Hi, 
Guest
Pic
Sign In

time-nuts@lists.febo.com

Discussion of precise time and frequency measurement

View all threads

Re: New design for a GPS disciplined OCXO or Rubidium

HM
Hal Murray
Wed, May 15, 2024 8:15 AM

Bob kb8tq said:

What you will run into on the F9T output are often called “hanging brid=

ges”.

PavelK said:

I am further trying to minimise such errors with averaging inside PI lo=

op.

Averaging doesn't work with hanging bridges -- that's why they hang.  You=
get=20
hanging bridges when the clocks are almost in sync -- the offset doesn't =
drift=20
much from sample to sample.  You will be averaging over several samples o=
f=20
almost the same value.

--=20
These are my opinions.  I hate spam.

Bob kb8tq said: > What you will run into on the F9T output are often called “hanging brid= ges”. PavelK said: > I am further trying to minimise such errors with averaging inside PI lo= op. Averaging doesn't work with hanging bridges -- that's why they hang. You= get=20 hanging bridges when the clocks are almost in sync -- the offset doesn't = drift=20 much from sample to sample. You will be averaging over several samples o= f=20 almost the same value. --=20 These are my opinions. I hate spam.
AT
Andy Talbot
Sat, May 18, 2024 11:49 AM

Hence the value of using an unstable or jittered clock in the processor.

On Sat, 18 May 2024 at 12:42, Hal Murray via time-nuts <
time-nuts@lists.febo.com> wrote:

Bob kb8tq said:

What you will run into on the F9T output are often called “hanging

bridges†.

PavelK said:

I am further trying to minimise such errors with averaging inside PI

loop.

Averaging doesn't work with hanging bridges -- that's why they hang.  You
get
hanging bridges when the clocks are almost in sync -- the offset doesn't
drift
much from sample to sample.  You will be averaging over several samples of
almost the same value.

--
These are my opinions.  I hate spam.

time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com

--
Andy (out keeping fit)

Hence the value of using an unstable or jittered clock in the processor. On Sat, 18 May 2024 at 12:42, Hal Murray via time-nuts < time-nuts@lists.febo.com> wrote: > Bob kb8tq said: > > What you will run into on the F9T output are often called “hanging > bridges†. > > PavelK said: > > I am further trying to minimise such errors with averaging inside PI > loop. > > Averaging doesn't work with hanging bridges -- that's why they hang. You > get > hanging bridges when the clocks are almost in sync -- the offset doesn't > drift > much from sample to sample. You will be averaging over several samples of > almost the same value. > > > -- > These are my opinions. I hate spam. > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com > -- Andy (out keeping fit)
PK
Poul-Henning Kamp
Sun, May 19, 2024 9:24 AM

Andy Talbot via time-nuts writes:

Hence the value of using an unstable or jittered clock in the processor.

That may look like it improves the situation, but it does not.

The fundamental problem with "hanging bridges" is that they do not
average to zero over any predictable time interval, not even the
12 hour orbital period, so there is no way to "filter them out".

You can make it look like you filtered them out, but you will find
that all you have done is to introduce noise at longer taus, probably
a LOT of noise.

Once you get to the level of precision where you can see the first
hints of the hanging bridges, you /have/ to start applying the "residual"
communicated by the receiver.

That has been attempted in "pure hardware" both with programmable
delay-lines and DAC + adder + Sample & Hold circuitry.  I have yet
to hear anybody praise the results, temperature stability or ease
of implementation.

Hanging bridges is where sane people switch to a software PLL.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Andy Talbot via time-nuts writes: > Hence the value of using an unstable or jittered clock in the processor. That may look like it improves the situation, but it does not. The fundamental problem with "hanging bridges" is that they do not average to zero over any predictable time interval, not even the 12 hour orbital period, so there is no way to "filter them out". You can make it look like you filtered them out, but you will find that all you have done is to introduce noise at longer taus, probably a LOT of noise. Once you get to the level of precision where you can see the first hints of the hanging bridges, you /have/ to start applying the "residual" communicated by the receiver. That has been attempted in "pure hardware" both with programmable delay-lines and DAC + adder + Sample & Hold circuitry. I have yet to hear anybody praise the results, temperature stability or ease of implementation. Hanging bridges is where sane people switch to a software PLL. -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.
Empathy v1.0   2025 ©Harmonylists.com