Very basic crystal question

E

Ecnerwal

Guest
When I go looking for information on crystal filters, everything I can
find seems to be aimed at using a bunch of crystals and capacitors to
widen the bandwidth, or else it's marketing for people who would be
happy to sell me crystal filters of the above sort. I don't want to
widen the bandwidth...

What I'm presently interested in is an exceedingly narrow-band filter
and/or amplifier. 60Khz for the monent, as I'm playing with WWVB - and
while I can buy an off-the-shelf amplifier or even amplifier and digital
buffer chip for $3-4 (CME6005 and CME8000) which uses an external
crystal to get very narrow bandwidth, there's no real discussion of
what, precisely, that's doing. As a practical person, I may end up
buying one of these units as a major building block, but I want a better
general understanding going into it, as I've really done very little
with radio, and this is as much about learning what's what as the actual
end result.

All the DIY WWVB receivers I've found are using LC filters, which is
classic, but seems to be more prone to noise than a very narrow
bandwidth filter would be, given that in this case the frequency of the
signal is _very_ well defined - in fact, given that I can get "the
correct time" rather easily from any network attached computer these
days, the frequency reference is as much or more of a driving reason as
the time codes are. I don't have any logical reason to spend money
sending my elderly ovenized crystal reference out for calibration, but
I'd still like to be able to maintain it on spec as much as possible -
which is, in theory, possible from the WWVB signal. The units mentioned
above claim to get about a 10HZ bandwidth, though I'd call the datasheet
"sketchy" at best.

Is it as simple as "slap a 60KHz crystal in line with the input (or the
feedback path of an amplifier), and only 60KHZ will get through"
(perhaps with some bypassing to block any multiples: 120KHZ, 180KHz,
etc.)? That seems a bit too simplistic, but other than a reference to
Walter Cady's 1922 paper (which I have not tried to go track down in
person) using single crystals for very narrow bandwidth coupling (in
this article: http://www.ieee-uffc.org/freqcontrol/crystal.html ), I've
somehow managed to miss any discussion of very narrowband crystal
filters at the most basic (or too basic to be discussed) level.

I need to clear the paper and clutter off my bench and get to playing
with this, having finally gotten a sweep generator.

--
Cats, coffee, chocolate...vices to live by
 
On 11/2/07 8:53 AM, in article
LawrenceSMITH-6211C7.11530102112007@news.verizon.net, "Ecnerwal"
<LawrenceSMITH@SOuthernVERmont.NyET> wrote:

When I go looking for information on crystal filters, everything I can
find seems to be aimed at using a bunch of crystals and capacitors to
widen the bandwidth, or else it's marketing for people who would be
happy to sell me crystal filters of the above sort. I don't want to
widen the bandwidth...

What I'm presently interested in is an exceedingly narrow-band filter
and/or amplifier. 60Khz for the monent, as I'm playing with WWVB - and
while I can buy an off-the-shelf amplifier or even amplifier and digital
buffer chip for $3-4 (CME6005 and CME8000) which uses an external
crystal to get very narrow bandwidth, there's no real discussion of
what, precisely, that's doing. As a practical person, I may end up
buying one of these units as a major building block, but I want a better
general understanding going into it, as I've really done very little
with radio, and this is as much about learning what's what as the actual
end result.

All the DIY WWVB receivers I've found are using LC filters, which is
classic, but seems to be more prone to noise than a very narrow
bandwidth filter would be, given that in this case the frequency of the
signal is _very_ well defined - in fact, given that I can get "the
correct time" rather easily from any network attached computer these
days, the frequency reference is as much or more of a driving reason as
the time codes are. I don't have any logical reason to spend money
sending my elderly ovenized crystal reference out for calibration, but
I'd still like to be able to maintain it on spec as much as possible -
which is, in theory, possible from the WWVB signal. The units mentioned
above claim to get about a 10HZ bandwidth, though I'd call the datasheet
"sketchy" at best.

Is it as simple as "slap a 60KHz crystal in line with the input (or the
feedback path of an amplifier), and only 60KHZ will get through"
(perhaps with some bypassing to block any multiples: 120KHZ, 180KHz,
etc.)? That seems a bit too simplistic, but other than a reference to
Walter Cady's 1922 paper (which I have not tried to go track down in
person) using single crystals for very narrow bandwidth coupling (in
this article: http://www.ieee-uffc.org/freqcontrol/crystal.html ), I've
somehow managed to miss any discussion of very narrowband crystal
filters at the most basic (or too basic to be discussed) level.

I need to clear the paper and clutter off my bench and get to playing
with this, having finally gotten a sweep generator.
Click to expand...
I'm not familiar with WWVB, but a word of caution is advised: If you wish
to recover modulation, the filter bandpass must be wide enough to at least
match the highest modulating frequency; an L/C circuit or a multiple crystal
network will do well. If all you want is to recover the carrier, then a
single crystal will do an admirable job.

Here are a couple links to look at:

http://www.k8iqy.com/testequipment/pvxo/Atlanticon2002V1R5.pdf

The Bode Plot of figure 2 is valuable in showing that the pass and reject
frequencies are different, and you need to deal with that by tuning the
network.

And


http://darleys.pwp.blueyonder.co.uk/radio_07/advanced/techasp12.htm

This has an assortment of filter design insight, but mainly, it has a
simple, generalized schematic for a single crystal filter network, that
might be helpful.

Assuming you will build a TRF receiver, the filter would be at 60 kHz., and
the transformer would be fairly simple to build or buy.
 
On Fri, 02 Nov 2007 15:53:00 +0000, Ecnerwal wrote:

Is it as simple as "slap a 60KHz crystal in line with the input (or the
feedback path of an amplifier), and only 60KHZ will get through"
Click to expand...
Pretty much, yeah, with a couple of caveats. The crystal has to have a
series resonance at that peak, and you will get a VERY narrow bandwidth;
and there are phase considerations either side (which I guess there are
with any resonant circuit). And the passband will be _exactly_ at the
freq. of the crystal; I don't want to guess what the bandwidth or
damping factor might be, but I guess the Q can be in the thousands.

Many years ago, I saw some crystal filter designs, and they used sets
of crystals that were different in frequency by about the bandwidth you
wanted - they were used in SSB transmission. Also mechanical and ceramic
filters; I've even seen mechanical filters centered at 455KHz, so 60KHz
should be no problem, if you could find one or design it. :)

Good Luck!
Rich
 
On Fri, 02 Nov 2007 15:53:00 GMT, Ecnerwal
<LawrenceSMITH@SOuthernVERmont.NyET> wrote:

When I go looking for information on crystal filters, everything I can
find seems to be aimed at using a bunch of crystals and capacitors to
widen the bandwidth, or else it's marketing for people who would be
happy to sell me crystal filters of the above sort. I don't want to
widen the bandwidth...
Click to expand...
There's a pretty nice freeware filter design program at
http://www.aade.com . I used it to design an SSB filter recently.
After I'd characterized my crystals, the results I measured from the
completed filter compared surprisingly well to the numbers the AADE
program generated.

Tom
 
Ecnerwal (LawrenceSMITH@SOuthernVERmont.NyET) writes:
When I go looking for information on crystal filters, everything I can
find seems to be aimed at using a bunch of crystals and capacitors to
widen the bandwidth, or else it's marketing for people who would be
happy to sell me crystal filters of the above sort. I don't want to
widen the bandwidth...

It's far easier to build a filter that is narrow than one that is
Click to expand...
wide.

For a long time, the only crystal filter in receivers was a single
crystal, with a phasing capacitor to knock out the capacitance of
the crystal's holder. That was when narrow selectivity first hit
shortwave receivers, and even when other things came along, that sort
of filter continued. It was really selective, though of course the
skirt wasn't so great (ie it was nice and narrow at 3db down, but
further down the curve it would broaden out). Later, when they
wanted to use it for voice bandwidth, they'd load down the crystal to
broaden the peak.

A lot of later work is about getting a wider bandwidth, and maybe
more important, a controlled passband. So early wider crystal filters
were of a lattice nature, using a pair of crystals (or more if you
wanted to improve the skirt) on different frequencies. The spacing
set the bandwidth.

Later, people started playing with ladder filters, where the crystals
were all on the same frequency, and the associated capacitors and loading
resistors affected the bandwidth. The greater the number of crystals,
the greater the skirt selectivity.

So for a narrow filter, a single crystal in the signal path may be fine,
if the requirements aren't too heavy. More crystals won't narrow the
passband, because it's already narrow enough, but it will improve the
skirts.

You may find it is too narrow, and hence need some sort of loading,
and you may find a need to null out the capacitance of the crystal
holder.

What you will find is that it's likely a very narrow filter, which
won't matter for WWVB but would likely mess things up if you were
trying to receive traditionally modulated signals.

One neat trick, that seemed to circulate when ceramic resonators
came along, was to simply used them instead of cathode/emitter bypass
capacitors, so they'd show low impedance at the resonator's frequency
but high impedance elsewhere, so the gain of the stage was sharply
peaked at the frequency of the resonator. Cascade a few stages
of that, and you get your improved skirt selectivity. Nothing was
needed beyond the crystal, so it made for a very easy method
of improving selectivity.

The same thing can be done using crystals.

