Chen,
You are right about the ideal DSP being able to handle the selectivity
processing. But for our project, we are not assuming that is the case at
all. The filter specifications are under our control - the DSP is not - and
we would like to use the filter as best we can to set the selectivity
bandwidth.
Your comment, in summary, is that the optimum bandwidth should be 300+ hz.
Let me ask you and the group reading along about the effect of the roll off
at the edge of the passband.
The filter plots in this range that are showing a "flat top" with a 1.6
shape factor get this result at the expense of ringing. In our application,
where we need to have this filter serve dual duty with the CW contester, we
are prioritizing the ringing criteria over the bandwidth. So the group
delay characteristics are much more even across the passband. The
compromise is that the filter will have a bit more rounded top than these
more "square top" plots.
So for example, a true 300 hz @ -6db with 1.6 shape factor may look like
this at the very top:
-1db 150 hz
-3db 240 hz
-6db 300 hz
That means that a perfectly centered RTTY signal will be roughly -2 db from
the center peak points (170 hz). The sidebands on the left and right will
see a bit of attenuation - depending on the point you measure the sideband
point.
>From a weak signal decoding standpoint, how will this roll off effect the
ability to copy these signals?
A common data point seems to be the Inrad/Elecraft K3 filter @ 250 hz.
Looking at the top of that filter, it is about -3db down at 240 hz and -6db
at about 370. Guys seem to have no negative comments about that filter.
The 300 hz @ -6db filter above would have a better shape factor (perhaps 1.6
vs. 2.1) and is 70 hz more narrow at the -6db point.
As we have discussed the importance of the sideband today, I can see the
benefit of the improved shape factor. But I am beginning to wonder if a
slightly wider -6db point would be more optimum for RTTY.
So perhaps 350 hz @ -6db dropping down to about 550 @ -60db would be a more
optimum filter for RTTY contesting purposes???
73/jeff/ac0c
--------------------------------------------------
From: "Kok Chen" <chen@mac.com>
Sent: Tuesday, August 25, 2009 5:57 PM
To: "RTTY Reflector" <rtty@contesting.com>
Subject: Re: [RTTY] Crystal filter width preferences for RTTY contesting
>
> On Aug 25, 2009, at 8/25 2:27 PM, John Barber GW4SKA wrote:
>
>> Simple maths for this used to be shift plus twice the baud rate; 170
>> + (45 x
>> 2) = 260Hz
>
>
> But why did you use 45 * 2?
>
> The fundamental RTTY keying waveform is 45.45/2 Hz, not 45 Hz.
> (Imagine a signal that switches between mark and space at 22
> millisecond intervals; the period of the square wave waveform is 44
> ms, not 22 ms).
>
> Secondly, even if you had intended the equation to represent 2*22.5 Hz
> (twice the fundamental keying signal) on each side of the two keyed
> carriers (thus a total of 2*45), the third harmonic (and arguably even
> up to the fifth harmonic) of the keying signal still has sufficient
> energy to be very useful for an "ideal" RTTY matched filter (i.e.,
> something that convolves the input with a 22 ms rectangular pulse).
>
> The minimum bandwidth I would use for a very weak RTTY signal under
> AWGN is 170 * ( 22.725 * 3 )*2, or a tad over 300 Hz.
>
> When Doppler spreading exists, you would require an even wider
> passband to capture most of the energy from the pulse. The CCIR
> profile (using the Watterson model) for high latitude flutter has a 2-
> sigma Doppler spread of 10 Hz. This is why copy of a fluttered signal
> usually improves when you open the received passband up, even when the
> demodulator is not matched to the fluttered signal's pulse profile
> (probabilistically, a rectangular pulse that is convolved with a
> Gaussian).
>
> In today's sound card world, all that the I.F. filter provides is a
> "roofing filter" to keep the sound card (and the receiver stages
> between the I.F. filter and the audio output) from saturating. If a
> signal, however wide, reaches the sound card without clipping it, good
> software filters are much better than any crystal filters at rejecting
> QRM.
>
> 73
> Chen, W7AY
>
>
>
>
>
>
>
> _______________________________________________
> RTTY mailing list
> RTTY@contesting.com
> http://lists.contesting.com/mailman/listinfo/rtty
>
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|