EMRFD Message Archive 8859
Message Date From Subject 8859 2013-07-30 20:04:06 Bob L Crystal Selection for 20 Meter Filter
I can't find references in the popular literature to the suitability of these crystals for a 6 filter of Cohn or Butterfield design.
What say yee? Are third overtone filters acceptable?8860 2013-07-30 20:15:03 Ashhar Farhan Re: Crystal Selection for 20 Meter Filter
A simple two transistor rig as described in EMRFD and a few minutes
with the frequency counter should have you all set with the motional
parameters.
-farhan
On 7/31/13, Bob L <bobledoux@proaxis.com> wrote:
> I have a lot of 50 Ecliptek ECX2449 crystals. They appear very consistent
> and oscillate at 8.87 Mhz. I looked them up and they appear to be 26.590
> Mhz third overtone crystals.
>
> I can't find references in the popular literature to the suitability of
> these crystals for a 6 filter of Cohn or Butterfield design.
>
> What say yee? Are third overtone filters acceptable?
>
>
--
Sent from my mobile device8861 2013-07-30 21:35:13 Kerry Re: Crystal Selection for 20 Meter Filter
One of my references says "... it is important to understand that every crystal is able to operate in fundamental or any overtone mode, at series and parallel resonance. It is just a matter of matching the crystal manufacturer's matching condition to the condition applied to the crystal terminals by the surrounding circuitry.".
The reason that either fundamental, 3rd OT, 5th OT etc must be specified is that the overtones are not true harmonics; their frequency does not exactly equal 3x, 5x etc of their fundamental.
Makers calibrate the crystal to the required final frequency so, for instance, if you want a 60MHz 3rd OT crystal, the maker will supply a crystal whose 3rd OT is 60MHz at the load conditions you have specified; the fundamental frequency of this crystal will not be exactly 20MHz (although it will be close).
So, as Farhan advised, measure the motional parameters of the crystals; the G3UUR method as per EMRFD requires the least amount of test equipment but the downside is that this method doesn't give Rs or Q so, if you have the equipment, measurement of Q would be advisable.
EMRFD covers this and there is also the very fine paper by Jack Smith K8ZOA;
http://www.cliftonlaboratories.com/Documents.htm
See about halfway down, Crystal Motional Parameters.
I think that measurement of Q is important; I find many modern crystals have poor Q as they are very much "made to a price".
Have fun; measuring 50 crystals by manual methods is a "mission" but it's a good "hands-on" experience.
I'm served my time on manual methods (all those described in the K8ZOA paper); that's why I love my N2PK & DG8SAQ VNAs!
Kerry VK2TIL.8862 2013-07-31 09:45:27 Bob L Re: Crystal Selection for 20 Meter Filter
My third overtone crystals have shown the following:
1. Significant variance in fundamental frequency. Among the 50 crystals my plus or minus one standard deviati8863 2013-07-31 14:07:20 Kerry Re: Crystal Selection for 20 Meter Filter
It's good to have your results; as I said, I've never tried OT crystals in filters and your experiment is the only one of its kind that I've seen.
Thanks for the feedback; I, and I expect many others of us, now know more than we did yesterday. :)
Kerry VK2TIL.
PS. I recently found this video on crystal manufacture in the olden days; I found it fascinating, so much so that I downloaded it for closer study.
http://archive.org/details/6101_Crystals_Go_to_War_01_20_16_218864 2013-07-31 16:52:27 Bob L Re: Crystal Selection for 20 Meter Filter
About 20 years ago I bought a lot of 200, 13.5 mhz crystals that I then sorted into groups by frequency variation.
I pulled 24 similar frequency crystals out and tested them using the G3UUR oscillator in figure 3.35 in EMRFD.
Sorting them by increasing frequency, I then selected each contiguous group of six crystals for a filter. The smallest variation for a group of six was 5 hz. The biggest variation was 20 hz. That's nice.
I've calculated various crystal parameters using the Java Applet provided by qrp.pops.net. Its a nice piece of software. Its basic shortfall is it reports Lm in henries to four decimal places. All my crystals reported as .0049, or .0050.
Here is the question: Which of the measured crystal parameters is most important for ladder filter design? I see recommendations of all crystals being within a range of 50 hz. But filter design programs ask you to enter an average value for your group of crystals.
