really slow PLL...

[server]

On Tue, 26 Jul 2022 08:05:49 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7



I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

cheers,
Gerhard

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV. Seems like it would be worth a bit of thought, though.

Cheers

Phil Hobbs

Imagine a single circuit/pcb that has N crystal oscillator circuits,
injection locked and summed, in an oven.

XOs near one another, namely in the same room, like to injection lock.

 

[Phil Hobbs]

jlarkin@highlandsniptechnology.com wrote:

On Tue, 26 Jul 2022 08:05:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7



I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

cheers,
Gerhard

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV. Seems like it would be worth a bit of thought, though.

Cheers

Phil Hobbs

Imagine a single circuit/pcb that has N crystal oscillator circuits,
injection locked and summed, in an oven.

XOs near one another, namely in the same room, like to injection lock.

Sure. That only gets you 10*log(N), though, AFAICT. Looking at it from
a phase noise POV, you win improved <delta phi> like sqrt(N), just as
you gain lower <delta V> by parallelling JFETs.

Cheers

Phil Hobbs


--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Joe Gwinn]

On Tue, 26 Jul 2022 08:40:45 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

Joe Gwinn wrote:
On Mon, 25 Jul 2022 20:51:18 +0200, Gerhard Hoffmann <dk4xp@arcor.de
wrote:

Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7


I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

As you suspect, no, it won\'t. Only better oscillators will help.

As Kipling might say, \"Not so, but far otherwise.\"

Well, for phase noise test sets, the rule is that noise reduces as
10Log10[ Sqrt[N] ], or simply 5Log10[N], where N is the number of
correlations performed, where what\'s being correlated is
data-collection runs with a specified data-collection window
durations. These windows can be over time or over devices, which
should be equivalent in the noise floor.

So, getting to very low PN levels this way soon becomes impractical,
and by far the best approach is to use a better oscillator. State of
the art these days is a noise floor at around -170 dBc/Hz at 10 MHz.
As always, 1/f^a noise can be a big problem, and it doesn\'t average
out all that well.

We may be talking about different things.


Joe Gwinn

 

[Les Cargill]

Lasse Langwadt Christensen wrote:

lørdag den 23. juli 2022 kl. 03.57.33 UTC+2 skrev Les Cargill:
bitrex wrote:
On 7/20/2022 8:22 PM, John Larkin wrote:
On Wed, 20 Jul 2022 19:32:20 -0400, Phil Hobbs
pcdhSpamM...@electrooptical.net> wrote:

John Larkin wrote:


Suppose I have several rackmount boxes and each has a BNC connector on
the back. Each of them has an open-drain mosfet, a weak pullup, and a
lowpass filtered schmitt gate back into our FPGA.

I can daisy-chain several boxes with BNC cables and tees.

Each box has a 40 MHz VCXO and I want to phase-lock them, or at least
time-align them to always be the same within a few microseconds,
longterm.

I could call one the leader (not \"master\") and make the others
followers (not \"slaves\") and have the leader make an active low pulse
maybe once a second. A follower would use her (not \"his\") clock to
measure the incoming period and tweak its local VCXO in the right
direction. That should work.

Don\'t GPS receivers lock their 10 MHz oscillators to a 1 PPS pulse
from the satellites?

My system should work from a 1 PPS GPS pulse too, all boxes as
followers.

The PLL algorithm might be interesting.


It\'s certainly possible. However, within whatever tiny loop bandwidth
you wound up with, the lockers would still have

20 log(40e6) = 152 dB

higher phase noise than the lockee.

GPS has that problem too.


It would be interesting to do the math to see whether it\'s possible to
generate a concensus lock for the group: if you get everybody close
enough, just sum their sine wave outputs and lock each one of them to
that, with some bit of AC coupling or something so that they don\'t all
wander together off to the edge of the tuning range.

Maybe have one doing the locking with a phase shifter and the others
with VCOs, or something like that.

Definitely a partly-baked idea, but surely one could do better than
152 dB!

