Richard

Your assumption that if the 2 oscillators didnt drift in opposite
directions you wouldnt be able to see any differences in the datasets is
fallacious.
Unless off course, the time intervals being measured by the TICs are
sufficiently long that the short term instabilities of the 2 100MHz TIC
oscillators mask the TIC quantisation noise.

If your oscillators are sufficiently unstable for a fixed isolation
between the 2 then injection locking will not occur.
The amount of isolation required to prevent injection locking increases
dramatically as the frequencies of the the 2 tuned circuits approach one
another.

Independent timestamps of a PPS edge will differ according to the
timestamp quantisation.
This is entirely independent of the PPS edge to PPS edge jitter either
random or determinisitic.

With 10ns TIC quantisation there will be variations of up to 20ns or so
in 2 independent TIC measurements of the same time interval provided
each TIC uses synchronisers.
If no synchronisers are used then all bets are off.
After averaging this variation will be reduced, but will still be present.
The observed differences will be a combination of the 2 oscillator
relative instabilities and the TIC quantisation errors.

Averaging the 2 (or more) sets of phase error measurements will improve
the determination of the short term GPS PPS error by perhaps 40% for 2
oscillators by 1/SQRT(N) for N oscillators (if and only if TIC
quantisation noise dominates over the GPS timing noise) however this
doesnt generate any useful information for more closely disciplining the
2 (or more) oscillators.

You can achieve the same improvement by increasing the TIC resolution to
the point where GPS PPS random noise dominates.
However in this case you also generate data that can be used to
discipline each oscillator more closely.

Bruce

Bruce,
Once again, thanks for the explaination. I am using a common 100M OCXO
and not independent XOs for the two TICs in the dual design. That is why
the plots are so similar. What effect does this have on the disciplining
of the individual oscillators?

Thanks,

Richard

Richard H McCorkle wrote:

Richard

When the transitions (PPS and divided down OCXO ouputs) are close to the
100MHz clock edges the 2 TICs measurements may differ.
If you dont use synchronisers then the differences between the TIC
readings can occasionally be large and the averages of the TIC outputs
will be biased.

Sharing a common TIC oscillator will have no effect on the individual
oscillators although it will tend to increase the correlation between them.

Ideally the TIC resolution should be increased so that the PPS timing
jitter (after sawtooth correction) is greater than the TIC resolution,
and the TIC oscillators should then be phase locked to their respective
OCXOs to improve their stability. The PPS timing noise will then be
sufficient by itself (if the TIC uses synchronisers) for the averaged
TIC outputs to be effectively unbiased estimators of the phase errors of
their respective oscillators.
When using a PPS signal it only remains to determine the optimum
algorithm for discipling the OCXOs.
The improvement achieved by increasing the TIC resolution further is
somewhat marginal.

A Kalman filter technique may have some advantages over a PLL
disciplining technique.

If you are using software to correct the TIC measurements for the PPS
sawtooth error then you may need (depending on your GPS receiver's PPS
timing noise) a higher resolution TIC to improve the performance. If you
are using hardware to correct the PPS signal for  sawtooth error then a
simpler scheme usng a single D flipflop may suffice together with the
receiver's TRAIM capability (if any).

Bruce

From: Bruce Griffiths bruce.griffiths@xtra.co.nz
Subject: Re: [time-nuts] GPS noise reduction
Date: Sun, 06 Apr 2008 11:16:36 +1200
Message-ID: 47F80854.7040007@xtra.co.nz

Actually, in this application will a bit of interlocking not hurt, as the two
oscillators should longterm have the same frequency and interlocking will pull
them together to their average frequency. If you further aid this interlocking
by externally couple them together for higher frequencies (up to say 10-100 Hz)
then you can sum their outputs and get a reduced noise response, a 3 dB
improvement. The noise processes internal to the oscillators will be
uncorrelated where as the locking causes the signals to be in phase.

If we go back to the original problem, the two oscillators is to be steered to
the same frequency... so any interlocking between them is as such not a problem
as it achieves part of the goal. However, the bandwidth of the interlocking is
of interest if we also wants to measure the noise sources. Below the
interlocking bandwidth (above the interlocking tau) then oscillators will
behave more and more as one oscillator and differences will change. Above the
interlocking bandwidth, the individual noise sources is more distinct.

Still, you need to measure the A-B to make conclusive measures. Also, the TIC
resolution is indeed problematic for high frequency noise analysis.

Cheers,
Magnus

...

That's an interesting concept. One could couple a number of oscillators that
way (provided it is practically feasible) and improve phase noise, like you
can make a better LNA by paralleling several amplifiers.

For that to actually be beneficial however, the locking effect should be
mutual, i.e. each oscillator has the same effect on the other so that
"averaging" (probably not the right term, but you get the idea) takes place.
If coupling is assymetrical, then the dominant oscillator (the one that is
less affected by the others but affects the others) will drive the result
and you won't have any improvement over that oscillator alone. Since it is
very hard in practice to control coupling between two devices unless you do
it intentionaly (that means the devices are designed to be coupled in a
controlled fashion), it may be best (easier in practice) to do everything
you can to reduce coupling.

Didier

No virus found in this outgoing message.
Checked by AVG.
Version: 7.5.519 / Virus Database: 269.22.5/1358 - Release Date: 4/3/2008
6:36 PM

This is very true. I saw that HP did this in their cal lab with a
half-dozen(?) hand-picked 10811 in order to get a combined
virtual OCXO with better phase noise than any one physical
OCXO. Rick, maybe you remember some details of this cool
10811 ensemble?

/tvb

That also leads to a thought I've had -- if you have a pair of roughly
comparable oscillators at say 5 MHz, what about combining their outputs
in a mixer and using the 10 MHz sum output which (apart from noise added
by the mixer) ought to be sqrt-2 better than either unit alone?

John

Tom Van Baak said the following on 04/06/2008 12:04 PM:

Tom Van Baak wrote:

Tom

Phase noise is also reduced when an oscillator uses several crystals
(with the same resonant frequency) connected in series.

Bruce

John Ackermann N8UR wrote:

John

Lowest mixer noise is achieved when the difference frequency is
reflected back into the mixer whilst terminating the sum frequency in a
matched load.

You can also use several mixers and combine their outputs to reduce the
effective mixer noise.

Bruce

From: Bruce Griffiths bruce.griffiths@xtra.co.nz
Subject: Re: [time-nuts] GPS noise reduction
Date: Mon, 07 Apr 2008 09:19:11 +1200
Message-ID: 47F93E4F.2040502@xtra.co.nz

Yes, but it only reduces crystal related noise... where as amplifier noise is
not reduced.

Cheers,
Magnus