EMRFD Message Archive 1725
Message Date From Subject 1725 2008-06-18 17:50:47 Glen Leinweber crystal pkg for LSB ladder filter
choice than HC-49 crystals when applied to wider
bandwidth (LSB) ladder filters. 49US crystals are in
low-profile metal cans.
The equivalent circuit for crystals has four components:
Lm, Cm, Rm = series motional equivalents
Cp = parallel capacitance of crystal plates
Of these four, Cp is directly measurable, usually at low
frequency where Lm, Cm don't interfere. Cp adds zeros
higher in frequency than the poles of the main filter
components Lm,Cm. These zeros distort the filter passband,
and put an upper limit on filter bandwidth. We'd like
Cp to be as small as possible.
I consistently measure Cp of a HC-49 crystal nearly
twice the Cp of an equivalent HC-49US crystal. I'd guess
that many of the surface mount crystals, and the surface-
mount plastic-case crystals might have low Cp too.
-Glen1728 2008-06-19 12:12:08 Wes Hayward Re: crystal pkg for LSB ladder filter
Generally, we have found just the opposite. That is, the HC49
crystals are preferred over the HC-49US, low profile. The low
profile parts do indeed have lower parallel C. But this means that
the motional C also decreases. This comes from the basic physics
of the devices that tells us that Cm = C0/K where in this case K is
typically around 200 to 220. The total parallel C is in two
components. One is the parallel plate C of the crystal while the
other is stray C. If the motional C drops (and motional L increases
to maintain resonance) this means that the shunt coupling caps in a
filter will also decrease. The tuning caps also drop in value.
There is a commercial transceiver that used the small profile
crystals in a SSB filter. The filter operated at VHF with the
crystals operating at the 3rd overtone (as I recall.) However, they
found that it was almost impossible to get the bandwidth up to a
desired 2.2 kHz for reasonable SSB. The radio was general coverage,
but the AM did not sound very good, I'm told. I never heard one.
I don't know if they every did get the problem fixed. Larger
crystals (HC-49) would have done the trick, although with complete
redesign of the filter. Also, you can get some control over these
things by playing the now popular "super VXO" game of paralleling
crystals. Going to two parallel rocks will essentially double Cm
and C0, and half Lm.
Clearly, simulations and constructi1729 2008-06-19 15:07:05 ve7ca2 Re: crystal pkg for LSB ladder filter
As a further note, if you wish to build a SSB bandwidth xtal filter
with low bandpass ripple try using higher freq. xtals.
I just designed and built an 8 sec. Butterworth 2.7 KHz BW filter
using 12 MHz xtals that has a measured band pass ripple of only 0.02
db, (measured with a N2PK Network analyzer). I also played around
with a wider BW filter at 3.6 KHz and it also had very low ripple,
under 0.03 dB.
On an other note, it would be helpful if experimenters posted their
measured xtal characteristics and the exact part number for xtals that
are ordered from parts suppliers such as Digikey, Mouser etc. If this
information were available it would help in selecting xtals for a new
xtal filter project without having to ordering a few xtals, testing
them and finding out that they are not suitable and then having to try
another brand, etc. One would still have to measure the xtals to
confirm that they are similar to the posted information.
73
Markus VE7CA
1731 2008-06-19 17:52:54 wimmie262000 Re: crystal pkg for LSB ladder filter 1732 2008-06-19 19:30:37 Jim Kortge Re: crystal pkg for LSB ladder filter
-snip-
>Those that I've measured are here:
> On an other note, it would be helpful if experimenters posted their
> measured xtal characteristics and the exact part number for xtals that
> are ordered from parts suppliers such as Digikey, Mouser etc. If this
> information were available it would help in selecting xtals for a new
> xtal filter project without having to ordering a few xtals, testing
> them and finding out that they are not suitable and then having to try
> another brand, etc. One would still have to measure the xtals to
> confirm that they are similar to the posted information.
http://k8iqy.com/miscellaneous/XtalParms/xtal_parameters.htm
I agree with Markus and others, crystals from the same source with the same part
numbers will probably not match prior units. Each batch seems to have its own
characteristics. Each batch needs to be measured to know what they are.