Cheers

Phil Hobbs

Each box is basically a multichannel power supply, but channels can be
programmed to do stuff in timed sequences. I want different box
outputs to time align within, say, one millisecond longterm once
programs are kicked off together. So, many microseconds of equivalent
RMS phase noise is OK as long as we stay time aligned longterm.

It sounds like you\'re looking for a protocol like DMX if what you want
is to trigger sequences of events across boxes to within a millisecond,
I don\'t understand what this lock-the-40 MHz across boxes is about.

https://en.wikipedia.org/wiki/DMX512




DMX for this is like hunting deer with an artillery piece. DMX is for
the big-ass risk scenarios in distributed topologies; this is a lot
less profound.

? it\'s a 250kbit uart on RS485, hardly rocket surgery

That\'s the physical layer; DMX is pretty holistic. I dunno - maybe
there\'s a COTS DMX doohickey that can be pressed into service.

--
Les Cargill

 

[Les Cargill]

jlarkin@highlandsniptechnology.com wrote:

On Fri, 22 Jul 2022 21:10:35 -0500, Les Cargill <lcargil99@gmail.com
wrote:

jlarkin@highlandsniptechnology.com wrote:
On Thu, 21 Jul 2022 11:42:28 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:
snip
Phil Hobbs

Mathematicians often like music. In my experience, music fandom is
negatively correlated to engineering design skill. Different brain
structure or something.


Engineering is composition. Composition is the thin edge of the musical
wedge. Musicianship is different; it\'s pattern identification. As is
composition but in a different way. But it is all the same thing.

It all depends on which wall you prefer to have your back against.

I\'ve always wondered about musicians. They have to play a piece
hundreds of times to get it right.

Some do; some don\'t. Session players from back when studio time
was the dominant cost probably played the parts on a song you later
heard on the radio on the first take.

Some have surely performed
something thousands of times. Don\'t they get bored? Apparently not.

There\'s too broad a spectrum to generalize. Some forms are better for
people with mild forms of OCD.

I design something, finish, and then want to design something entirely
different.

It depends on boredom thresholds.

Much does.

One other thing I see a lot is undue respect for standards. As in \"you
can\'t do that because it violates SCPI standards.\" Where are the SCPI
Police when you need them?

Over where they MATLAB.

SCPI is send-and-forget. There is some query you can send to ask if
the last command worked. And you can have an error queue that you can
interrogate now and then for historical forensics.

I told the customer that damn the specs, every command is going to
reply with data, an error message, or \"OK\". They agree.

And there you go turning a perfectly good full duplex channel into a
half duplex walkie-talkie channel

It\'ll be fast enough.

--
Les Cargill

 

[Phil Hobbs]

Joe Gwinn wrote:

On Tue, 26 Jul 2022 08:40:45 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Joe Gwinn wrote:
On Mon, 25 Jul 2022 20:51:18 +0200, Gerhard Hoffmann <dk4xp@arcor.de
wrote:

Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7


I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

As you suspect, no, it won\'t. Only better oscillators will help.

As Kipling might say, \"Not so, but far otherwise.\"

Well, for phase noise test sets, the rule is that noise reduces as
10Log10[ Sqrt[N] ], or simply 5Log10[N], where N is the number of
correlations performed, where what\'s being correlated is
data-collection runs with a specified data-collection window
durations. These windows can be over time or over devices, which
should be equivalent in the noise floor.

Interesting. With phase-locked sources, when you simply average the
signals directly, the flatband amplitude and phase noise amplitudes go
down like sqrt(N), i.e. the noise power goes as 1/N. It\'s just like
additive noise.

Inside the loop BW, things are no doubt more complicated.

There are other situations such as OTDR where the amplitude dependence
is sqrt(sqrt(N)), but that\'s on account of electrical power going as the
square of optical power.

