Posted by
Geoffrey Mackenzie-Kennedy-2 on
Nov 14, 2004; 8:16am
URL: http://elecraft.85.s1.nabble.com/Talking-filter-Bandwidths-tp371577p371581.html
Agree that the 60db/6db shape factor is an essential indicator of a filter's
"selectivity performance". But the filter's rsponse above the 6db points is
also very important if you want a good sounding SSB signal, or a Rx's SSB
filter that can really dig in a pile-up. If the filter's skirts come up from
the stopband, through the 60db and 6db points and continue to 0db at the
same rate of attenuation vs frequency, the SSB signal will usually sound
"raspy". To avoid this, the "top" of a good SSB filter's response between
0db and 6db (or thereabouts) is designed to be a little rounded, approaching
Gaussian, while still meeting the 3db bandwidth required by the signal which
is to be passed through. A signal through a 14 crystal SSB filter without
the rounded top does not sound good, even with a 3db bandwidth of 2.7kHz.
The same applies to Rx CW filters. Very narrow bandwith filters with rounded
tops do not "ring" as badly as those with sharp edged tops.
Totally agree with you about sweeping filters to obtain their responses. It
can be intersting when the stopband is down 90 - 100db, because at no point
must the filter be overdriven. These are hombrew filters. I wish that filter
manufacturers would specify their filters signal level limits, because too
often receiver designers forget that crystal filters especially can be
overdriven and introduce distortion. But perhaps if the manufacturers did,
some might not sell any filters!
73 Geoff GM4ESD
----- Original Message -----
From: "Guy Olinger, K2AV" <
[hidden email]>
To: "Deni" <
[hidden email]>; <
[hidden email]>
Sent: Sunday, November 14, 2004 4:53 AM
Subject: Re: [Elecraft] Talking filter Bandwidths
> For a lot of us (I have seen this discussed before) the 3 db points
> are fairly useless as a real predictor of performance. The 6 and 60 db
> points do a much better job, telling where the skirts start, and where
> they are diving.
>
> I will never forget my disappointment in the Collins 500 kHz CW
> mechanical filter (3 db) for my MP. I just assumed.....
>
> Best thing is to get a sweep of the passband. Funny how those are so
> hard to get for filters unless you're getting them from INRAD's
> webpage. They have the Collins curves as well. You can compare curves
> and see yourself. Interesting browsing.
>
> You can see from the curves whether the skirts are still going strong
> and diving past 60 db or whether they're dying and on the way to
> having a poor ultimate rejection.
>
> 73, Guy
>
_______________________________________________
Elecraft mailing list
Post to:
[hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
http://mailman.qth.net/mailman/listinfo/elecraft
Help:
http://mailman.qth.net/subscribers.htmElecraft web page:
http://www.elecraft.com