On Sat, 2007-06-16 at 11:54 -0700, Kevin Purcell wrote:
> This is very common amongst VHF/UHF weak signal DXers (including
> EMEers) too.
>
> It's also little know but wider bandwidths for SSB improve
> intelligibility too especially for the very weak signal case (see the
> graphs on articulation index versus bandwidth in Sabin's book on SSB).
>
> The reason to use narrower filters is to reduce QRM that your brain
> (or the RX) can't deal with.
>
> Why do wider filters work:
>
> 1. It avoids the problem of impulse noise being smeared in time by
> the limited bandwidth of the filter (and by group delay in the
> filter) so in noisy environments a wider filter can help a lot.
The time response of various filter designs is very important. The
classical Tchebychev filter approximations typical of crystal lattice
filters and vintage Collins mechanical filters have a rotten time
response. The amplitude response of those filters has sharp shoulders
and a flat passband. The filters with the best time response
(Butterworth or even better Bessel) have the least flat frequency
response, rounded shoulders, and gentle skirts. Its hard to make ladder
filters give a Tchebychev response while Butterworth and Bessel are
easy. Tentec filters are ladder filters. I don't know about Inrad
filters. I do know that an aftermarket Collins filter I bought for CW in
my FT-857 doesn't ring significantly. And on VHF it sometimes enhances
S/N slightly on very weak signals and always does that better than the
audio DSP.
It is possible to have sharp corners, flat frequency response, steep
skirts, and a good time response but that is ungodly expensive in
hardware because the design has to use many extra resonators to correct
the phase response. In a digital filter it is handier to have the extra
poles to achieve both goals, at the cost of total time delay.
The worst CW filter I've ever used was a 270 Hz from Kenwood in a
TS-430. Pure noise going in, had a distinct pitch coming out. It wore my
ears out in a couple hours of FD fighting that tone to copy desired
signals.
>
> 2. Your brain is an interesting signal processor. It can act as a
> very narrow (50Hz or so) adaptive filter but without the usual
> problems of group delay and the like of a real filter.
Probably because it does a time correlation and the wider the noise
bandwidth, the poorer the correlation of the noise so cancellation is
easier. The narrower the noise bandwidth, the more coherent the unwanted
noise is the harder it is for brain or DSP to eliminate.
>
> It was mentioned in EMRFD and you can find it on VHF DXer websites
> too. SM5BSZ comes to mind.
>
> On Jun 16, 2007, at 6:19 AM, Jeff Frank wrote:
>
> > I almost always have easier and clearer copy on weak cw signals
> > when I use the 1.8 kHz filter vs. the .5 or .25 kHz filters in my
> > 6.3 MHz IF. I've got nothing in the N1 position. Is this the way
> > it's supposed to be?
>
> --
> 73 DE N7WIM / G8UDP
> Kevin Purcell
> kevinpurcell@pobox.com
>
--
73, Jerry, K0CQ,
All content copyright Dr. Gerald N. Johnson, electrical engineer
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|