Within the base ladder filter topology there are subgoups of type.
They can designed as minimum loss, butterworth, linear phase or
Chebychev not unlike LC filters. Each has it's differences and
sailent characteristics. When designing any of these filter the
meaurement of the crystals used is important for the result to be
close to the design prediction. Measurement includes frequency,
motional L, motional C, holder C and Q. All interact. Using 4
crystals of the same type and frequency does not assure the builder
the expected result though usually it will be close. Once the
crystals are known and a design selected the termination impedence
will also be a factor. That is one design criteria, the filters
sensitivity to termination.
When I select crystals from a group of the same lot I often find
besides frequency variation some units that have significantly
different Q [usually inferior] to the rest of the lot. The
variations can be as great as 3:1. An inferior crystal such as
that will be acceptable in an oscillator but would degrade a filter
if used in one.
While this is not meant to be exhaustive by any means. It helps
explain why theory and result will be dissimilar.
Allison
KB1GMX
toggle quoted message
Show quoted text
--- In BITX20@..., "Ruud Jongeling" <pe2bs@t...> wrote:
Hi Paul and Max,
Thanks for your info about the design of a ladderfilter. The
ladderfilter was subject in this group before. Building the BITX20 I
made a shape of the ladderfilter with the program SpectrumAnalyzer
(see also msg 460: "Shape of Ladderfilter easy to measure" by Chris
v.d. Berg). I noticed some differences between the theory and the
actual shape of the filter, see the pictures in the Photo-box and the
messages about them.
From Chris I received an article from G3JIR in QST nov. 80 about
Ladder Filter Design. A copy is dropped in the Files-box. The filters
discribed looks like the filters in the Russian program Max put in
the File-box. I am working on a 9MHz filter following the steps G3JIR
discribed. With the SpectrumAnalyzer it is not difficult to measure
the bandwith of the filters and Excel can calculate the impedance of
the filters and the C's in the filters.
Doing the measurements I noticed that matching the impedance is very
important for the filtershape. Changing the C's makes great impedance
changes. (33 pF, R= 328 Ohm, C=76 pF, R= 142 Ohm on the 2nd order
test filter) That's why I asked some information about matching the
impedance in the amplifierstages in the BITX20 (msg 868). I am still
waiting for a reply....
When I completed the design of the filter I will drop the info in the
Files-box and the shape pictures Photo-box. I hope you will drop
pictures of your filtershapes too.
73
Ruud.
--- In BITX20@..., "Max" <m_orwell@y...> wrote:
de 4N1ZM
I added older easier for understanding version of ladder filter
desingn program in file section under 4N1ZM folder.
--- In BITX20@..., "Max" <m_orwell@y...> wrote:
de 4N1ZM
There is two programs and alternative aproach to ladder filter
design.
Since altavista translation can overcame language bariers it is
possible to check this link through systran translation engine.
Program give obvious relation between impedance capacitance and
crystal unit mesurmen tools:
--- In BITX20@..., "ajparent1" <kb1gmx@a...> wrote:
That is a good starting point but I'd add much has been
published
since then on building crystal filters.
Using the series ladder design and the right constants you can
build
USB and LSB filters as well as symetrical. A good source is
the
ARRL
publication Experimental Methods in RF Design which devotes a
fair
amount of text to crystal filters and has a good listing of
citings
for further reference.
I've use their techniques to build excellent filters using 6
and
8
crystals with symetrical shapes to greater than 70db and low
ripple.
The gausian to 6db shape happens to sound the best to me for
2.2khz
SSB filters. I'm using a 6 crystal version of same in the
first
version of BITx I've built as an transverter IF.
Allison
KB1GMX