73 and keep those crystal posts coming,
Jim, K8IQY1740 2008-06-23 11:24:02 Glen Leinweber Re: crystal pkg for LSB ladder filter
Went back and measured my crystals having a range of
plate capacitances. The ratio of plate capacitance to motional
capacitance was nearly constant for all - in the 250 ballpark.
It is this ratio that determines pole/zero spacing, with a low
ratio being desirable for filters. No advantage goes to
HC49US (tiny) crystals.
Then tried both type crystals in the EMRFD program
XLAD to see what kind of SSB filters worked out....
Crystal A: 7.0pf plate , Lm=.0588h, Qu=120000
Crystal B: 3.5pf plate, Lm=.1176h, Qu=120000
Filter was Butterworth, bandwidth 2000 Hz, 4 crystals
Filter A: End R=1741 ohms, C1,C3=35pf. C2=54.5pf.
Filter B: End R=3482 ohms, C1, C3=17.5pf, C2=27.2pf
Wow, these are VERY different filters, just changing Cp
from 7pf to 3.5pf and holding Cp/Cm ratio constant at 260.
As Wes suggested, the crystals having higher plate capacitance
made a more reasonable-looking SSB filter.
And as Markus suggested, filters are best designed from
measured crystal parameters - plop junkbox crystals into
someone else's filter design at your peril.
mea cupla, -Glen1744 2008-06-24 10:55:58 ve7ca2 Re: crystal pkg for LSB ladder filter
Glen and Wes.
Last year I ordered (from a surplus electronics supplier) 6 sets of 6
xtals per set for a total of 36 xtas. They were supposed to be matched
xtals. It turned out that the xtals indeed were matched quite close
in freq. but some of the xtals in each set of six came from different
sources so Cp, Cm and Lm varied considerably from xtal to xtal. I had
to measure each xtal individually then entered the info. into an
excel file and then sorting them. I was able to obtain 8 xtals with
very similar parameters for a SSB filter and 5 for a CW filter but it
took a lot of time.
Recently I ordered a bunch of 5 MHz xtals from Digi-Key #535-9025ND
and it is obvious they were from the same producti1749 2008-06-26 16:38:59 wimmie262000 Re: crystal pkg for LSB ladder filter 1750 2008-06-27 07:00:16 wimmie262000 Re: crystal pkg for LSB ladder filter 1751 2008-06-27 07:13:00 Leon Re: crystal pkg for LSB ladder filter
1752 2008-06-27 12:25:56 Jim Kortge Re: crystal pkg for LSB ladder filter
>1753 2008-06-27 14:24:14 t-online Re: crystal pkg for LSB ladder filter
why didn't you measure Cp ? Yes it is not needed directly, but there
is a constant relation between Cm and Cp. If you match for Cp, you also
match for Cm and if the frequency is the same you also match for Lm.
Some people match for Lm/Cm for SSB filters. Others for same frequency
( <50 Hz for CW-filters, 2.5% of SSB bandwidth).
73 de Guenter
dl5sdc
----- Original Message -----
1754 2008-06-27 15:45:11 wimmie262000 Re: crystal pkg for LSB ladder filter 1755 2008-06-27 15:45:12 wimmie262000 Re: crystal pkg for LSB ladder filter 1756 2008-06-27 18:03:43 kerrypwr Re: crystal pkg for LSB ladder filter
original poster) and Joop.
I have regarded Co as "stray" and, therefore, the smaller the better.
But Wes (who knows this stuff) tells us that "This comes from the
basic physics of the devices that tells us that Cm = C0/K where in
this case K is typically around 200 to 220 ".
Incidentally, for the "small" crystals that I have, K is about 350.
So it seems that Co is not just "stray"; it is a basic property of the
crystal.
