Luciano,
Thanks it gets saved in one of my folders. Pretty sure I will have a use
for it.
Regards
Paul
WB8TSL

On Thu, Apr 11, 2013 at 12:42 PM, Luciano Paramithiotti <timeok.it@gmail.com

Luciano Paramithiotti wrote:

It is easy to use filter software to design a filter like this,
but how well it actually works is highly dependent on the
quality of implementation.  Suppression is a function of the
Q of the resonators, the tuning tolerance, and possibly the
tempco.  Tuning tolerance gets into the issue of tuning granularity.

In the 5071A, what I elected to do is design the filters to use
off the shelf Coil Craft air core inductors and pairs of standard
value capacitors in parallel (coarse and fine).  The capacitor values
were chosen to account for PC board parasitics.  I lived with whatever
suppression I could get under these constraints.  I then designed
a filter with multiple traps per frequency such that I could achieve
100+ dB suppression.  This is not a minimum parts count design.
These boards were automatically loaded with components off a tape
and they simply worked when we turned them on.  Every time.
When it came to getting complaints from the production engineer,
I was as lonely as the Maytag repairman.  Nothing ever went wrong.
Compare this to the finicky 5061 frequency multiplier.
Filter software (at least what I have seen) doesn't do this
type of design.

Rick Karlquist N6RK
(5071A RF designer)

Maybe a silly question but isnt the phase response of the filter important
in this application ?? notches have fairly vicious phase shifts.

Alan
G3NYK
----- Original Message -----
From: "Luciano Paramithiotti" timeok.it@gmail.com
To: time-nuts@febo.com
Sent: Thursday, April 11, 2013 5:42 PM
Subject: [time-nuts] Low-pass Filter for 5 and 10 MHz

Actually, the opposite is true.  Notches have the least phase
shift at the frequency being passed, which is what matters.
It is true that the phase shift at the notch frequency is
uncontrolled, but that is not important.  The HP8662A
had an interesting PLL synthesizer where they had 10 notch
filters for the first 10 harmonics of the sampling frequencies.
This minimized phase shift within the loop bandwidth that
detracted from phase margin.  I designers of the 8662
definitely know what they were doing.

Rick Karlquist N6RK

On 4/11/2013 5:02 PM, Alan Melia wrote:

...but what about the phase jitter of the filter itself? While absolute
phase shift may not (or may?) be an issue I guess that passive filters
do have a phase jitter, too, due to mechanical vibration, tempco, and
what else.

Particularly at frequencies where the filter response has sharp slopes
(resonance or corner frequency) the phase variation (d phi / d f) is
quite big. Thus small frequency changes lead to considerable phase shift
variation what in turn should lead to phase jitter added to our holy
signal - what about overall ADEV?

Wouldn't it be better to not filter the 10 MHz signal when used solely
as a frequency standard?

I understand, that a high Q filter in a PLL reduces the phase noise of
that oscillator - until the jitter of the filter becomes important. Am I
wrong?

Volker

Am 12.04.2013 02:31, schrieb Richard (Rick) Karlquist:

On 4/12/2013 7:58 AM, Volker Esper wrote:

Most of the time, you don't have to worry about jitter in passive
filters unless you are in a high vibration environment.  The other
common cause of jitter is permeability modulation of magnetic cores
in inductors from power line fields.  However, in an atomic clock,
the clock stability is so high that phase tempco becomes significant.
A temperature ramp becomes a phase ramp which becomes a frequency
offset.  In the 5071A, I multiply the 10811 crystal oscillator to 80
MHz.  This 80 MHz is used for both the microwave synthesizer chain and
also to generate a stepped approximation to a 10 MHz sine wave.  This
greatly reduces harmonics up to 60 MHz, which allows a much less severe
output filter to be used.  This alleviates the effects noted above.
The multiplier to 80 MHz, another possible source of phase drift,
is common moded out in this architecture.

For further details, see my FCS paper.

Rick

Hi

A simple way to look at filter stability is to first look at group delay at
10 MHz (or what ever your signal frequency is). The higher the group delay,
the more likely you are to get into the sort of trouble you thinking about.
Low Q / wide band filters are your friend in this case. Band pass filtering
is going to be more of an issue than a low pass.

Unless you are playing with 1x10^-15 at tau=1 sec sort of signals, or really
crazy environments (missile launch), you can get away with some filtering on
the signal. It's likely that the harmonics you remove would have caused you
as much grief as the stability you compromise with the filter.

Bob

-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of Volker Esper
Sent: Friday, April 12, 2013 10:59 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Low-pass Filter for 5 and 10 MHz

...but what about the phase jitter of the filter itself? While absolute
phase shift may not (or may?) be an issue I guess that passive filters
do have a phase jitter, too, due to mechanical vibration, tempco, and
what else.

Particularly at frequencies where the filter response has sharp slopes
(resonance or corner frequency) the phase variation (d phi / d f) is
quite big. Thus small frequency changes lead to considerable phase shift
variation what in turn should lead to phase jitter added to our holy
signal - what about overall ADEV?

Wouldn't it be better to not filter the 10 MHz signal when used solely
as a frequency standard?

I understand, that a high Q filter in a PLL reduces the phase noise of
that oscillator - until the jitter of the filter becomes important. Am I
wrong?

Volker

Am 12.04.2013 02:31, schrieb Richard (Rick) Karlquist:

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.

Am 12.04.2013 18:24, schrieb Bob Camp:

That's what I meant (d phi / d f), I just wrote it in a roundabout way...

Yep, my apprehension...

I haven't launched missiles, yet :-), but am playing with GPSDO signals,
and planning to build a distribution amp - so I am a bit afraid about
using filters, at least high Q ones.

Thank you, Bob, for your comment.