of sample points between channel 1 and channel 2.       

Running/Exploring the Program 

Suggestions for Further Explorations

Lots---