Does it matter whether you select your crystals based on variation in parallel capacitance, Lm, Cm or frequency? A different set of crystals may result on the variable used for selection. Does this have any practical implications for filter performance?
I know the frequency difference between 20 meters and a 13.5 mhz filter could result in a sub-megahertz VFO. But I've also built an SI570 DDS VFO that could work at the sum of the frequencies, about 27 mhz. That should eliminate birdies and other intermodulation effects.8865 2013-07-31 17:20:44 William Carver Re: Crystal Selection for 20 Meter Filter
sideband" ladder filter (coupling capacitors to ground) because the
crystals are used at series resonance in this filter topology.
The series Lm and Cm are resonant at the series frequency of the crystal
so if you know one you can compute the other. Because they're related by
resonance, the software I've seen let you design just knowing the L and
you never have to compute Cm.
R is related to L (or C) by the crystal Q.
So if you know L and R, Q = 2*pi*f*Lm divided by R.
Or, if you measure Q, and know L, R = 2*pi*f*Lm divided by Q.
Or, if you measure Q and R, L = Q*R divided by 2*pi*f.
Some test methods measure R by substitution: the seminal QST article way
back of a tester built by Doug DeMaw found R in this way. The you
measured His tester had you find the Q by finding the frequencies on
each side of resonance, fH and fL, where impedance increases by 1.414 (3
dB) the computing Q as resonant frequency divided by the difference
between fH and fL. This actually gives you the Q of the crystal IN THE
TESTER, so you have to "correct" it knowing the tester resistance and
the crystal resistance. Knowing the unloaded Q of the crystal alone, and
its series reistance, you can compute L.
G3URR has an alternate way to find the parameters. And a vector network
analyzer (hp, N2PK, DG8SAQ) have the ability to compute R, L, C and Q
after measurements around the resonant frequency of the crystal.
So there are several ways to find Q, R and L numbers for the design
process. They SHOULD all give the same result, of course. Depending on
what kind of instruments you have, or want to build, it can be
approached from whichever method suits you.
W7AAZ
On Wed, 2013-07-31 at 23:21 +0000, Bob L wrote:
>
> The third overtone crystals previously tested are out of the picture.
>
> About 20 years ago I bought a lot of 200, 13.5 mhz crystals that I
> then sorted into groups by frequency variation.
>
> I pulled 24 similar frequency crystals out and tested them using the
> G3UUR oscillator in figure 3.35 in EMRFD.
>
> Sorting them by increasing frequency, I then selected each contiguous
> group of six crystals for a filter. The smallest variation for a group
> of six was 5 hz. The biggest variation was 20 hz. That's nice.
>
> I've calculated various crystal parameters using the Java Applet
> provided by qrp.pops.net. Its a nice piece of software. Its basic
> shortfall is it reports Lm in henries to four decimal places. All my
> crystals reported as .0049, or .0050.
>
> Here is the question: Which of the measured crystal parameters is most
> important for ladder filter design? I see recommendations of all
> crystals being within a range of 50 hz. But filter design programs ask
> you to enter an average value for your group of crystals.
>
> Does it matter whether you select your crystals based on variation in
> parallel capacitance, Lm, Cm or frequency? A different set of crystals
> may result on the variable used for selection. Does this have any
> practical implications for filter performance?
>
> I know the frequency difference between 20 meters and a 13.5 mhz
> filter could result in a sub-megahertz VFO. But I've also built an
> SI570 DDS VFO that could work at the sum of the frequencies, about 27
> mhz. That should eliminate birdies and other intermodulation effects.8866 2013-07-31 18:36:20 Bob L Re: Crystal Selection for 20 Meter Filter 8867 2013-07-31 19:17:38 Kerry Re: Crystal Selection for 20 Meter Filter
I began by using a simple test fixture; on the right here with its variable resistor;
http://i39.tinypic.com/24cayj5.jpg
I used attenuators on each side to set the 50-ohm environment; a later version had in-built attenuators.
I found that the 100Hz step of my R & S SMS generator was too coarse so I had to resurrect my old Marconi TF144H from the basement;
http://i40.tinypic.com/34zi07a.jpg
I later got one of K8ZOA's fixtures; the "manual" for this fixture on his site contains good information.
The latest iteration (because I like my home-made items to look professional) is this pair;
http://i40.tinypic.com/154tzeg.jpg
The upper one is for crystals; it uses impedance-changing attenuators as Jack's fixture did to put the DUT in a 12.5 ohm environment as per the EIA standard.
