On Sat, 2009-10-10 at 14:36 -0400, Gary Hoffman wrote:
> I've built crystal lattice filters (and others) in the past. I also have
> books on designing them,
> Basically you are all correct !
>
> You CAN pull them with capacitors.
>
> It WILL change the input and output impedances, more if you pull them more.
>
> Crystal Frequency is the design parameter you use to change the filter's
> operation.
>
> So, your mileage may vary, but you can pull them with capacitors, and you
> might be sorry you did, depending on how far off the impedance matches end
> up being. Its better to change the crystals.
>
> 73 de Gary, AA2IZ
>
>
In the ladder filter, both the crystal series resonant frequency and the
Q of that resonance is critical. I think a high Q crystal comes from how
close the faces of the (AT cut) crystal are to parallel. At least that's
how it seemed when I was grinding (and etching) my own crystals. Polish
is critical, the highest quality crystals I have are in glass envelopes
and the quartz is clear, it does not have a ground surface. Then I
suspect the weight of the electrodes is critical to Q and that weight is
the final frequency adjust in plated crystals. In the AT cut square or
rectangular crystal there are ratios of thickness to width that produce
high Q and low Q crystals. As you grind only the thickness, the crystal
Q (activity) changes cyclically. Often when you get it to frequency you
have to grind on the edges to get the activity up so the circuit will
oscillate reliably. Murphy's law at work.
The cheapest crystals these days are for 3.579 MHz and they are the
lowest quality rocks around. Unless the 32.768 KHz tuning for crystals
can make that claim. Neither are great quality crystals.
When choosing crystals for a narrow (and 250 Hz is narrow) ladder filter
the crystal frequencies have to match to better than 50 Hz, (5 ppm) and
the Qs have to be high. Some of the design techniques can start with
crystals having a range of frequency and adjust each one with series
capacitors (not part of the simplest ladder filter). For wider filters
there are at least two techniques. The working crystal series resonant
frequencies need to be spread over most of the passband so either
crystals are selected from the distribution (buy 20 crystals, use 6 sort
of thing) to fit the design or some or all the crystals have their
series resonant frequency adjusted with series capacitors (usually in
the 10 to 75 pf range). The troubles are that you may have to accept a
change in center frequencies which is OK for a one of project, but in
production you have to match carrier frequencies and LO frequencies so
that having the filter edges where you wish is important. Though in my
FT-857D, there is a set of memory settings for those things to
accommodate lower tolerance ceramic filters. Computer control can be
programmed to accommodate those variations.
Low cost crystals these days are made with 50 ppm calibration and
temperature drift distributions. You can buy crystals with better
calibration and stability than that, but you pay a price. You can
specify the L and C plus Q of a crystal too, but you can't buy them in
large batches for 46 cents a crystal, probably have to pay a domestic
producer (like ICM) $46 a crystal. Being able to select L and C
(motional parameters) as well as Q and frequency tolerance can give
addition degrees of freedom for the filter design to optimize time as
well as frequency response and terminal impedance. Buying crystals from
an off shore facility you might have to take the range they produce and
do your own sorting for all the parameters. There will be a reject bin
of parts that won't fit any filter (or BFO) criteria.
73, Jerry, K0CQ
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|