Why?? I don't know. :(
And I have the same problems as Joop in selecting crystals.
In VK, and particularly in a rural area, many parts are
difficult/impossible to get; I envy the US & Europeans their parts
sources.
Crystals are a case in point; I become green with envy when I read
about bags of crystals for a dollar!
At up to A$4 each from our suppliers, a large assortment is out of the
question so I have to make do.
This raises Joop's question; which parameter is most important?
I think it's Q; that number is beyond the control of the filter
designer. Other parameters can be "tuned" but not Q.
I think frequency is #2 or even equal #1; the usual references advise
matching of F to within some % of the filter bandwidth.
I get crystals with the same (or similar) F but rather different Lm
(and therefore Cm) values.
Lm & Cm are inter-related (except for that pesky Co :) ).
So I look for a group of crystals with, firstly, good Q plus good
frequency similarity. I sort initially by frequency and find a group
by inspection, looking for good Q.
It's a bit of a "judgment" process, art mixed-in with science.
But I'm just a learner so far as crystal filters are concerned; some
"expert" comment would be appreciated.
Kerry.1757 2008-06-28 07:39:58 t-online Re: crystal pkg for LSB ladder filter
yes Co is related to Cm by physics (sorry I don't know more). Also by
measuring Co you have
stray C, you have to know how many (there is a document on Wes homepage). Co
can be measured
be AADE type of C-meter (sometimes you find on Ebay or elsewhere schematics
and PCBs on the web
for much lower price). Measuring Co with other methods may add stray (you
have to know how many, usually
about 1pF, see Wes paper on his homepage).
By measuring Co of small xtals you might have to keep more attention to the
stray, because
Co might be lower. So if you have K=350 there might have been more relative
stray measured.
By the was your Cm fits very nice to Co=6pF.
Co is needed for the Dishal method of filter design, which is supposed to be
the most accurat method.
Another way for the crystal paramters I have seen is to calculate the
arithmetic mean of Lm and Cm for
all the xtals you have measured. Design the filter for theses parameter set,
preferebly with the Dishal
method and 0.5dB ripple.
You can go on like:
A) Mix your xtal set. and put the needed number of xtals in. By chance you
should get a good filter.
or
B) You can do some of selection of F or Q.
73 de Guenter
dl5sdc
----- Original Message -----
1758 2008-06-28 07:40:07 wimmie262000 Re: crystal pkg for LSB ladder filter 1759 2008-06-28 15:14:17 kerrypwr Re: crystal pkg for LSB ladder filter
Thanks for the link to Hi-Q Australia; I hadn't found their "specials"
section. The crystals are certainly "odd" but some might be useful.
Dieter;
"By the wa(y) your Cm fits very nice to Co=6pF".
Co is 1.85 pF (AADE meter, verified with another meter with a zero
facility); Lm is in the region of 0.18 mH and Cm is around 5.5 fF.
ESRs are high; around 35-45 ohms. Qs are around 120K-180K.
I used the small crystals in my first foray into filter-building; a
narrow filter for my version of Wes' Measurement Receiver. It came
out really well; 215 Hz BW (I wanted 200-250) and nice smooth curve.
I wanted a wide filter option so, flushed with success, went to work
on a 2 kHz filter.
After several disasters I hit the books/internet and have learned a
great deal; the postings in this thread came at the right time.
I have some HC-49 crystals and have managed to select five with fairly
good parameters so I am now working on a new design.
One paper that really interested me was K8ZOA's;
http://www.cliftonlaboratories.com/Documents/Crystal%20Motional%20Parameters.pdf
There's great information there.
It describes crystal parameter measurements with a vector voltmeter;
since I have a HP8405A I will do some measurements as a check on my
series-resonance and G3UUR measurements.1760 2008-06-30 23:27:45 kerrypwr Re: crystal pkg for LSB ladder filter
:) ) has driven me to re-read material that I thought I understood;
again as usual, I didn't understand it at all!!