So, getting to very low PN levels this way soon becomes impractical,
and by far the best approach is to use a better oscillator. State of
the art these days is a noise floor at around -170 dBc/Hz at 10 MHz.
As always, 1/f^a noise can be a big problem, and it doesn\'t average
out all that well.

We may be talking about different things.

I expect so!

Cheers

Phil Hobbs

Joe Gwinn

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Anthony William Sloman]

On Wednesday, July 27, 2022 at 12:34:40 PM UTC+10, Phil Hobbs wrote:

Joe Gwinn wrote:
On Tue, 26 Jul 2022 08:40:45 -0400, Phil Hobbs
pcdhSpamM...@electrooptical.net> wrote:

Joe Gwinn wrote:
On Mon, 25 Jul 2022 20:51:18 +0200, Gerhard Hoffmann <dk...@arcor.de
wrote:

Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7


I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

As you suspect, no, it won\'t. Only better oscillators will help.

As Kipling might say, \"Not so, but far otherwise.\"

This might be the better oscillator.

https://spectrum.ieee.org/for-precision-the-sapphire-clock-outshines-even-the-best-atomic-clocks?utm_campaign=post-teaser&utm_content=7190c3vu

It\'s a big lump of sapphire in a pool of liquid helium. Long term stability isn\'t wonderful, but short term stability is uniquely good.

\" The team is re-engineering the device to work at 50 K by increasing the concentration of magnetic impurities in the crystal without introducing additional losses. That\'s a temperature that liquid nitrogen can\'t quite get to, but it\'s way easier than 6 K. \"

Apparently the Australian Air Force has two of the 6K devices for their radar network. Probably beyond John Larkin\'s budget.

--
Bill Sloman, Sydney

 

[Martin Brown]

On 26/07/2022 13:05, Phil Hobbs wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7

I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

Entrainment of weakly coupled oscillators at frequencies near to each
other can be quite strong (a problem if you don\'t want that to happen).

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

I think the catch is that to do that you would have to provide hardware
to compute the cross correlation of every pair of oscillators so that
correlator complexity goes up as N(N-1)/2 too. I can\'t immediately see a
way to exploit this to get a better average oscillator though.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV.  Seems like it would be worth a bit of thought, though.

VLBI typically disciplines a hydrogen maser using some other long term
stable centralised terrestrial time source. Getting it just a little bit
wrong just makes the white light fringe much harder to find later. Local
clock short term stability stability is the key to it working well.

I expect they are a lot better at it by now. In my day it involved
moving around furniture van loads of tweaked VHS video tape cassettes
from the big dishes to the correlator centres.


--
Regards,
Martin Brown

 

[server]

On Tue, 26 Jul 2022 19:56:53 -0500, Les Cargill <lcargil99@gmail.com>
wrote:

jlarkin@highlandsniptechnology.com wrote:
On Fri, 22 Jul 2022 21:10:35 -0500, Les Cargill <lcargil99@gmail.com
wrote:

jlarkin@highlandsniptechnology.com wrote:
On Thu, 21 Jul 2022 11:42:28 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:
snip
Phil Hobbs

Mathematicians often like music. In my experience, music fandom is
negatively correlated to engineering design skill. Different brain
structure or something.


Engineering is composition. Composition is the thin edge of the musical
wedge. Musicianship is different; it\'s pattern identification. As is
composition but in a different way. But it is all the same thing.

It all depends on which wall you prefer to have your back against.

I\'ve always wondered about musicians. They have to play a piece
hundreds of times to get it right.

Some do; some don\'t. Session players from back when studio time
was the dominant cost probably played the parts on a song you later
heard on the radio on the first take.

Some have surely performed
something thousands of times. Don\'t they get bored? Apparently not.


There\'s too broad a spectrum to generalize. Some forms are better for
people with mild forms of OCD.

I design something, finish, and then want to design something entirely
different.

It depends on boredom thresholds.


Much does.


One other thing I see a lot is undue respect for standards. As in \"you
can\'t do that because it violates SCPI standards.\" Where are the SCPI
Police when you need them?

Over where they MATLAB.

