[RTTY] Narrow Filters for RTTY

[Kok Chen]

The use of narrow receiving filters has cropped up consistently on the reflector.  

Andy K0SM and I have been looking at narrow transmit filters, but there is something from that study which can be "translated" to narrow receive filters.  After all, transmit filters and receive filters are cascaded (with the ionosphere in between).  A filter that is too narrow for transmit is also too narrow for receive.

To answer the question of how narrow can you make a transmit filter, I have written this article:

http://www.w7ay.net/site/Technical/RTTY%20Transmit%20Filters/index.html

Pretty much only modem developers are interested in transmit filters, together with perhaps rig manufacturers who are interested in applying DSP transmit filters to their FSK transmitters.  So, if you are only interested in receive filters for RTTY, just scroll down the web page until you hit Figure 2.2.

There, you will see a plot of the absolute practical minimum bandwidth for any filter that you want to place in the path of a 45.45 baud RTTY signal.  The absolute minimum bandwidth for an "ideal" data filter is of course the 261 Hz Raised Cosine that I often cite.

Notice from the figure that for no loss in Character Error Rate, the perfect practical filter will need to be at least 270 Hz to 280 Hz wide.  By perfect, I mean that the filter has to (1) be flat to a fraction of a dB, and (2) is phase linear (i.e., zero group delay), within that 270 Hz to 280 Hz.

If a filter allows more Mark energy to go through than Space energy, you will suffer the equivalent of sustained selective fading.  And group delay will show itself as something similar to a sustain multipath (you *really* don't want that!).

So, if you are going to place a crystal filter in between the antenna and the modem, you need to make sure that whatever filter you use has 280 Hz worth of a flat passband that has no group delay.  If not, you will incur extra errors.

How much error?  Notice from the plot that when you narrow the "perfect" DSP filter by just 60 Hz, you will double your error rate.  And it climbs much steeper as you narrow further.

This does not include the need to widen the filter when propagation conditions deteriorate and you have more Doppler scattering.  Count on perhaps even doubling the filter bandwidth when you hear audible flutter.

73
Chen, W7AY