Then I am quite lucky with my 5065A  so far :) :) :)

Well, as out of the box solutions there would still be these new optical
clocks (Inflection Tiqker, Vector Artomic EG-30, Vescent-DFM ...) that
promise active hydrogen maser performance for less than the (new) price
of an active hydrogen maser, but that is still $$$. Moreover the
experience with these is limited, I think.

Am 22.04.26 um 16:13 schrieb Bob kb8tq via time-nuts:

Hi

Based on the prices I’ve seen, an “entry level” active maser is about half the price of a optical clocks.
Since none of these folks publish price lists, it’s a bit tough to doccument that.

Bob

That's very interesting, I even have access to some mechanical expertise
with CNC lathes to modify it for the new filter. I'll have to see if I can
pick one of those up (in case anyone has one lying around...)

ADEV plot out to 1K seconds

I emailed them and asked them to do just this.

and we have a cleanup loop in the lab where's used.

I'd be very curious to hear what you use for the fiber distribution and
cleanup loop.  KaTS and the New Horizons receiver are basically what I'm
going for, though KaTS measures velocity down to something like 10
microns/sec out at Jupiter which is stability something like 6e-14@60s 🫪

I'll be very curious to hear people's experience with these as they become
more used.  I know that Shimadzu sells a strontium clock but at $3.5
million per it's probably out of our range for the time being =D.

Sean

On Wed, Apr 22, 2026 at 7:21 PM zfe--- via time-nuts <
time-nuts@lists.febo.com> wrote:

Bob, Sean,

For better results use more than one reference and apply
cross-correlation / 3-hat tricks. See 3 BVA here:

http://www.ke5fx.com/bva.htm

For reliable ADEV plots you want the reference to be several times
better (maybe 3x, or even 10x) than the oscillator you're trying to
measure. So at tau 60 s, 1e-13 is a challenge with a PHM, but AHM will
be fine. See 3 masers here:

http://leapsecond.com/museum/mhm-2010/

It sounds like your project is at the bleeding edge. IIRC, there are
informative JHU or JPL papers about selecting oscillators for
spacecraft. I can dig up the old links if you want. I think Jim Lux
posted about this recently also.

What impressed me was that they'd buy tens of oscillators, thoroughly
test them, pick the top few for even deeper testing, and then only one
would be selected for the mission. The eBay / time nut in me always
wonders: where on earth did all those "rejects" end up? Is there a
landfill somewhere with dozens of 9e-14 oscillators because they didn't
meet some 8e-14 spec?

/tvb

For USOs
They build a batch of physics packages, test em all, and cherry pick.  Then they finish the final assembly, test them and cherry pick.

Some of the "less excellent" wind up as a flight spare (in case a bolt of lightning hits the spacecraft while being assembled, or falls off a ladder, damaging it, or hooks up the wires wrong) - so it's fully flight qualified, and can be swapped in.  Those often go on to be used for a future mission (with an appropriate cost transfer, etc.) 
 (For example, the flight spare radio for MRO (I think) wound up being used on one of the MER rovers, leading to an interesting situation of two spacecraft at Mars being on the same channel - the radio requires a crystal tuned to the channel frequency, and with a 2 year lead time.)

The spare physics packages probably go on the shelf for some number of years, in hopes that they can be used on a future mission. 

The same sort of thing happens with Traveling Wave Tube Amplifiers - the yield is hard to predict so you might start 10 tubes, to make sure you can deliver 3 (prime, redundant, spare).  If you wind up with 4 or 5 good tubes, you hang on to them.   

Depending on the contract, who owns the spares/in process varies.  If it was a Firm Fixed Price - the mfr owns em, and "got lucky".  If it was Cost plus Fixed Fee, the government owns all the piece parts and work in progress.   It's super common at the end of a build, just after launch (when you know you won't need the spares) for the contractor to send you this HUGE list of all the spare parts they have and ask you for the disposition.  Do they put them in a big crate and ship them to you? (It's happened.. I've seen big boxes of carefully bagged and tagged resistors and transistors and parts), or do you negotiate a sale back to the contractor? Or, do you transfer them to some university (there's a special provision in some law for this - I transferred a bunch of equipment from a test that we didn't need any more in New Mexico to BYU this way.

This kind of thing is why APL came up with a design to integrate a DDS with a USO, because typically, you purchase a USO for a specific frequency - The USO on MRO is, I think, 24 MHz.  The ones for GRAIL (a pair measuring gravity at the Moon) were at two different frequencies.  It's a pain to customize, so the ability to use a standard 10 MHz or 100 MHz or whatever APL finds best is great.

Similarly, all the new spacecraft radios use digital synthesis chains in FPGA, so they can all use the same frequency reference oscillator.  The older transponders required the oscillator to be tied to the spacecraft channel (e.g. if you're on Channel 14 8415.000000 MHz downlink, you'd want the oscillator to evenly multiply up to that.  Typically, some multiple of 9.5625 MHz, so the crystal might be cut for 76.5 MHz (8x) then multiplied up by 110 to the X band downlink. And another synthesis to make 7152.75, the LO for the receive of 7162.3125, so the IF is exactly at 9.5625.

