[1000mp] Filters

[Kok Chen]

On Sep 5, 2009, at 9/5    7:32 AM, Bill Axelrod wrote:

> I almost
> always [use] the 250 Hz filters which improves the S/N ratio and  
> cuts out most of
> the QRM.

The use of real 250 Hz filters when there is no QRM will actually  
degrade the SNR of a 45.45 baud/170 Hz shift RTTY signal, not improve  
it.

This can be seen by the fact that the Matched Filter for a mark (or  
space) pulse has a sin(x)/x profile in the long term RF spectrum --   
the spectrum looks like that from the sin(x)/x of a 22.73 Hz square  
wave.

You can find the explanation of how SNR is optimized here (scroll down  
to the section on "Example of matched filter in digital  
communications" as it pertains to digital signals),

http://en.wikipedia.org/wiki/Matched_filter

Notice that the result from a matched filter is not a sinusoid, but a  
triangle wave.  You can also see that same triangular waveform of a  
matched filtered RTTY signal in Figure 1 of this page

http://homepage.mac.com/chen/w7ay/cocoaModem/More/Contents/part2.html

A triangle is merely the inverse Fourier transform of the square of  
sin(x)/x.  I.e., when you convolve two square waves.

The reason I said "real" 250 Hz filters in the first paragraph is that  
most crystal filters that you buy are much broader than you think they  
are.  But narrower is not always better in the realm of digital  
communications.

73
Chen, W7AY