The learning process was greatly assisted by a comment from Bill
Carver; " you cannot make a filter wider than the difference between
series resonance of the crystal and parallel resonance with the
"holder" capacitance ".
That makes sense.
Wes, as ever, has covered this matter; you just have to look & think!
The "pole-zero" concept of crystal resonance is described on p. 104 of
Introduction to RF Design; Fig. 3.25 on that page shows the resonances
of a "real" crystal and a "perfect" one. The real crystal has a
pole-zero spacing determined by Co whilst I guess from the plot that
the perfect crystal has an infinite pole-zero spacing.
If my guess is correct, reducing Co brings the parallel resonance
frequency closer to the series-resonance frequency and, of course, the
converse applies.
I don't know why this occurs; I would have thought that increasing C
in a parallel LC circuit would reduce F and therefore narrow the
pole-zero gap.
Someone might care to comment on this.1761 2008-07-01 09:38:59 Kenneth Stringham Re: crystal pkg for LSB ladder filter
Ken - AE1X1766 2008-07-01 19:52:22 kerrypwr Re: crystal pkg for LSB ladder filter
This means that the capacitance doesn't add directly. It as Cm in
series with Co.
But increasing Co still increases the net C of Cm & Co, eg;
Cm = 0.005 pF, Co = 2 pF; total C = 0.00499 pF.
Cm = 0.005 pF, Cm = 5 pF; total C = 0.004995 pF.
The increase is tiny as we would expect since the shift in frequency
is very small.
If the series-resonance frequency is less than the parallel-resonance
one, increased Cm should bring the two closer together.
I have the loan of a copy of Zverev; it arrived yesterday.
He gives equations for both series & parallel resonance frequencies;
in the parallel case, his equation is such that increased Cm widens
the separation of the two frequencies.
I know, of course, that this is correct but I'm frustrated that I
can't see why; I'm missing something that's blindingly obvious, I'm sure.1776 2008-07-03 13:42:29 Glen Leinweber Re: crystal pkg for LSB ladder filter
determined by the physics of the device. It seemed
remarkable to me that crystals of radically different
shapes maintain similar ratios.
So I characterized some ceramic resonators of
various types, from 500 KHz up to 4 MHz. to
find that THEIR Co/Cm ratio is much lower, in the
10 to 15 ballpark, whereas quartz crystals have
Co/Cm ratio in the 200 to 250 ballpark.
This test supports Wes' claim (is he ever wrong?;-)
It also seems remarkable that quartz has at least
three properties that make it so very good:
-it is WAY piezoelectric
-it is temperature-stable
-it has very high Q (rings like a bell).
BTW: ceramic resonators make lousy LSB ladder
filters, with high loss and very low end R. But a
quick test in the USB ladder configuration looked
promising, especially for wider bandwidth filters.
-Glen1780 2008-07-04 23:11:50 kerrypwr Re: crystal pkg for LSB ladder filter
about why increasing Co doesn't narrow the difference between Fseries
and Fparallel.
I suggested that he post it here as he deserves credit but he hasn't
done so; I will do it but with full credit to Glen.
From my earlier post;
But increasing Co still increases the net C of Cm & Co, eg; Cm = 0.005
pF, Co = 2 pF; total C = 0.00499 pF.
Cm = 0.005 pF, Cm = 5 pF; total C = 0.004995 pF.
The increase is tiny as we would expect since the shift in frequency
is very small.
If the series-resonance frequency is less than the parallel-resonance
one, increased Cm should bring the two closer together.
*****************
Glen replied;
No...
Increased Cm moves both series/parallel resonance down in frequency,
but moves series resonance down more than parallel resonance.
Extending your example above:
Cm=0.01pF, Co=2pf, total C=0.0099502 pf
Cm=0.01pf, Co=5pf, total C=0.00998 pf
Now total C has changed 3%.
Your example above (with less Cm) changed only 1.5%
***********************
Simple once someone points it out!!!