SCPI is send-and-forget. There is some query you can send to ask if
the last command worked. And you can have an error queue that you can
interrogate now and then for historical forensics.

I told the customer that damn the specs, every command is going to
reply with data, an error message, or \"OK\". They agree.



And there you go turning a perfectly good full duplex channel into a
half duplex walkie-talkie channel

It\'ll be fast enough.

One might feel a little silly, having sent 14,000 commands to a box
and then discovering that the power strip is off.

 

[Gerhard Hoffmann]

Am 27.07.22 um 08:21 schrieb Anthony William Sloman:

On Wednesday, July 27, 2022 at 12:34:40 PM UTC+10, Phil Hobbs wrote:
Joe Gwinn wrote:

As you suspect, no, it won\'t. Only better oscillators will help.

As far as crystals go, we have reached the end.

This might be the better oscillator.

https://spectrum.ieee.org/for-precision-the-sapphire-clock-outshines-even-the-best-atomic-clocks?utm_campaign=post-teaser&utm_content=7190c3vu

It\'s a big lump of sapphire in a pool of liquid helium. Long term stability isn\'t wonderful, but short term stability is uniquely good.

\" The team is re-engineering the device to work at 50 K by increasing the concentration of magnetic impurities in the crystal without introducing additional losses. That\'s a temperature that liquid nitrogen can\'t quite get to, but it\'s way easier than 6 K.\"

Apparently the Australian Air Force has two of the 6K devices for their radar network. Probably beyond John Larkin\'s budget.

More on this:

<
https://scholar.google.de/scholar?q=poseidon+sapphire+oscillator&hl=de&as_sdt=0&as_vis=1&oi=scholart
>

IIRC, Poseidon Scientific Inst. , a spin-off from that
Australian univ belongs to the Microchip timing empire now.



< https://sci-hub.mksa.top/10.1109/FREQ.1995.483927 >

(shorten that ridiculous filenme)
Driscoll seems to be everywhere it is interesting.

< http://rubiola.org/ >
The entire web site is most interesting,
Alone the reference section!
In this context:

< http://rubiola.org/pdf-articles/journal/2019-UFFC--TDDS-Sapphire.pdf >

< http://rubiola.org/pdf-articles/journal/2016-JPCS-8FSM--Sapphire.pdf >

and so on.
He is also into optical oscillators .


cheers, Gerhard

 

[Gerhard Hoffmann]

Am 27.07.22 um 10:36 schrieb Martin Brown:

On 26/07/2022 13:05, Phil Hobbs wrote:
Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

Entrainment of weakly coupled oscillators at frequencies near to each
other can be quite strong (a problem if you don\'t want that to happen).

But I want to lock them to the same frequency and phase anyway
with dedicated slooow PLLs. Injection locking would just
take away some control over the corner frequencies of the PLLs.
As long as the corners a low enough that is not a desaster.
Say below 0.5 Hz.

I did not see that the 2 Morion MV89A oscillators I use as a
cross corr reference talk to each other. The double oven and
the hermetically soldered box probably help. And there is a
frequency doubler to the output.

I think the catch is that to do that you would have to provide hardware
to compute the cross correlation of every pair of oscillators so that
correlator complexity goes up as N(N-1)/2 too. I can\'t immediately see a
way to exploit this to get a better average oscillator though.

My Timepod can do the 3 cornered hat only. Since 2 corners are
already reserved for the X-Band extension, I cannot remove the
noise of the reference by cross correlation via the 2
independent oscillators.

But then, the X-Band sources are not that wonderful. It\'s not
necessary to swim faster than the sharc, It\'s enough to swim
faster than your neighbour.

BUT IT\'S STILL FOR THE SPORTS!


Cheers,
Gerhard, DK4XP

 

[Jasen Betts]

On 2022-07-23, whit3rd <whit3rd@gmail.com> wrote:

On Friday, July 22, 2022 at 2:38:45 PM UTC-7, Don wrote:
Joe Gwinn wrote:

snip
Also, I\'d lose the BNC connectors. Threaded connectors like SMA, TNC,
and Type N are far better.

Or use shielded twisted pair to carry the 1PPS pulses.

Twisted pair (TP) sounds even easier than BNC. So, what\'s the
\"catch\" with TP? Where\'s the \"gotcha\" to make TP harder than BNC?

Biggest \'gotcha\' is the lack of good shielded TP connectors. I had only
UHF-style twisted pair shielded connectors last time I wanted some, and
that\'s a polarity-insensitive connector. We applied paint markings
to get it straight.

MiniDIN 3 (don\'t trust the shield connector) was what Apple used for their
LocalTalk/Appletalk hardware,

SVHS ran a higher bandwidth, unbalanced, over the 4 pin version.
localtalk was only a few hundered kilobaud.

Shileded pair conectors are fairly common now, USB, HDMI, SATA all give
multiple shielded pairs. (SATA is actully untwisted). Oh shit! I
forgot \"RJ45\" is also used for STP.

--




Jasen.

 

[server]

On Tue, 26 Jul 2022 11:03:15 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

jlarkin@highlandsniptechnology.com wrote:
On Tue, 26 Jul 2022 08:05:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7



I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

cheers,
Gerhard

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV. Seems like it would be worth a bit of thought, though.

Cheers

Phil Hobbs

Imagine a single circuit/pcb that has N crystal oscillator circuits,
injection locked and summed, in an oven.

XOs near one another, namely in the same room, like to injection lock.

Sure. That only gets you 10*log(N), though, AFAICT. Looking at it from
a phase noise POV, you win improved <delta phi> like sqrt(N), just as
you gain lower <delta V> by parallelling JFETs.

Cheers

Phil Hobbs

I\'m beyond my pay grade here, but summing jfets can be done with an
ideal isolated n-port summer and the s/n improvement indeed goes as
sqrt. But injection locking 10 oscillators is different. Each one
pulls towards the mean of the other nine. They herd one another.

 

[Phil Hobbs]

Martin Brown wrote:

On 26/07/2022 13:05, Phil Hobbs wrote:
Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7

I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

Entrainment of weakly coupled oscillators at frequencies near to each
other can be quite strong (a problem if you don\'t want that to happen).

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme,
but it seems as though it ought to be possible to combine the
measurements to produce N-1 oscillator signals, each one N times
quieter, so that averaging _those_ would get you to the N(N-1)/2 level.

I think the catch is that to do that you would have to provide hardware
to compute the cross correlation of every pair of oscillators so that
correlator complexity goes up as N(N-1)/2 too. I can\'t immediately see a
way to exploit this to get a better average oscillator though.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from
a total-hardware POV.  Seems like it would be worth a bit of thought,
though.

VLBI typically disciplines a hydrogen maser using some other long term
stable centralised terrestrial time source. Getting it just a little bit
wrong just makes the white light fringe much harder to find later. Local
clock short term stability stability is the key to it working well.

I expect they  are a lot better at it by now. In my day it involved
moving around furniture van loads of tweaked VHS video tape cassettes
from the big dishes to the correlator centres.

As the wise man said, \"Never underestimate the bandwidth of a truck full
of tapes.\"

Also, variously, a 747 full of tapes, CDs, DVDs, MicroSDs, etc. A
747-load of 256-GB MicroSDs is about
256e12 B * 113,400 kg / 0.25 g = 1.16E+23 bytes.

Six of them would be over 1 Avogadro.

Of course reading them out in less than the lifetime of the universe
would take quite a few boxes--it would need a bandwidth of
1.16E+23 / 3.156e+7 / 15e+9 = 245 kB/s just to do that.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Phil Hobbs]

jlarkin@highlandsniptechnology.com wrote:

On Tue, 26 Jul 2022 11:03:15 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

jlarkin@highlandsniptechnology.com wrote:
On Tue, 26 Jul 2022 08:05:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7



I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

cheers,
Gerhard

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV. Seems like it would be worth a bit of thought, though.

Cheers

Phil Hobbs

Imagine a single circuit/pcb that has N crystal oscillator circuits,
injection locked and summed, in an oven.

XOs near one another, namely in the same room, like to injection lock.

Sure. That only gets you 10*log(N), though, AFAICT. Looking at it from
a phase noise POV, you win improved <delta phi> like sqrt(N), just as
you gain lower <delta V> by parallelling JFETs.

Cheers

Phil Hobbs

I\'m beyond my pay grade here, but summing jfets can be done with an
ideal isolated n-port summer and the s/n improvement indeed goes as
sqrt. But injection locking 10 oscillators is different. Each one
pulls towards the mean of the other nine. They herd one another.

Yup. Inside the locking bandwidth, it\'s probably possible to make the
phases chaotic at some level, so the close-in noise might even be worse.

Outside that, though, as long as the peak phase error is smallish, both
amplitude and phase noise look additive, so the usual theorems apply.

For small epsilon,

sin(t + epsilon) = sin t cos epsilon + cos t sin epsilon

~= sin t + epsilon*cos t, (1)

so

(sin(t) + sin(t + epsilon))/2 ~=

sin t + (epsilon / 2) cos t. (2)


Using (1) backwards,

(sin(t) + sin(t + epsilon))/2 ~= sin(t + epsilon / 2).

With N different epsilons, you have a random phasor sum, which winds up
with an average phase error going like 1/sqrt(N).

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[server]

On Wed, 27 Jul 2022 10:29:44 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

jlarkin@highlandsniptechnology.com wrote:
On Tue, 26 Jul 2022 11:03:15 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

jlarkin@highlandsniptechnology.com wrote:
On Tue, 26 Jul 2022 08:05:49 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs

\"A geometric view of closure phases in interferometry\", DOI:
https://doi.org/10.1017/pasa.2022.6

.<https://www.cambridge.org/core/journals/publications-of-the-astronomical-society-of-australia/article/geometric-view-of-closure-phases-in-interferometry/5E8A5A8D58A2FC72ADFA0587347C4DA7



I\'m still digesting it, but basically deducing the underlying geometry
allowed for some significant improvements.

I have not yet digested it, but can I assume that it won\'t help
me to create a carrier that is phase noise wise better than
averaged over 16 oscillators created equally bad?

More suitable for post-processing after-the-fact?

U. Rohde has the math for n injection locked oscillators in one
of his books, but the formulas probably fall apart when you have
to insert hard numbers for real oscillators you can buy, or build.
Methinks he is more into multiple coupled resonators.

cheers,
Gerhard

I\'m not sure--as I say, I haven\'t got a properly-thought-out scheme, but
it seems as though it ought to be possible to combine the measurements
to produce N-1 oscillator signals, each one N times quieter, so that
averaging _those_ would get you to the N(N-1)/2 level.

It probably needs a whole lot of phase shifters or weighted summers
(like a Wilkinson with attenuators), so it may well not be a win from a
total-hardware POV. Seems like it would be worth a bit of thought, though.

Cheers

Phil Hobbs

Imagine a single circuit/pcb that has N crystal oscillator circuits,
injection locked and summed, in an oven.

XOs near one another, namely in the same room, like to injection lock.

Sure. That only gets you 10*log(N), though, AFAICT. Looking at it from
a phase noise POV, you win improved <delta phi> like sqrt(N), just as
you gain lower <delta V> by parallelling JFETs.

Cheers

Phil Hobbs

I\'m beyond my pay grade here, but summing jfets can be done with an
ideal isolated n-port summer and the s/n improvement indeed goes as
sqrt. But injection locking 10 oscillators is different. Each one
pulls towards the mean of the other nine. They herd one another.


Yup. Inside the locking bandwidth, it\'s probably possible to make the
phases chaotic at some level, so the close-in noise might even be worse.

