[RTTY] Dual Peak Filter question
RLVZ at aol.com
RLVZ at aol.com
Sat Apr 9 07:44:53 PDT 2011
I appreciate the discussions on this reflector as I still have a lot to
learn about RTTY.
Based upon the information below that basically contends that "software
based dual peak filters outperform the a rigs dual peak DSP filter" can I
assume then that I'd do better during RTTY contests by turning my K-3's dual
peak filter off" and turning on the MMTTY dual peak filter? (I have never
tried MMTTY's dual peak filter) In case it matters, I have 250hz and 400 hz
roofing filters in the K-3.
Thanks & 73,
Dick- K9OM
As such, optimal filtering for digital modes is better achieved in the
software modem than by using approximations in the DSP stages of the rig. And
software UI can be simplified to anything that you wish.
With the processors in today's computers, there are not that may things
DSP processors can do in real time that the computers cannot also in real
time, where latency is not a problem (the exception is the decimation process
of the direct SDR that starts at 70 MHz).
A dual peak filter (usually done in the DSP stages of a rig) is just an
approximation of two narrow bandpass filters around each of the two RTTY
tones. This is what produces the "depression" between the two tones. It
approximates what a pair of matched filters does in software that supports
matched filtering. In fact, one thing to note is that if the software already
implements matched filtering, engaging a dual peak filter will make copy
worse, not better. If copy gets better, it means that the software filter is
not optimal.
A simple dual peak filter cannot track the baud rate of the RTTY signal
(at least I have not seen one that does). I think RITTY's matched filter is
also fixed at 45.45 baud, but there are other software that implements
matched filtering which tracks the baud rate when you change it. In the case
of RTTY, a matched filter has an impulse response that corresponds to the
rectangular pulse of the demodulated RTTY signal, i.e., a pair of sin(x)/x
shapes in the frequency domain.
73
Chen, W7AY
More information about the RTTY
mailing list