AD Digital Filter

[Mike Walne]

Sorry newguy. I disagree.

My method computes R.M.S. values for ANY waveform, bipolar, uni-polar or any anything in-between.
It is a bog standard algorithm which worked over half a century ago when I studied statistics. I don't see why it should not work now.

The real beauty; it is a single pass process, any offset falls out at the end.

Mike

[newguy]

I put together a spreadsheet with a sinusoidal waveform and added random noise. Computed RMS the traditional way and your way and the values agreed closely.

I then added a DC offset to the sinusoid and the two methods suddenly no longer agreed. The greater the offset, the higher the difference.

For example:
- create a spreadsheet with 500 samples, 0-499
- omega column = 2*pi()*sample/500
- sample noisy waveform = 18.3 + 5.15*sin(omega)+(3.52*(rand() - .5)))

Using the traditional method of calculating RMS value, I get 18.702. The magic number method gives me 3.821. If I make the offset term 0, the traditional RMS value is 3.832 and the magic number method gives 3.850.

My original assertion stands: this method doesn't take DC offsets into account properly.

[Mike Walne]

First thoughts.

You're testing with a sine-wave of amplitude 5.15.
Whichever method you use, the RMS value of the pure waveform should be 5.15/(2^0.5)
That's ~3.64.
My method gives the RMS value of the AC component of the waveform and ignores the DC offset, which is what the O/P appeared to want.

With noise you will get higher values.

I'll do a repeat of your noisy spread sheet analysis and come back.

Mike

[Mike Walne]

OK. I've done a spreadsheet analysis on noisy sine-waves with Excel.

I performed the calculations on 1000 values for one cycle of the sine-wave.
I varied the sine-wave & noise amplitudes, and the offset

Each time F9 is pressed, Excel generates a new set of random Nos and new RMS values.
I calculated by both the traditional and the faster method on the each set of values.
Both methods produced exactly the same RMS value on each iteration (within Excel's limitations).

Mike