The other is 50 ohms with impedance-stabilising attenuators on each side.
I found it useful to prepare & print "worksheets" for the different measurement methods; these had spaces for entry of the crystal number and the measured quantities such as Fc & F-3dBx2, Rs etc and following spaces for entry of calculated parameters.
I used a HP39G calculator into which I programmed the required equations.
Those who, unlike me, are clever with Excel could, no doubt, develop more sophisticated methods. :)
Kerry VK2TIL.8868 2013-07-31 19:55:50 William Carver Re: Crystal Selection for 20 Meter Filter
a stable, clean generator and a sensitive detector that can accurately
measure a 3 dB change in amplitude. If you have the test equipment to
support the simpler EMRFD setup you're in good shape.
Not mentioned in the EMRFD 3.34 text: the oscillator needs to have
fairly low harmonic content. The crystal will filter out oscillator
harmonics when you read the first amplitude. When you replace the
crystal with a pot, the pot will NOT filter out harmonics. An oscillator
with significant harmonic content (say an unfiltered transistor crystal
oscillator) will require a higher pot value so fundamental + harmonics
gives the same reading as the (filtered) fundamental only. IE, the pot
reading will be on the high side.
W7AAZ8869 2013-07-31 20:09:48 William Carver Re: Crystal Selection for 20 Meter Filter
> as Jack's fixture did to put the DUT in a 12.5 ohm environment as per
> the EIA standard.
>
With most HF crystals a 50 ohm environment will load the crystal down
dramatically. Extrapolating the loaded-Q measurement in the fixture, to
a loaded Q can produce some large errors.
For example a good HF crystal can have a series resistance of 5 ohms. In
a fixture with 50 ohms source and load, the corrected Qu will be 105/5
or 21 times the measured Q. If the source and load are not really 50
ohms, but, say, 47 ohms, then the correction should be 99/5=19.8
instead of 21, so you'll end up with corrected Qu being 6% high and
inductance will be off by 6%.
Jack's fixtures, since they don't load the xtals as much, start out with
a higher measured Q, and the correction to Qu involves multiplication by
a smaller ratio, so the overall accuracy of Qu is improved.
I don't want to make too big a deal out of this and discourage anyone,
but the more accurately you can characterize the crystals the closer
your finished filter will be to what you expect.
W7AAZ8870 2013-07-31 22:26:12 Kerry Re: Crystal Selection for 20 Meter Filter
Well, Bill, so you should; I learned most (if not all) of it from you during our e-mail discussions in 2008!
That discussion really helped me towards an understanding of crystals & crystal filters; I'm still learning though, it never really ends.
I still have copies of the e-mails and documents that you sent me.
Regards, Kerry VK2TIL.8871 2013-07-31 23:02:04 Kerry Re: Crystal Selection for 20 Meter Filter
I look for sets with high Q (as far above 100 000 as possible) and fairly close frequencies. I just do it "by inspection".
Here are a couple of sets I made-up recently after testing a batch of, I think, fifty;
http://i42.tinypic.com/9qbtee.jpg
I regard these as well-matched.
I got about five sets of five in all, ie half the batch is suitable for filters.
Wes' Ladpac programs allow "tuning" to bring each crystal to the same frequency so frequency is not necessarily so important.
From memory I think that the DJ6EV program also allows "tuning"; I don't think the AADE program does but it's been a few years since I've used either program.
Kerry VK2TIL.8872 2013-08-01 08:43:08 William Carver Re: Crystal Selection for 20 Meter Filter
continued well past that.
About 30 years ago (pre internet!) I called Wes Hayward on the
telephone. I was extremely nervous, but the lure of making filters was
strong. Wes was very gracious, explained some of the things I should
have understood, and sent me a paper design of a filter. He "launched"
me in the right direction and I'm forever in his debt. A GREAT guy.
Learning how to make filters, after decades of thinking they were some
kind of black magic, was an eye opener.
W7AAZ8873 2013-08-01 09:08:44 Vern Re: Crystal Selection for 20 Meter Filter
Keep forging ahead...
73,
-Vern
Sent from my iPhone
8876 2013-08-01 11:03:39 Philip Pollock Re: Crystal Selection for 20 Meter Filter
I am not sure about the use of a pierce oscillator for crystal frequency
measurement.
