22 replies to this topic

[mbruner63]

This great!  I will certainly use this for my AGENT performance algorithm.  For the circadian rhythm component, I use a double sine wave.

[mbruner63]

[Paul Newton]

Hello all,

 

When Hanzibal first posted his sine & Cosine table code, I tried it out and wished for a few updates.

 

The first thing I did was to write a test harness where I spotted the same issue reported by Dan T.

The test harness showed that around zero (or 90 degrees for cosine) and each time the table wraps the waveform was not symetrical:

Sin( -2) =  -1Sin( -1) =   0Sin(  0) =   0Sin(  1) =   1Cos(254) =   1Cos(255) =   0Cos(256) =   0Cos(257) =  -1

It turns out that I solved the problem in the same way Dan T did - by adding an extra value to give an odd number of samples in the byte array. This allows it to be symetrical around each quarter cycle.

 

I also noticed that the peak value was not being reached at the correct moment. I solved that by a small modification to the generation of the table data.

 

Next I added the features I wanted for myself - adjustable table length, and adjustable peak value.

These are specified by setting the two private parameters at the start of the class. _peakValue and _tableLength. Everything else is calculated from these, so no other code should need to be edited.

I also added two new functions that return the peak, and the total virtual table size if the program needs them.

 

The code is commented, so it should be easy to use.

 

My updated class and test harness program.cs file are attached in the zip.

The test harness prints the cross over points of the waves, and the entire table to the debug output, the printing is formatted so that the output can be saved as a .csv file and opened in a spreadsheet for plotting.

 

Have fun - Paul