When I asked about this on another Group a while back, I was told the phase
noise made it an undesirable solution.

Although I never tried it, I would imaging balanced shielded twisted pair
would be a better way to go.  Immune from ground loop problems than can
beset coax distribution and a whole raft of ICs support the medium, from
RS422 to CAN bus drivers and receivers.

Andy
www.g4jnt.com

On Sat, 28 Aug 2021 at 17:51, AC0XU (Jim) James.Schatzman@ac0xu.com wrote:

On 8/28/21 9:51 AM, AC0XU (Jim) wrote:

Take a look at the standard fiber optic modules that plug into a
switch.  They typically have a very simple logic level interface to the
diode.

here's a $27 dollar example - but I think you might want to iterate
around - I recall finding cheaper ones - part of it is which fiber
cable/connector you use, and part of it is what the backside interface
looks like. I can't find it now, but you can get the mating electrical
connector, or you can solder wires.

https://www.fs.com/products/11555.html?attribute=77&id=219732

The OVRO-LWA uses fiber to connect the remote antenna preamp with ~1km
fiber to the receiver at 30-90 MHz  - they use a laser diode on the Tx
end and a receiver on the Rx end. They're pretty cheap. They've done
quite a bit of testing of various schemes, and measuring noise
performance and loss.

dig around here: http://www.tauceti.caltech.edu/LWA/lwamemos.html

In particular memo 12

http://www.tauceti.caltech.edu/LWA/memos/memo12.pdf - gives a couple
part numbers.

If you need Sandy Weinreb's report (from 2019, referenced by Larry
D'Addario in memo 12) - you could probably email him, or I can look for
a copy.

On 8/28/21 9:56 AM, Andy Talbot wrote:

We use fiber to distribute high quality 10 and 100 MHz at JPL. I can
visualize the rack downstairs from my office where the cables terminate,
but I can't for the life of me remember the brand name. It's a 3U card..

DSN does it as well, with obsessive attention to temperature, mechanical
stress, etc.

From 20 years ago:
https://tmo.jpl.nasa.gov/progress_report/42-148/148L.pdf - probably overkill

PTS has a Fiber distribution amp: https://www.ptsyst.com/DA1-100-10-FO-B.pdf

There's this, but Im going to bet it's pricey:
https://emcore.com/products/reference-timing/

Look at Synchronous Ethernet.....

On Sat, 28 Aug 2021, 17:51 AC0XU (Jim), James.Schatzman@ac0xu.com wrote:

I was looking for something similar about 18 months or so ago.  Although I
haven't taken any
action yet, I concluded that one could do a nice job for under $200 per
segment, including the
transmitter and receiver modules and lots of connectorized multimode
fiber.  What I don't know
is what the phase noise performance would be, except that I do know that
the fiber's VF is
materially influenced by temperature.

I was looking primarily at the HFBR-2416 for the fiber receiver, and the
HFBR-1412 (standard
power) or the HFBR-1414 (high power option) for the transmitter  Unlike
most of the available
models, these are fundamentally analog devices, meaning that you can
transmit sinewave
10 MHz through them.  At the time I was looking, Mouser was selling these
for about $20 each.

See the datasheet at
https://www.mouser.com/datasheet/2/678/AV02-0176EN_2019-02-15-1827546.pdf
with particular attention to page 21 regarding the HFBR-2416 receiver.

I was also looking at Fiber Instrument Sales for patch cords.  We bought a
lot of fiber stuff from
them at Arecibo, and I was always happy with them.  See:
https://www.fiberinstrumentsales.com/catalog-cable-assemblies?gclid=Cj0KCQjwvaeJBhCvARIsABgTDM7eNTkP2nQbyFzhcwDE38VnSEP879MBKV1ZyDq2YrnEtOn7_VfzjbkaAtpfEALw_wcB

Somebody had pointed out yet another source of connectorized fibers ("patch
cords") to me, but I cannot find the name
at the moment.

Dana  K8YUM

On Sat, Aug 28, 2021 at 11:52 AM AC0XU (Jim) James.Schatzman@ac0xu.com
wrote:

White rabbit seems pretty good for time synced over fiber.  PTP with
enhanced stability profile.  It's a IEEE standard.  It seems to work for
CERN.  Lately I've been looking at Cisco devices that can support it.  It
whould be nice if I could find hardware that'll support it more easily...
Maybe it'll come soon?

If you need optics I highly recommend AMC optics they'll code (Cisco,
Intel, NetApp) SFP+ and QSFP for any vendor you want.  Prices are great and
it's all made the USA no relation to me.  I've been getting into 100G
Ethernet over OM4 fiber.  The new profile for PTP over fiber is legit.

Bill

On Sat, Aug 28, 2021, 12:04 PM Dana Whitlow k8yumdoober@gmail.com wrote:

Hi

With a whole variety of pretty good OCXO’s going for cheap money
on eBay, it’s likely less expensive to do cleanup oscillators on the link
compared to going crazy with low noise optical this or that. Yes, you
will be getting something in the high 150’s for noise, but still pretty good
for $10.

If you need better, spend $50 or so on a 10811. Still less money than
some of the crazy fiber stuff.

With either one, send over a high enough frequency that the loop isn’t
dealing with reference spurs in any significant way. Given the clock rates
Ethernet runs at these days, that part should be fairly straightforward.

This assumes that there is a crossover somewhere practical between
the fiber noise and the noise on the optical gear. You should be able to
work out what it is with some fairly normal phase noise or ADEV testing.

Bob

Bob, my own motivation  for going to fiber was entirely different.  I
simply wanted to
run 10 MHz all over the place from reference sources in disparate
locations in the
house, and I quickly discovered that cable leakage was embarrassingly
severe.  So
I shut down and began contemplating a fiber link.  But then reality set in
and I realized
that running fiber across the house had its own big problems, so I set the
project aside.

Dana

On Sat, Aug 28, 2021 at 3:32 PM Bob kb8tq kb8tq@n1k.org wrote:

Hi

Anything you do that has a bunch of 10 MHz cables running out from here or there
will impact your ability to listen to WWV at 10 MHz :)  Other than killing all the sources,
there is no silver bullet.

Bob

On 8/28/21 4:57 PM, Dana Whitlow wrote:

just thinking out loud here.. there's a bucket load of cheap "media
converters" around from fiber to twisted pair ethernet or RS422 -
including older 10 and 100 base T.  I wonder if they'd be repurposeable
for 10 MHz distribution.

or video over fiber widgets - used a lot for cameras and such. There's
been some discussion on the list about using video distribution amps for
10MHz.

Bob,

In my case simply disconnecting the long interconnecting cables at both
ends almost
completely eliminated the problem.  Hence my expectation that use of fiber
optics
would be effective.

Dana

On Sat, Aug 28, 2021 at 7:27 PM Bob kb8tq kb8tq@n1k.org wrote:

What about double shielded RG-223, or Thinnet ethernet cable?
Most Thinnet has both a braided shield and a foil wrap.

Robert

On 08/28/2021 08:26 PM, Bob kb8tq wrote:

Hi

Skin depth is what gets you with most coax at 10 MHz, even with the “double
shield” stuff. Hardline with a nice heavy outer would do better. It’s not terribly
practical around the typical lab. Might not be a bad choice for a “burry in the yard”
setup though. With normal cable, the losses from the soil at normal burial
depths make a good shield, that makes dealing with it questionable even there ….

Bob

I believe that a significant problem in my case was not leakage through the
shield, but
rather common-mode currents on the cable.  My experiences were with RG-6
style cable,
namely a cable that mostly went around the outside of the house for TV
distribution.
I get all the TV I want with an indoor dipole, and I had had plans of using
that TV
cable for ham purposes.  One thing I have noted is that adding ferrite
common mode
chokes near the ends of the cable did help quite a  bit, just not enough.

Those who are conversant with RF shielding concepts will recognize that
this puts most
of the blame squarely on my own shoulders, for not putting my kluges in
proper enclosures.
But I'm retired, old, shaky, tired, etc, and am not too ambitious any
more.  But I take
some solace in the observation that I'm not the only one; much of the
commercially-built
equipment I own has similar (and quite obvious) flaws.

I worked in the Tektronix spectrum analyzer group some years back, and one
of my side
tasks was taking instruments over to the EMI lab for compliance tests.  On
more than
one occasion those people told me that they were always delighted to test
stuff designed
by our group, because it was designed and built well enough that it almost
always passed
FCC specs on the first round.

Dana

On Sun, Aug 29, 2021 at 8:52 AM Bob kb8tq kb8tq@n1k.org wrote:

Hence my observation that balanced twisted pair might be a better solution
than coax for 10MHz distribution
Removes all possibility of ground loops

Andy
www.g4jnt.com

On Sun, 29 Aug 2021 at 15:22, Dana Whitlow k8yumdoober@gmail.com wrote:

Andy
The balanced line does work and does reduce the emissions on a 160 ft run.
Used it for 1, 5, 10 MHz.
Wellll my only issue is every time we had a lightning storm it seemed to
fry the ICs.
The line drivers and receivers were analog devices ADSL type technology.
Was a bit of a pain to deal with as they were smt devices. Tried various
suppression methods. I am sure there was a way. But I went back to coax.
I have also considered optics. HP had written a paper on using them circa
1990 approx. for frequency distribution.
Regards
Paul
WB8TSL

On Sun, Aug 29, 2021 at 10:34 AM Andy Talbot andy.g4jnt@gmail.com wrote:

Andy,

Balanced twisted pair may be helpful, but only if adequate precautions are
taken to
avoid creation of common mode currents on the line.

Dana

On Sun, Aug 29, 2021 at 9:34 AM Andy Talbot andy.g4jnt@gmail.com wrote:

exactly

On Sun, Aug 29, 2021 at 10:45 AM Dana Whitlow k8yumdoober@gmail.com wrote:

Robert wrote:

IME, thinnet cable is notoriously leaky.  I have also not had success
with double-shielded cables such as RG-223.

Unless you go all the way to hardline (some of which can be bent to some
degree for installation), there is not much you can do with shielded
coax.  No amount of braided and/or foil shielding will really do the job
effectively.  A braid-and-foil shield (or double-braid + foil) over a
twisted pair is better, but in practice (for our purposes) not usually
enough (again, IME).

Properly specified and installed triaxial cables can get you there,
but understand that you would need to change the connectors on your
instruments to gain the full benefit.  Also understand that proper
installation of triax cables and connectors is not trivial.

Best regards,

Charles

Andrew Sureflex (FSJ) is a lot more flexible then the LDF4 Healix, which
is what you find a
lot of on ebay and at hamfests.

Here is some 1/4"  Sureflex, new:

Andrew Type N Male to Type N Male Sureflex Cable 30 feet
F1RNA-PNMNM-30-L1 NEW!
https://www.ebay.com/itm/173863049072
US $30.00 + $13.02 Shipping.  For two pieces shipping goes to $16.33,
for three $22.90

You can cut it with a hacksaw, do NOT use wire cutters.  Connectors can be
installed with just hand tools.

If you move the test equipment around it's not very practical, but if
it's setup on
a bench or in a rack it will work.

Robert

On 08/29/2021 09:51 AM, Bob kb8tq wrote:

Hi

With any cable setup, proper termination is vital. An improperly terminated
balanced line can be just as bad as a poorly terminated coax. Things like
common mode chokes are part of a lot of setups.

What gets messy is that a lot of gear does not properly terminate the cable.
It’s high-z so you can run a bunch of them with Tee connectors or some other
strange approach. >20 db of return loss is not what most cables get terminated
in.

It doesn’t take a lot of coax to get a quarter wave at 10 MHz. Anything nearly
that long is going to be a radiator if you let it.

This all gets a bit insane if you have multiple 10 MHz sources (like devices
you are testing). Close in phase noise and ADEV both can get really strange
as a result.

Lots of issues ….

Bob

This has been an interesting thread.

I purchased some 1/4" FSJ with BNC connectors from the usual auction site years ago for time nuts use but never installed it.  It does seem to be a bit of a hassle to use in my time lab (I do use some FSJ with N connectors in parts of my home amateur radio station.)

I suspect I haven't looked hard enough for troubling leakage at 10 MHz from the RG400 I mostly use.  I have the FSJ to fall back on if I take my time nuts hobby to the next level.

Some of the fiber ideas seem promising as well.

Mark Spencer
mark@alignedsolutions.com
604 762 4099

On 8/29/21 10:29 AM, Robert DiRosario wrote:

ya know.. system cost for optical fiber is probably cheaper than
hardline - the cable certainly is (about $15 for a 100 ft duplex single
mode). And the analog/optical converters at either end are probably
cheaper than hardline coax connectors. And way easier to install.

We just need to figure out which COTS media converter is right for
"time-nuts" usage.

On Sun, Aug 29, 2021 at 10:43:51AM -0400, paul swed wrote:

While they may not have ideal behavior with temperature (and

related phase/group delay changes) the old antique 10Mbs era balanced
twisted pair ethernet  transformer/filter modules did eliminate common
mode energy getting into the receivers (and transmitter) which it sounds
like was the cause of your lightning EMP problem.  For 10 MHz these
ought to work pretty well with twisted pairs as they provide galvanic
isolation breaking ground loops.  Not completely clear what the common
mode  Z of the things is at 10 MHz... but unless there is a deliberate
cap from a center tap to local ground I'd guess it would be
significantly high given just winding to winding capacitance.

I do certainly beleive that optical transmission is the 

better solution... however.

--
Dave Emery N1PRE/AE, die@dieconsulting.com  DIE Consulting, Weston, Mass 02493
"An empty zombie mind with a forlorn barely readable weatherbeaten
'For Rent' sign still vainly flapping outside on the weed encrusted pole - in
celebration of what could have been, but wasn't and is not to be now either."

David I. Emery writes:

Twisted pair is 135 Ohm.

--
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.

On Sun, 2021-08-29 at 09:21 -0500, Dana Whitlow wrote:

Chiming in with a maybe tangential issue.

I had some really unreliable strange behaviour in my lab, and I traced
it to some low quality BNC crimp plugs. The noise at my receiver would
fluctuate by tens of dB, as I wiggled the connector. With a good
connector, it's dead quiet all of the time.

My conclusion was that the outer conductor was not making contact when
the plug was mated with a socket, and so outer currents were flowing
through the spring, and through the part of the plug that you rotate
when locking it to the socket. That adds some impedance, and your
receiver is now also listening to the common-mode current, that should
flow harmlessly to the chassis.

Visually, you can spot these particular bad plugs from the lack of
slots in the outer conductor. There's no way for it to compress as it's
mated with the socket, so they are under-sized instead. They often feel
loose. It may work sometimes, but not with all sockets, and only if
gravity is pulling on the cable just right.

All of the ones I've received from China have been like this. They went
in the bin. The ones from Jaycar, my local electronics shop, appear to
be identical. I used them anyway, because I was travelling a lot. Since
that time, I have been cutting them off my cables, hopefully I've
gotten them all.

It's been a sad lesson. Time and money down the drain, but at least I
worked out what was going on in the end.

Maybe someone else is struggling with this issue? Try wiggling all the
connectors :)

Have fun,
Darren

FWIW we've played with various SMA-to-SFP converter boards in the lab for
10MHz or 100MHz distribution.

A bare-bones SFP board is e.g.
https://github.com/aewallin/SFP-Breakout-Board
That interfaces directly to the differential AC-coupled TX and RX pins of
an SFP.

If you want some buffering (and limited bandwidth at the same time) you can
put an op-amp as a buffer and a transformer to convert to/from the
differential TX/RX.
Here's a design I've used https://github.com/aewallin/SFP2SMA_2018.03
We also made a box around that board:
http://www.anderswallin.net/2020/12/sfp-board-in-a-box/

If your requirements are somewhere in 1e-14 or worse then this type of
passive frequency distribution should work fine for maybe a few km fiber
runs.

Anders

On Mon, Aug 30, 2021 at 8:13 AM Poul-Henning Kamp phk@phk.freebsd.dk
wrote:

Hi

This sort of trouble with BNC’s has been going on a lot longer
than they have been coming in from China. A lot of folks use
them way past the point they should. They do indeed wear
out. They also come loose on the cable.

We went a bit nuts “killing off” all the BNC cables in the area
back in …errr … 1976. The result was a significant bump in the
supples budget that month and the elimination of a whole bunch
of problems on a number of tests. Back in that era Motorola
could afford to buy / fab a few cables.

Did the saved labor hours justify the expense? We claimed it
did……It most certainly made life easier.

Bob

On 8/29/21 11:07 PM, Anders Wallin wrote:

This is really cool..