Michael
 
On Fri, 02 Nov 2007 15:53:00 +0000, Ecnerwal wrote:

When I go looking for information on crystal filters, everything I can
find seems to be aimed at using a bunch of crystals and capacitors to
widen the bandwidth, or else it's marketing for people who would be
happy to sell me crystal filters of the above sort. I don't want to
widen the bandwidth...

What I'm presently interested in is an exceedingly narrow-band filter
and/or amplifier. 60Khz for the monent, as I'm playing with WWVB - and
while I can buy an off-the-shelf amplifier or even amplifier and digital
buffer chip for $3-4 (CME6005 and CME8000) which uses an external
crystal to get very narrow bandwidth, there's no real discussion of
what, precisely, that's doing. As a practical person, I may end up
buying one of these units as a major building block, but I want a better
general understanding going into it, as I've really done very little
with radio, and this is as much about learning what's what as the actual
end result.

All the DIY WWVB receivers I've found are using LC filters, which is
classic, but seems to be more prone to noise than a very narrow
bandwidth filter would be, given that in this case the frequency of the
signal is _very_ well defined - in fact, given that I can get "the
correct time" rather easily from any network attached computer these
days, the frequency reference is as much or more of a driving reason as
the time codes are. I don't have any logical reason to spend money
sending my elderly ovenized crystal reference out for calibration, but
I'd still like to be able to maintain it on spec as much as possible -
which is, in theory, possible from the WWVB signal. The units mentioned
above claim to get about a 10HZ bandwidth, though I'd call the datasheet
"sketchy" at best.

Is it as simple as "slap a 60KHz crystal in line with the input (or the
feedback path of an amplifier), and only 60KHZ will get through"
(perhaps with some bypassing to block any multiples: 120KHZ, 180KHz,
etc.)? That seems a bit too simplistic, but other than a reference to
Walter Cady's 1922 paper (which I have not tried to go track down in
person) using single crystals for very narrow bandwidth coupling (in
this article: http://www.ieee-uffc.org/freqcontrol/crystal.html ), I've
somehow managed to miss any discussion of very narrowband crystal
filters at the most basic (or too basic to be discussed) level.

I need to clear the paper and clutter off my bench and get to playing
with this, having finally gotten a sweep generator.
Click to expand...
Try searching for "crystal ladder filter". If that doesn't work, go to
the ARRL site (http://www.arrl.com) and get this year's Handbook or one of
their radio circuits design books. They may even have a book specifically
on crystal filters.

At any rate, it can be done. When you're building a cheap crystal ladder
filter you often find that the center frequency isn't exactly the marked
crystal frequency -- this would be a problem if you want a 60kHz filter,
'cause you'd have to buy custom-made, low-frequency crystals.

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
 
In article <b9ednU8U9a3qdbbanZ2dnUVZ_qbinZ2d@web-ster.com>,
Tim Wescott <tim@seemywebsite.com> wrote:

At any rate, it can be done. When you're building a cheap crystal ladder
filter you often find that the center frequency isn't exactly the marked
crystal frequency -- this would be a problem if you want a 60kHz filter,
'cause you'd have to buy custom-made, low-frequency crystals.
Click to expand...
They certainly seem to available as mass-produced, $1-2 parts from
Mouser and Newark, most specified to + or - 30PPM (about 1.8 hz, if I
didn't slip a digit). Mouser has a + or - 20 PPM for a buck. Same for
100KHz (if one was going after Loran-C, another possible frequency
reference.)

--
Cats, coffee, chocolate...vices to live by
 
On Sat, 03 Nov 2007 15:05:02 +0000, Ecnerwal wrote:

In article <b9ednU8U9a3qdbbanZ2dnUVZ_qbinZ2d@web-ster.com>,
Tim Wescott <tim@seemywebsite.com> wrote:

At any rate, it can be done. When you're building a cheap crystal ladder
filter you often find that the center frequency isn't exactly the marked
crystal frequency -- this would be a problem if you want a 60kHz filter,
'cause you'd have to buy custom-made, low-frequency crystals.

They certainly seem to available as mass-produced, $1-2 parts from
Mouser and Newark, most specified to + or - 30PPM (about 1.8 hz, if I
didn't slip a digit). Mouser has a + or - 20 PPM for a buck. Same for
100KHz (if one was going after Loran-C, another possible frequency
reference.)

If you just build a ladder filter from the ARRL references you'll get a
Click to expand...
center frequency that's a bit off from the crystal's marked frequency.
Hence the "custom made" in my post.

However, thinking about this in a new day, it occurs to me that you could
probably load each crystal in the ladder with an inductance or capacitance
that would bring the filter center frequency into line, maybe.

Some SPICE simulations or just computations on paper should tell the truth.

--
Tim Wescott
Control systems and communications consulting
http://www.wescottdesign.com

Need to learn how to apply control theory in your embedded system?
"Applied Control Theory for Embedded Systems" by Tim Wescott
Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top