Message-ID: <4C5AFD94.4080902@iup.uni-heidelberg.de>
Date: Thu, 05 Aug 2010 20:06:12 +0200
From: =?UTF-8?B?U3RlZmFuIFNjaMOkZmVy?= <schaefer@iup.uni-heidelberg.de>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.11) Gecko/20100711 Thunderbird/3.0.6
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <4C58579F.30406@telus.net> <8CD01F14619A000-1C48-3D75@webmail-d073.sysops.aol.com> <8CD02ACE55ECBAC-1CF4-9FFF@webmail-m086.sysops.aol.com> <4C5ABD7D.2080301@iup.uni-heidelberg.de> <008501cb34c4$51e99cd0$4001a8c0@lark>
In-Reply-To: <008501cb34c4$51e99cd0$4001a8c0@lark>
Subject: Re: LF: Re: Bandpass filter design
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: quoted-printable
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


Thanks for all answers so far.

Alan, i think that filtering is essential for me since DLF (153 kHz,=20
500kW) is just 40km distanced. I am afraid that the amplifier gets=20
overdriven else. For DCF39, which is now almost not filtered (maybe=20
10...20 dB) it is the same. This level is 60 dB above noise and i need=20
some 20 dB more gain after the BF981 since the level is still to low for=
=20
the soundcard. SpecLab can do a lot, clear, but what if the amplifier=20
becomes nonlinear due to those high levels? This is my worry. In UK, DLF=
=20
or DCF39 is not that issue i assume (?) ;-)

What is the problem with rapid phase changes? I am no communication=20
engineer, sri, but i like to learn on that! (remembering that before=20
some months i asked what is the "gain" when going from qrss3 to qrss30=20
;-) ).

Michels idea sounds good. In principal, this is the same what i have=20
done before the mixer. Applying 2 further band filters means parts=20
effort but this doesn't matter really. If i put one filter to 12,7 kHz=20
(QRSS3) and the other to say 11,6 kHz (CW and european transmit window)=20
this would work, regarding the attenuation of DCF39. And i can adjust=20
the filter characteristic directly by watching the SpecLab window and a=20
broad band noise source (my soldering station does a good job there when=
=20
placing the antenna near it :-) ).

I will keep your suggestions in mind. Maybe a call to Markus can=20
convince my completely ;-)

73, Stefan



Am 05.08.2010 17:00, schrieb Alan Melia:
> Hi Stephan....why do you think you need a narrow filter at 12kHz?? Why=
 not
> let the sound card sort it out? Provided you have killed the image (113k=
Hz)
> there should not be a problem. If you use a narrow passive filter you ri=
sk
> rapid phase changes near the wanted frequency. This is probably not a go=
od
> idea. I suspect that a fairly "benign" low pass filter (Butterworth??)=
 just
> above 12Khz (to aid the anti-alias filtering) and another Butterworth to
> remove any 50Hz and low harmonics of that say below 1kHz. this leaves a
> fairly flat pass-band with a slowly changing phase response.
>
> What may be more important may be getting a good low noise amp to feed=
 the
> sound-card. It is worth a look at some of the softrock workand circuits
> here. Also Paul did some work on this some time back. Jim may have some=
 more
> helpful ideas in this area. I have not found conventional filtering in=
 front
> of an FFT does a lot of good and it certainly has the potential to "mudd=
y"
> things up.
>
> Alan G3NYK
>
> ----- Original Message -----
> From: "Stefan Sch=C3=A4fer"<Stefan.Schaefer@iup.uni-heidelberg.de>
> To:<rsgb_lf_group@blacksheep.org>
> Sent: Thursday, August 05, 2010 2:32 PM
> Subject: LF: Bandpass filter design
>
>
>   =20
>> Dear LF,
>>
>> Currently i am setting up my active antenna for the planned LF grabber
>> here in Heidelberg. It is an active E field antenna, using a BF981 and=
 a
>> 125 kHz signal that transforms the 137 kHz down to 12 kHz where some
>> band filtering has to be applied. Then, i need another amp stage to
>> drive the soundcards input (BF862). The high impedance of the wire inpu=
t
>> is first down transformed by a BF862 stage as a source follower, then=
 i
>> allpy a double LF bandpassfilter that is coupled by a C of some pF
>> (about 4...8 pF). This signal is applied to the 2nd Gate of the BF981..=
.
>>
>> My question is: There may be better suited filter designs than taking=
 a
>> L parallel C resonated at 12 kHz (after the mixing stage), between
>> signal and ground since this gives a sharp filter, ie 137,7 kHz is
>> already attenuated by 25 dB compared to 137,0 kHz. What i want to have
>> is a filter with a specific bandwith and edge frequencies with about
>> constant low attenuation in the transmission range and relative sharp
>> slopes so that 137,7 kHz is not really attenuated but 138,83 kHz
>> (DCF-39) as much as possible. DCF39 is 60 dB above noise here although
>> it gets already attenuated by the input band filter!
>>
>> Jim/M0BMU has designed a filter for his VLF loop RX that looks quite
>> good. Is there a web page where i just can type the filter oder, edge
>> frequencies, input- output impedances and so on and get the values?
>> I have found such one at
>> http://www-users.cs.york.ac.uk/~fisher/cgi-bin/lcfilter but i am not ye=
t
>> experienced too much about this stuff so i am not sure if this gives
>> really useful answers.
>>
>> Any other simple ideas to come quickly to the optimal filter type, oder
>> and values? I do not want to spend too much time for that, so an
>> "excellent filter design book" is not the best hint ;-)
>>
>> What about a cauer filter? I have read that it has the sharpest edges
>> but this may cause QRM in the pass band? (like clicks in a too sharp CW
>> filter?)
>>
>> The picture shows what i have done so far. Watching the spectrum on the
>> roof  of the institute (the future QTH) from 0...48 kHz in SpecLab look=
s
>> very promising so far (see picture). DLF is 60 dB above noise although
>> already attenuated about 60 dB! So filtering before mixing and further
>> amplification is necessary in my case, i assume...
>>
>> Tnx for helping ideas.
>>
>> 73, Stefan/DK7FC
>>
>>     =20
>   =20