On 11/15/2010 7:44 AM, John Ackermann N8UR wrote:

John, You don't say what width you require or any other details other
than it should be 50 Ohm, but these are relatively easy to build in a
ladder configuration with stock crystals. Google produces a lot of hits
for "crystal ladder filters"; the first:
http://www.aade.com/filter32/dishal.htm

Dan

Hi John I have a collection of filters but I have not seen any common ones
at those frequencies with 50 ohm termination. The characteristic would
probably require transformer input and output coupling, the inherent
impedance of most filers in this frequency range will be around 1kohm. The
narrower the lower the impedance.

Those crystal frequencies are available relatively cheaply it should be
possible to make a ladder filter with some form of matching to 50 ohms quite
simply. There are lots of references around (the recent QEX has on on
overtone crystals, but does reference some of the articles on fundamental
crystal filters.)

Alan G3NYK

----- Original Message -----
From: "John Ackermann N8UR" jra@febo.com
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Monday, November 15, 2010 3:44 PM
Subject: [time-nuts] 5 and 10 MHz crystal filters

https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts

A friendly warning about crystal filters.  Crystals, whether
used in an oscillator or filter, have intrinsic phase noise.
You cannot improve the phase noise of a crystal oscillator
with a crystal filter unless the filter crystals have lower
phase noise than the oscillator crystals.  In general, post
filtering of an oscillator is rarely done for the reasons
I mentioned.  One place where it made sense was in the HP
8662 with the 640 MHz output option where a 10811 is multiplied to 80
MHz and then filtered to reduce the noise floor.  It is further
multiplied to 640 MHz and filtered again by a SAW filter,
that was made at HP in those days.  As others have mentioned,
filter characteristic impedance is a function of the crystal
physics.  There is a specific impedance that you must use for
a given bandwidth.  Fortunately, it is easy to put transformers
at the ports to convert to 50 ohms.

Rick Karlquist
N6RK

On 11/15/2010 7:44 AM, John Ackermann N8UR wrote:

Rick -

Unless I  misunderstood, I do not believe what you said is correct. A
crystal filter  can indeed improve the far out phase noise of a crystal
oscillator. The oscillator's noise is not only a function of the crystal,
but, the combination of the crystal noise and the amplifier noise used for
feedback and the Q of that circuit. All of those additional components
degrade crystal noise. Consequently, a 100 Hz wide filter at 10 MHz can
really do a nice job on just a decent 10 MHz oscillator. Lincoln Labs proved
this in their EDM of a satcom system in the early 80's that then I took over
in industry to replicate and improve in 1986. We did have a 10 MHz crystal
filter in there for cleanup. In our application, the final transmit
frequency was close to 45 GHz, and it needed to be very clean, even after an
optimal architecture used to get us up there. 73 - Mike

Mike B. Feher, EOZ Inc.
89 Arnold Blvd.
Howell, NJ, 07731
732-886-5960 office
908-901-9193 cell

-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Richard (Rick) Karlquist
Sent: Monday, November 15, 2010 4:44 PM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] 5 and 10 MHz crystal filters

A friendly warning about crystal filters.  Crystals, whether
used in an oscillator or filter, have intrinsic phase noise.
You cannot improve the phase noise of a crystal oscillator
with a crystal filter unless the filter crystals have lower
phase noise than the oscillator crystals.  In general, post
filtering of an oscillator is rarely done for the reasons
I mentioned.  One place where it made sense was in the HP
8662 with the 640 MHz output option where a 10811 is multiplied to 80
MHz and then filtered to reduce the noise floor.  It is further
multiplied to 640 MHz and filtered again by a SAW filter,
that was made at HP in those days.  As others have mentioned,
filter characteristic impedance is a function of the crystal
physics.  There is a specific impedance that you must use for
a given bandwidth.  Fortunately, it is easy to put transformers
at the ports to convert to 50 ohms.

Rick Karlquist
N6RK

On 11/15/2010 7:44 AM, John Ackermann N8UR wrote:

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

I hadn't considered phase noise, thanks for bringing that up!  I know tempo can also be an issue with high Q filters.

These aren't significant issues for my current application, which is to clean up a synthesizer running at a slight offset from the nominal 10.000 MHz frequency.  So I'm starting with much more noise than an OCXO.

John

On Nov 15, 2010, at 4:43 PM, "Richard (Rick) Karlquist" richard@karlquist.com wrote: