Message-ID: <01C26537.C1466D20.g4jnt@thersgb.net>
From: "Andy talbot" <g4jnt@thersgb.net>
To: rsgb_lf_group@blacksheep.org
Date: Thu, 26 Sep 2002 08:36:04 +0100
Importance: high
MIME-Version: 1.0
Subject: LF: RE: Narrow CW filter using 32kHz clock crstals?
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Sender: <owner-rsgb_lf_group@blacksheep.org>
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

Its was I ...........

1)  Take a handful of watch crystals, from the same batch  (buy 10 or 20 from 
Farnel / Maplin ??)
2)  Connect  5 in series to form the top elements of a ladder filter; you now 
have 6 connection nodes, taking 1 as the input, 6 the output
3)  Connect shunt capacitors to ground as follows, from node 1  47pF,  nodes 
2,3,4 and 5   82pF,  and node 6  47pF.   This gives a ladder filter with a 
characteristic impedance of around 50kHz, a 3dB bandwidth in the region of 
2.2Hz and 10 dB loss.  Centre frequency is 32765.5Hz - the series resonance of 
a 32768Hz device  cut for  || resonance.    The loss is high enough to smooth 
out significant ripples so I made no serious attempt to get a decent response 
shape - it was rounded - but with a lot of patience it could be improved no 
doubt.
4) All attempts to use the crystals in thier parallel mode, with top coupling 
capacitors lead to very high losses that would have needed a characteristic 
impedance of meg-ohms to get operational.

I didn't think much of the xtal filter idea idea and abandoned it for a DSP 
solution using the 56002EVM.

Andy  G4JNT

-----Original Message-----
From:	Tracey Gardner [SMTP:tracey.gardner@ntlworld.com]
Sent:	2002/09/25 18:49
To:	rsgb_lf_group@blacksheep.org
Subject:	LF: Narrow CW filter using 32kHz clock crstals?

 << File: ATT00001.txt; charset = Windows-1252 >>