Modern SDR techniques save us a lot. You can buy the same radio and tune it anywhere in the band today, even in flight.
 

On Thu, 23 Apr 2026 08:57:05 -0700, Tom Van Baak via time-nuts time-nuts@lists.febo.com wrote:

Bob, Sean,

For better results use more than one reference and apply
cross-correlation / 3-hat tricks. See 3 BVA here:

http://www.ke5fx.com/bva.htm

For reliable ADEV plots you want the reference to be several times
better (maybe 3x, or even 10x) than the oscillator you're trying to
measure. So at tau 60 s, 1e-13 is a challenge with a PHM, but AHM will
be fine. See 3 masers here:

http://leapsecond.com/museum/mhm-2010/

It sounds like your project is at the bleeding edge. IIRC, there are
informative JHU or JPL papers about selecting oscillators for
spacecraft. I can dig up the old links if you want. I think Jim Lux
posted about this recently also.

What impressed me was that they'd buy tens of oscillators, thoroughly
test them, pick the top few for even deeper testing, and then only one
would be selected for the mission. The eBay / time nut in me always
wonders: where on earth did all those "rejects" end up? Is there a
landfill somewhere with dozens of 9e-14 oscillators because they didn't
meet some 8e-14 spec?

/tvb

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

On Tue, 17 Mar 2026 12:46:35 -0600
Sean McAllister via time-nuts time-nuts@lists.febo.com wrote:

If I had no money constraints?

• Femto-Tech ULISS Gen2 for short term
• T4 Science iMaser 3000 for medium term
• Shimadzu Aetherclock OC020 for long term
• SpectraDynamics LNFS-1G for frequency synthesis
• Timeprocessor by INRIM/Claudio Calosso to tie all together

Unfortunately, the above costs more than than just a nice car.
But a time-nut can dream.....

To make this discusion a bit more informed, I attached an ADEV plot
of the performance (Bob's comment on applicability applies)
(sorry, I have not gotten around to add CSO data yet... they
are approximately two orders of magnitude better than the HSO14,
ie are below 1e-15 from 0.01s to at least 10'000s usually to 100'000s)

For further discussion, as you did not write where you are located,
I will assume you are from the US. If not, please let us know, so
I can adjust the recommendations.

So, for your application, a HSO14 Option 08 would do the job...
kind of.

The problem, as Bob mentioned, is that these only get to that performance
if they have been running continuously. Yes, they should get to those
specs in a month. But to be sure, you should run them for a year or
two to be on the safe side. Also: do not move them, touch them, heat
them or anything during that time...Or the counter starts again.

If you need just the ultimate shor-term performance, go to Femto-Tech
and get an ULISS CSO. These things are great and give you better
performance below 1000s than anything else on the market... unless you
go experimental. Also, you have a direct 10GHz output, which makes going
to K band very easy and low noise.

You can also get an optical cavity stabilized laser and down convert
it to RF by using a frequency comb. This option is probably cheaper
than buying a CSO (especially considering the political situation).
Stable Laser Systems from Boulder, CO would be the most suitable
contact for a company in the US. If you use a high repetition
rate frequency comb, like the ones from Menhir (which are probably
the most reliable on the market), then you can use them directly
to generate GHz signals.

If neither of that works, go for an active hydrogen maser.
I would recommend the iMaser 3000 from T4 Science/Safran.
This would also give you the added benefit having a good
long term performance that reaches almost 1e-16 around 10'000s
(note, the Microchip MHM perform as well short term, but not
as well long term).

Before you build a HSO14 - Caesium beam standard combination,
be aware that Oscillquartz sells that as a package already.
It's called "Enhanced Short-Term Unit" (ESTU) for the OSA3300.

As you have already ordered an HSO14, I would get an
cRb from SpectraDynamics. They deliver passive hydrogen maser
stability without the drift (ADEV stays <3e-15 indefinitely,
which is better than the 5071 high-performance).
This combo would give you decent short and long term performance
for a reasonable price.

Oh... and for all this, you will need a dedicated, specially
air conditioned room. And by air conditioned I mean a continuously
running air conditioning, not a cycling one that are usually sold.
Otherwise your ADEV will just go out of the window.

HTH

		Attila Kinali

--
Science is made up of so many things that appear obvious
after they are explained. -- Pardot Kynes

Hi

Off we go and pick up all the nice stuff on Attila’s list. We also get all of the various bits and pieces
needed to get it hooked up and verified as working.

On the “air conditioning” side of things:

The target you are after is +/- 0.1 C at any point in the lab. Yes, you could go for something better.
This is what most of the gear is specified at. Note: the lab can have a gradient of > 0.1C, it just has
to be a stable gradient.

Walk around the lab and you heat things up. Even with a “perfect” system your body heat will make
+/- 0.1 C a tough target to hit That’s even if the A/C system is ideal. Window with sunlight coming
through them …not so much. Uninsulated walls to adjacent spaces that are not controlled …. not
so much.

Simple answer for the infinite budget folks: Dedicated “walk in” temperature control chambers for
each of the various pieces of gear. Nobody goes in unless it’s scheduled maintenance. Yes this
does have an impact on any piece of gear that needs manual adjustment. Shop accordingly. Thinking
about putting multiple items in a chamber? That’s a single point of failure …. so not how it’s done.