Outside that, though, as long as the peak phase error is smallish, both
amplitude and phase noise look additive, so the usual theorems apply.

For small epsilon,

sin(t + epsilon) = sin t cos epsilon + cos t sin epsilon

~= sin t + epsilon*cos t, (1)

so

(sin(t) + sin(t + epsilon))/2 ~=

sin t + (epsilon / 2) cos t. (2)


Using (1) backwards,

(sin(t) + sin(t + epsilon))/2 ~= sin(t + epsilon / 2).

With N different epsilons, you have a random phasor sum, which winds up
with an average phase error going like 1/sqrt(N).

Cheers

Phil Hobbs

Of course the summing follows the usual linear equations AFTER the
oscillators are locked. But the things being summed are changed by the
phase locking, not independent sources any more.

If one oscillator is the big outlier, it gets all nine others pounding
on it to get in sync. Injection locking is fundamentally nonlinear.

 

[Jan Panteltje]

On a sunny day (Wed, 27 Jul 2022 12:12:42 +0200) it happened Gerhard Hoffmann
<dk4xp@arcor.de> wrote in <tbr32q$qg7c$1@solani.org>:

Am 27.07.22 um 10:36 schrieb Martin Brown:
On 26/07/2022 13:05, Phil Hobbs wrote:
Gerhard Hoffmann wrote:
Am 25.07.22 um 18:31 schrieb Joe Gwinn:
On Fri, 22 Jul 2022 09:03:16 -0400, Phil Hobbs


Entrainment of weakly coupled oscillators at frequencies near to each
other can be quite strong (a problem if you don\'t want that to happen).

But I want to lock them to the same frequency and phase anyway
with dedicated slooow PLLs. Injection locking would just
take away some control over the corner frequencies of the PLLs.
As long as the corners a low enough that is not a desaster.
Say below 0.5 Hz.

No idea what you are doing, but in the old days of Ampex broadcast videotape
recorders there were several loops on top of each other to get the color carrier to nano seconds precision in playback
1) slow one capstan tape speed
2) 4 head rotating scanning head to get the signal from tape
3) AMTEC (Automatic Time Element Compensator) a variable delay to get the video to micro second correct phase
4) Colortec a vaiable delay to sync the playback 4.43 MHz playback color carrier to the studio precision reference
all this so they could cross-fade and edit and use effects, and people\'s TVs would sync. and display color.

Old tech...
Slow PLL was not the way it worked,
From tape speed variations to nano second precision.
google Ampex AMTEC and Ampex Colortec
sixties and seventies

 

[Gerhard Hoffmann]

Am 27.07.22 um 16:13 schrieb Phil Hobbs:

I expect they  are a lot better at it by now. In my day it involved
moving around furniture van loads of tweaked VHS video tape cassettes
from the big dishes to the correlator centres.

As the wise man said, \"Never underestimate the bandwidth of a truck full
of tapes.\"

That was Andy Tanenbaum, either in his book \"Structured Computer
Organisation\" or in a guest lecture i saw at TU Berlin.
I was seldom more impressed by a prof.

He announced the \"Free Univerity Compiler Kit\", from the
Free Univerity Amsterdam.


Gerhard

 

[Gerhard Hoffmann]

Am 27.07.22 um 18:13 schrieb Gerhard Hoffmann:

He announced the \"Free Univerity Compiler Kit\", from the
Free Univerity Amsterdam. 

:gs/Univerity/University/

I hate that keyboard.


> Gerhard

 

[Gerhard Hoffmann]

Am 27.07.22 um 17:00 schrieb jlarkin@highlandsniptechnology.com:

Of course the summing follows the usual linear equations AFTER the
oscillators are locked. But the things being summed are changed by the
phase locking, not independent sources any more.

If one oscillator is the big outlier, it gets all nine others pounding
on it to get in sync. Injection locking is fundamentally nonlinear.

No, they all follow the Lucent GPS timing receiver and its own MTI-260
oven by a dedicated PLL in my case