From: "Mike Dennison" <mike.dennison@rsgb.org.uk>
Organization: Radio Society of Great Britain
To: rsgb_lf_group@blacksheep.org
Date: Sat, 2 Oct 1999 11:00:00 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Content-Transfer-Encoding: 8bit
Subject: Re: LF: re Dave's Comment on filters
In-reply-to: <004801bf0c5b$78aca1c0$23088cd4@default>
Message-ID: <E11XMDE-0001pn-00@mserv1c.u-net.net>
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

G3NYK wrote:
> I must admit to being a little confused now. I am told a 500Hz
> filter is not good enough for 136, but the theoretical advantage of a
> 300Hz filter is only a couple of dB better, I believe. A 100Hz filter
> would have a theoretical advantage of 7dB (I believe) this should be the
> same as going from 2.5kHz to 500Hz (where I find the practical advantage
> more like 3-4dB). Yes there is a lot of QRM on HF cw and you need to
> isolate a signal to listen to. My experience (with 'inexperienced cw ears'
> !) is that I don't get the theoretical advantage by going to narrow
> filters. Am I missing something? or am I maybe not sufficiently practiced
> to take full advantage of the difference? I guess as Toni says that 1 to
> 2dB can make the difference between working a station and not working it.
> 
> I can hear, and copy, most of the signals on 136 with a 2.5KHz SSB filter,
> and I don't get trouble from DCF39. I use lower sideband and put the
> carrier at 138.05 or 138.10. With a cheaper rx like the Lowe HF150 the SSB
> filter does not have steep enough skirts (and there is no cw option) and
> the carrier of DCF39 is only 60dB down, and I get a light trace of the
> signal on the waterfall display. Now listening like this does mean you
> have to tune the 'grey matter filter' to morse at frequencies up to about
> 2kHz. My tally of calls was about 20 or so listening like that, and it
> gave me the incentive to get better gear. On a receiver with a decent SSB
> filter shape factor the signal from DCF39 is 90dB down and no trouble
> (provided the front end has enough dynamic range)
> 
> I'm keen to get a few tens of dBs advantage on weak signals over the band
> noise but I'm afraid I dont see it as a reality yet. The problem seems to
> be that those last few dBs are beginning to get expensive now. 

Ah, the key words here are "grey matter". The DSP in the brain is 
far superior to anything you can get from MFJ, but only for certain 
things. Certainly, the theoretical advantage in signal/noise of a 
narrower filter is never achieved in practice with reception by ear 
because the brain is able to do its own filtering. The main advantage 
of a narrow filter is in reducing QRM which the brain has difficulty 
filtering out. And perhaps Alan would find the need for a narrower 
filter if he had to copy all of a QSO.

I use a 500Hz filter - the shape factor of the SSB filter in the IC-706 
is too poor for 136 - but have not felt the need to spend the money 
on a 250Hz filter. My score of 70 stations heard on 136 
demonstrates how good the Rx is. 

I do agree, though, that an absolute beginner may well have 
problems with a wider filter because he/she may not be experienced 
in digging signals out of the noise (the brain's DSP needs some 
programming) and often there will be spurious mixing products 
audible in-band until the beginner realises just how good the front 
end needs to be. I am sure there must have been many people who 
gave up when they could not get anywhere near the noise floor.

Of course, narrower filters (including DSP) can be useful and will 
give the brain a rest, so I am not suggesting that they should not be 
used. I have a DSP filter on my bench but do not often use it as I 
rarely find it improves on what the ear/brain combination can do.

One last point, someone (Vaino, perhaps) commented that some 
contest operators do not use narrow filters. Yes, this is common 
because in a contest most stations are both strong and off 
frequency, and overs are too short to do much fiddling with the Rx. 
So using too narrow a filter will result in lost contacts. The beauty of 
a DSP filter is that, once tuned in, a weak signal or one very close 
to QRM can be quickly selected by progressively reducing the 
bandwidth without retuning - something that is difficult to do with 
switched analogue filters.


Mike, G3XDV (IO91VT)
http://www.dennison.demon.co.uk/activity.htm