On Jan 10, 2013, at 7:22 AM, Joe Subich, W4TV wrote:
> 400 Hz is nearly the minimum bandwidth necessary for proper decoding
> (minimum intersymbol interference). Narrower filters work by cutting
> adjacent signals more than the desired signal but they impose a much
> higher group delay than a filter that is "flat" across the necessary
> 350 to 370 Hz bandwidth.
That is perfectly correct.
Lets assume the receiver uses a Raised Cosine filter which has been designed to
be the narrowest possible demodulation data filter with no intersymbol
interference (ISI). This means that the receiver's filter needs to be
absolutely flat for 45.45 Hz on either side of the shift (45.45 is twice of the
keying sideband fundamentals, which is 45.45/2 Hz). I.e., the receiving filter
has to be perfectly flat and has no phase errors for a range of 170+91 Hz, or
261 Hz.
Accounting for group delays and ripples from crystal filters, plus any tuning
inaccuracy, and Joe's 350 Hz is completely reasonable. I bought 400 Hz (INRAD)
filters for both my FT-1000MP and my K3 specifically for narrow band RTTY use,
and that is the narrowest I am willing to go. I complement my receiving system
with high dynamic range sound cards to handle adjacent channel interference.
This is not an abstraction either, one of the reasons 2Tone copies better than
MMTTY is because it uses Raised Cosine filters (a little too tight when there
is strong selective fading, but that will be rectified in the near future, if
it has not already been). When you use a receiving filter that is too narrow
ahead of it you are negating a lot of the advantage from good demodulators. The
intersymbol interference will probably degrade 2Tone's performance all the way
down to the performance of MMTTY's demodulator when both have a dual peak
filter ahead of them in the receiver.
The corollary is this: if you ever find that a modem performs better when you
engage a receiver's dual peak filter, it is because the modem's filter is not
optimal for 45.45 baud RTTY. This is nothing new, and I have mentioned it
before on this reflector. When you place a narrow filter in front of K6STI's
RITTY or ahead of cocoaModem, you will invalidate the Matched Filters that are
in those modems.
When there is selective fading, the data filter has to be broadened even more
since the envelope of the RTTY signal is now also scattered by the ionosphere.
If you have put an RTTY signal on an oscilloscope, when propagation conditions
become less than perfect, the signal envelope takes the form that is not unlike
a random AM signal instead of seeing constant power on the scope. This is even
more obvious when look at the individual Mark and Space signals. The 261 Hz
number I have stated earlier is the absolute minimum for a perfect 45.45 baud
signal (what we call the AWGN -- Additive White Gaussian Noise -- condition).
Any narrower, and the signal interferes with itself.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|