My understanding of this oscillator; is that the crystal inductance is
used with two capacitors to form a pie circuit.
The measured frequency will be lower than the marked. Maybe another
type would better measure the crystal frequency. I have just measured
ten 9.216MHz crystals with the pierce oscillator as described by W1FB in
his tester. They all have exactly the same frequency, highly suspicious!
The crystal C seems to be swamped out by the configuration.
Any comments?
Philip EI8JT
Crystals for Jan 2012 p32 75 metre superhet. Mine will use European S20
RX crystal in vxo with multiplier for 144MHz ssb use. I used the same
idea to put the Bingo 40M rig on Two metres. I am not an HF person. :>)
On 01/08/13 02:27, Bob L wrote:
>
>
>
> I found the DeMaw article you mentioned. The article was "A Tester for
> Crystal F, Q and R," originally printed in 12/1990, QST, page 21. I
> also found it in his "W1FB's Design Notebook"8877 2013-08-01 15:47:56 Kerry Re: Crystal Selection for 20 Meter Filter
A Pierce oscillator requires the crystal to operate in a parallel-resonance mode with specified load C which, in the operating circuit, is a combination of several capacitances.
The oscillator frequency is slightly higher than the crystal's series-resonance frequency.
Since series-resonance mode is the mode used in most filters we build the Pierce would seem to be not the best choice.
Kerry VK2TIL.8878 2013-08-01 15:51:25 kb1gmx Re: Crystal Selection for 20 Meter Filter
I've done this once and didn't use them for a ladder filter.
Tried the and it was just not good enough.
The crystals I had were 27mhz third overtone ex CB rocks for the
same channel (tx11 cb). I had maybe 50. While sorting them I found
I had 3 each of x and 3 more at x+1.45khz and nowhere near enough
for decent ladder filter as the delta on frequency was over 1khz!
The soluti8880 2013-08-01 16:53:05 William Carver Re: Crystal Selection for 20 Meter Filter
Chebychev, Gaussian, have pairs of crystals at opposite ends that are
on slightly different frequencies.
If you want to make a Butterworth filter, for example, and have a batch
of identical crystals, you actually have to move pairs of those
"identical" crystals the specified "offset" from the filter center
frequency. You have to make them not-identical, hi.
Wes' software tells you the required offset for each crystal in a
filter, and it has capability to tell you the value of series capacitor
that will move a crystal that amount.
So except for Cohn filters they don't HAVE to be identical frequencies.
You find the relative frequencies in a batch by plugging them into an
oscillator/counter test setup. Then you play "mix and match", finding
relative crystal frequencies that match the offsets the design software
specifies. Or real close, or maybe just needing a little moving with a
series capacitor again using Wes' software.
This is more tedious than just matching crystals, and trying this as
your first filter attempt is like putting on skis for the first time and
going to an Olympic slalom run. Cohn, with matched crystals, is a good
place to start. But once you figure out how to measure the Q, series
resistance and motional inductance of a crystal the sky is the limit.
W7AAZ8882 2013-08-01 22:42:09 Kerry Re: Crystal Selection for 20 Meter Filter
Although the maths is far beyond me, it's intuitive that the various meshes of a filter will be tuned differently depending on which polynomial (parrot's name? :) ) is adopted.
Kerry VK2TIL.8883 2013-08-01 22:54:53 Ashhar Farhan Re: Crystal Selection for 20 Meter Filter
for any measurements with the crystals, you need a signal generator and a
frequency counter. i suppose you have them both. if you scan the crystal in
parallel and series mode, you will be able to find out see the minima and
the maxima. These can lead you to calculating the Q, Lm, Cm and the
approximate internal resistance. Now all you have to do is plug them into
the filter design calculator.
you could alternatively, just use the cohn topology. Start with 100pf
capacitors all around, terminate it on around 200 ohms (for your crystals)
and listen. If it is too hollow, open up the filter (decrease the caps) and
listen again. ultimately, it is how good it sounds to your own ears that
matters. all the plots are only meant to indicate the goodness/badness of a
particular filter.
btw, if you have a separation of 2.4 Khz between the crystals, then you are
in luck! you can try a half-lattice filter instead! it has a better
passband and lower losses (the signal has two paths to traverse by,
lowering the losses of crystals).
- farhan
8884 2013-08-02 09:59:35 Philip Pollock Re: Crystal Selection for 20 Meter Filter
Sorry about that.
