Message-ID: <37F7C071.24C8@xtra.co.nz>
Date: Mon, 04 Oct 1999 09:45:37 +1300
From: "vernall" <vernall@xtra.co.nz>
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
Subject: Re: LF: Re: Receivers
References: <004101bf0cdf$11aaa6a0$0600a8c0@main> <001b01bf0d7f$0719c400$22c622c3@34.thefree.net>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Hi all,

Thanks to contributors of comments on receiver performance at LF.  Some
more points tossed in to the melting pot ...

The bandwidth and shape factor of a receiver IF filter is one of the
"hearty annuals" of amateur radio discussion.  What most users focus on
is the STEADY STATE frequency response, as one could verify by tuning
across a clean carrier as a test signal (and not forgetting about phase
noise and reciprocal mixing, but not to digress ...).  The parameter
that is not adequately mentioned in specifications is the TRANSIENT
RESPONSE of a filter but at any LF band (and MF and lower HF as well)
that is of great importance as the "static" is a type of unavoidable
impulse applied to the receiver.  For trying to copy weak signals, what
happens to the static peaks can end up as "ringing" in the IF filter
output, even if it seemed "cleaner" at the antenna terminals or filter
input.

Passive filter circuits can not avoid some amount of ringing.  It
necessarily gets worse as the passband is narrowed.  This can be thought
of as impulse energy being "converted" to ringing energy within the
passband.  Once the ringing energy is inband, it "pollutes" the wanted
signal.  The filter phase response (also known in its derivative form as
group delay) depends on the specific type of design (Butterworth,
Chebychev, Elliptic, Bessel, etc or a hybrid combination).  The filter
with least ringing is the Bessel design, but it also has the least skirt
selectivity.

So in summary, steady state filter response is only part of the story,
and transient response is one of the parameters that has a big influence
on the results of receiving with narrowband settings (not only to aural
CW, but to other modes such as RTTY or PSK-31).  Going for (passive
component) filters with test results of "square skirt responses" and
"narrower passbands" is a recipe for a lot of ringing, guaranteed.  I
would imagine that commercial designs of the narrow band crystal filters
(for receivers) do trade off skirt roll-off as bandwidths are narrowed. 
This is probably a topic where results are commercially or militarily
sensitive and could be why there is a lack of public domain
information.  Most of the amateur designs of narrow filters are entirely
focussed on selectivity results and appear to under-estimate the matter
of transient response.

The characteristics of an IF noise blanker can also be significant to
"improving" reception of weak LF signals, as the blanker tries to negate
the burst of QRN by temporarily switching off the IF gain.

I have had an audio DSP accessory with variable bandwidth for some time,
and in theory it can give incredible selectivity along with "no
ringing".  While it tests out rather well for audio signal to noise
situations (more signal than noise), I find it is not of practical
advantage when tried in noise to signal situations.  However, it is
noise to signal situations that is the attraction to amateur DX work! 
So in summary, I have found that audio DSP can make good signals better,
but it also makes bad signals worse (as judged by aural listening). 
Consequently I do not bother using audio DSP for weak signal listening.

I can not comment on receivers with DSP in the final IF, as I have not
tried one.  However, if they can do digital blanking of noise bursts as
well as filtering with minimal ringing then they could have something to
offer that passive filters do not?

When receivers are tested on the bench, with fairly pure signal
generator sources, and under steady state conditions, they can verify
the specifications that are in the advertised data for selectivity and
signal to noise performance.  Putting that receiver on LF (using a
converter if the tuning range does not go down that far) where there is
always "pops" of QRN will soon sort out how ringing impacts on narrow
bandwidth options and when receiving weak signals.  I am not aware of a
standard test method to produce measured results of how a receiver
performs in noise to signal ratio situations (especially with impulse
QRN on LF), but that particular aspect would be of great interest to
this discussion group.

Bob ZL2CA