On Sat, 02 Apr 2022 03:27:06 -0400, time-nuts-request@lists.febo.com
wrote:
time-nuts Digest, Vol 216, Issue 3

I know of Wenzel, but they don't make bench-top lab instruments.
Rack-mount is available, but as a custom part.

I did find NEL, and they do look very good.  The NEL 2030A and 2030B,
which are rack-mount, are plausible.

Both are likely expensive, with long lead times.

Yes.  I assume that it takes 6 to 12 months of screening the run of
production to get these few golden units.  Or they make a special
circuit version, but still screening is required, with low yield.

And so yes there is a tradeoff to be made.

Yes, voltage control is also needed, so a Rubidium can be used to
stabilize the OCXO.  Or the OCXO already has that built in.

What is also needed is a very quiet source of DC power for the Rb,
OCXO, et al.  None of the vendors seem to specify their PSRR (power
supply rejection ratio), and I assume that all published curves are
obtained using a battery-powered unit under test.

Mutually-exclusive, yes.  Impossible, no - Wenzel OCXOs locked to a
Rb already do this.

Yes, an OCXO is the only plausible answer.

Would be figured out using the measured performance of the Rb and
OCXO.  The exact crossover frequency isn't all that critical, so long
as it's reasonably close.  Wenzel OCXOs use 2nd order PLLs with a
loop bandwidth around one Hertz.

Buy and select is out.  I'm trying to buy a self-contained instrument
intended to live in the rough-and-tumble of a lab bench.

Yep.

Joe Gwinn

On 4/2/22 4:47 PM, Joseph Gwinn wrote:

Wenzel will be happy (for a price) to put it in any size or shape box
you want. Most people want rack mounts, but it's mostly a matter of
sheet metal work and they can certainly do that.

My experience with Wenzel is that the cost isn't huge - the dominant
thing will be the stuff that goes in the box (i.e. the oscillator). 
Essentially, you're paying for a few week's engineering and tech time to
lay out and fab the thing. At 5-10k/work week, yes, you're looking at
$10-20k for "the box and assembly" for a one-off.

I wouldn't have them put a $200 streamline OCXO in a box <grin>

If you're comfortable assembling the parts a linear supply from Acopian,
a box, using frontpanelexpress.com to do the machining of the front and
back panels, and the baseplate, and your oscillator. (that's what we've
done at JPL on multiple occasions - although considering technician time
it's probably cheaper to have Wenzel do it)

Wenzel also has a sort of intermediate level where you get a plate with
a bunch of parts on it to do some function, and you put that in a box.

Depending on your frequency and performance, you might find a "in
stock"  oscillator.  A 10 MHz Onyx or something like that might be
basically stock.

Most of them have an electronic tuning input. (which can be left out as
a special order).

My $200 (10 years ago) speedlines have a tuning input. Of course, it's
not hitting your -170 dBc requirement, but the ULN and other ones do (or
come close).

No, that's a "call them and ask", but most have pretty good PSRR
(although what are you looking for?) - the guy or gal on the phone will
tell you what it is in a "not guaranteed on the data sheet" sort of way.
Unless you want them to hit a spec, but you'll pay for it.

https://wenzel.com/model/btuln/ mentions that they have an internal low
noise regulator - it's about 5dB shy of your -170 at 100Hz requirement.

I wouldn't assume battery power - but this is where a phone call helps -
they'll be happy to tell you.

Hi,

This is nearly "off-the shelf" model:
https://www.axtal.com/cms/docs/doc100916.pdf
The ULN version does meet the -170 dBc/Hz

Bernd

-----Ursprüngliche Nachricht-----
Von: Lux, Jim [mailto:jim@luxfamily.com]
Gesendet: Sonntag, 3. April 2022 03:14
An: time-nuts@lists.febo.com
Betreff: [time-nuts] Re: Low Phase Noise 10 MHz bench signal source sought

On 4/2/22 4:47 PM, Joseph Gwinn wrote:

Wenzel will be happy (for a price) to put it in any size or shape box you want. Most people want rack mounts, but it's mostly a matter of sheet metal work and they can certainly do that.

My experience with Wenzel is that the cost isn't huge - the dominant thing will be the stuff that goes in the box (i.e. the oscillator). Essentially, you're paying for a few week's engineering and tech time to lay out and fab the thing. At 5-10k/work week, yes, you're looking at $10-20k for "the box and assembly" for a one-off.

I wouldn't have them put a $200 streamline OCXO in a box <grin>

If you're comfortable assembling the parts a linear supply from Acopian, a box, using frontpanelexpress.com to do the machining of the front and back panels, and the baseplate, and your oscillator. (that's what we've done at JPL on multiple occasions - although considering technician time it's probably cheaper to have Wenzel do it)

Wenzel also has a sort of intermediate level where you get a plate with a bunch of parts on it to do some function, and you put that in a box.

Depending on your frequency and performance, you might find a "in
stock"  oscillator.  A 10 MHz Onyx or something like that might be
basically stock.

Most of them have an electronic tuning input. (which can be left out as
a special order).

My $200 (10 years ago) speedlines have a tuning input. Of course, it's
not hitting your -170 dBc requirement, but the ULN and other ones do (or
come close).

No, that's a "call them and ask", but most have pretty good PSRR
(although what are you looking for?) - the guy or gal on the phone will
tell you what it is in a "not guaranteed on the data sheet" sort of way.
Unless you want them to hit a spec, but you'll pay for it.

https://wenzel.com/model/btuln/ mentions that they have an internal low
noise regulator - it's about 5dB shy of your -170 at 100Hz requirement.

I wouldn't assume battery power - but this is where a phone call helps -
they'll be happy to tell you.

Hi

These days a PLL is going to either be analog or digital. If it’s
analog, you get into size constraints related to capacitors
as you go to lower crossover frequencies. With digital, you
get into all of the noise issues that any digital circuit will have.
(Yes, they can be addressed but it’s not easy at very low
offset frequencies).

Regardless of design, you will always have some noise peaking.

in other problems. Setting the phase margin and other parameters
is either a mathematical design process or done with simulation.
It can be very frustrating doing it by trial and error. I still find
Phaselock Techniques by Floyd M Gardner to be a good reference
on this stuff.

The normal PLL control loops are fairly low order filters. If you
go high order, loop stability (and peaking) becomes difficult
to handle. Because it’s a simple filter, things like close in spurs
will only be attenuated by some finite amount.

So yes, you will have issues. Dealing with those issues means
an area ( = range of offset frequencies) that are not as nice as you
might wish. That’s just the way the real world works.

Bob