Philip EI8JT
On 01/08/13 15:51, Philip Pollock wrote:
>
> The article is in January 1990 QST.
> I am not sure about the use of a pierce oscillator for crystal frequency
> measurement.
>
> My understanding of this oscillator; is that the crystal inductance is
> used with two capacitors to form a pie circuit.
> The measured frequency will be lower than the marked. Maybe another
> type would better measure the crystal frequency. I have just measured
> ten 9.216MHz crystals with the pierce oscillator as described by W1FB in
> his tester. They all have exactly the same frequency, highly suspicious!
> The crystal C seems to be swamped out by the configuration.
>
> Any comments?
>
> Philip EI8JT
>
> Crystals for Jan 2012 p32 75 metre superhet. Mine will use European S20
> RX crystal in vxo with multiplier for 144MHz ssb use. I used the same
> idea to put the Bingo 40M rig on Two metres. I am not an HF person. :>)
>
[Non-text portions of this message have been removed]8885 2013-08-02 10:01:16 Bob L Re: Crystal Selection for 20 Meter Filter
I still think the best article I have read on Cohn filters is Wes Hayward's article, "Designing and Building Simple Crystal Filters," QST, July 1987, page 24; Also in the"W1FB's Design Notebook," page 186. It is not included on the EMRFD disk. It is available for download8887 2013-08-02 13:35:49 William Carver Re: Crystal Selection for 20 Meter Filter
> My SI570 generator has an ugly wave form so I'm building a low passGood show Bob! That filter is imperative for measuring crystals. The
> filter to clear it up.
very small tuning steps of the Si570 great for getting accurate
"delta-F" when using fixtures.
The square wave from the Si570 in theory, and pretty much in practice,
has mostly odd harmonics. So you're starting with the second harmonic
already smaller. This makes lowpass filtering a little easier.
W7AAZ8889 2013-08-02 13:47:15 William Carver Re: Crystal Selection for 20 Meter Filter
Measurement Approaches" (A GREAT PAPER!) page six evaluates the formula:
REFF = 2RL + RM = 2 * 12.5 * 6.52 = 31.52
The result is correct, but the second multiply sign should be a "+"
REFF = 2RL + RM = 2 * 12.5 + 6.52 = 31.52
Did you ever decide whether all your tests were wrong, or whether the
Sanders numbers were bogus? I'm chuckling. I've often found the Sanders
value for parallel capacitance was "off".
Bill - W7AAZ8890 2013-08-02 14:33:36 William Carver Re: Crystal Selection for 20 Meter Filter
> No big deal, butSorry guys, my reply finger broke: I didn't intend for that to be posted
to EMRFD. Ignore it.
W7AAZ8891 2013-08-02 16:42:16 Kerry Re: Crystal Selection for 20 Meter Filter
I quite agree that Jack's paper is the best that I know about crystal parameters and their measurement.
Just from memory I think that the parameters he obtained from his various measurent techniques agreed fairly well; it was the Sanders data that was the "odd one out".
Kerry VK2TIL.8892 2013-08-02 22:01:02 William Carver Re: Crystal Selection for 20 Meter Filter
technical significance. I did not intend it a group post.
Jack is extremely thorough. There's a handful of guys like him. N2PK,
PA3AKE among them. It would be interesting to know unequivocally what
the correct answer is for that crystal if you want to make a roofing
filter that has flat group delay, for example. Unfortunately it's like
having three watches that all show a different time, but two closer
together. Does that mean they are correct?
Ah well. At this point one of life's little mysteries.
W7AAZ
On Fri, 2013-08-02 at 23:42 +0000, Kerry wrote:
>
> It was clearly meant for Jack K8ZOA but I've often wondered the same
> thing; perhaps you or Jack could let us know.
>
> I quite agree that Jack's paper is the best that I know about crystal
> parameters and their measurement.
>
> Just from memory I think that the parameters he obtained from his
> various measurent techniques agreed fairly well; it was the Sanders
> data that was the "odd one out".
>
> Kerry VK2TIL.
>
>
>
>
>8894 2013-08-04 20:09:26 Gene Dorcas Re: Crystal Selection for 20 Meter Filter
somewhere where I could view those emails???
Thanks & 73,
Gene, W5DOR
gene@w5dor.com
www.w5dor.com