If power goes out, you are right back to the “maybe a month” sort of stabilization time. Having
dedicated power into the lab from two sub-stations at the power company is a good idea. Having
both power sources backed up with generators and batteries (generators take a while to get going.
Even a brief outage is a problem) is also a good idea.

As some have found out recently, 5 days of fuel on site for those generators is a “skimpy” allowance.
Two to three weeks is a better target. Also, make sure to test the generators and the switchover
at least once a month. Rotate the “time of day” that’s done to make sure the folks on all shifts are
familiar with what to do. I have a log of  empirical data on the need to do that …. Power outage at
3 AM on a Sunday? It does happen … Generator needs this or that “poked” to get it running? That’s
part of the fun.

Sounds like we’ve got it all set up. Not quite yet.

In order to be sure that each of your fancy standards are working, you need to compare them to
something that is “just as good”. The net result is that you need three of each of your fancy devices
as well as the gear to continuously monitor how they are doing.

At this point, the budget is well past what all but a few national governments can afford. We haven’t
even started talking about staff and maintenance. We also have not dug into replacement / repair  cost
when this or that wears out in 6 years (high performance 5071) or maybe 10 years (various other items).
When this or that is “gone” for a few months, do you have a backup on site? Now you need 4 of each
of your fancy devices.

Now, you have a set of reference standards. You are confident of their performance. You still need to
set up for whatever you actually want to do. All we have at this point is plumbing. The real gear is
still to be discussed. Just how this all interfaces to that gear will impact things. Since we’re talking
a giant pile of cash, those impacts are not going to be a minor thing.

Fun

Bob

Excellent suggestions thank you, I hadn't heard of most of those. I'm
actually in the Denver area so I could pay a visit to Stable Laser Systems
in person if needed.

On Fri, Apr 24, 2026 at 10:27 PM Attila Kinali via time-nuts <
time-nuts@lists.febo.com> wrote:

Moin

On Sat, 25 Apr 2026 09:12:21 -0400
Bob kb8tq via time-nuts time-nuts@lists.febo.com wrote:

That's slightly overkill. ±1°C is plenty. That's what most NMI keep
their labs at. Some do ±0.5°C, though. But for an industrial metrology
lab? ±2°C would probably also be fine.

A walk-in is recommended for high-precision applications, but not
required. Also, keep in mind we are talking about an industrial
application. These are used as frequency standards, not as time keepers,
which means short excursions don't matter, as long as they return
to normal soon after.

[...]

Ok, this is way overkill! Unless you run an NMI, you don't need
that much. Just need to protect against what is likely to happen.
If you live in a civilized country and only have small, a few seconds
to a few minutes outages every 5 to 10 years, then that's what you
protect against. And for that, you only need a decent UPS. Maybe, put
the HSO-14 on a dedicated UPS so it can continue to run even when
power is out for a few hours or days, but that's all. The atomic clocks
don't care that much if they are powered off. At least not when we
are talking about industrial applications. You'll have a day or so
with degraded performance, but that's something one can deal with
easily. Especially when the rest of the facilities lost power too
and are still recovering anyways.

Things change a bit when you have to keep time, e.g. for legal
reasons. Then you need to ensure that all clocks and other equipment
used for the timescale are continuously powered. But even then, you
don't need to protect against a multi-day outage that happens once
a century, which would take out the consumers of time as well.
Once you get power back, you have plenty of time to rebuild the whole
time tracability chain and get your accurate time back during the time
it takes the rest of the company to reboot.

So no, a dedicated, properly air conditioned room with a decent UPS
is all you need. No need to go overboard.

		Attila Kinali

--
The driving force behind research is the question: "Why?"
There are things we don't understand and things we always
wonder about. And that's why we do research.
-- Kobayashi Makoto

Hi Attila,

What a fun list!  What do some of these instruments cost?  How
reliable are they?

The cold sapphire oscillators are a very promising technology. I
haven't seen a ULISS but its specs look incredible.  How much does it
cost?  Also, continuing this thought experiment...If one is
developing a lab to have "top-notch time/frequency stability", the
investment in cryogenic pumps can be spread over a few instruments.
So, from ambient air in the room that contains the instruments to, in
principle, the cryogenic pumps, a lab can spread the cost to get down
to even liquid helium temps across a few instruments (get one or two
bigger pump systems, i.e., get more cooling capacity for various
instruments.  This could also potentially increase reliability.)
Also, the CSO does seem to pair quite well with the cRb clock from
Spectradynamics based on your ADEV chart. Is that also a reliable
instrument?

As a separate question...An optical frequency comb based oscillator
seems to promise a similar performance for less money.  Who here has
experience operating a commercial optical comb system for
time/frequency purposes?  (As opposed to strictly optical comb
research.)  Are they reliable?  What kind of accuracy and stability do
you get compared with the specs?

Alex

On Sat, Apr 25, 2026 at 12:27 AM Attila Kinali via time-nuts
time-nuts@lists.febo.com wrote: