On 8/14/2022 10:08 AM, Corby Dawson via time-nuts wrote:

Any crystal can operate with any load capacitance, or in series
mode with "infinite" load capacitance.  This is a function of
the oscillator circuit, not the crystal.

OTOH, the frequency calibration is specified for some load
capacitance.  If you want to operate on that frequency, you
have to use the load capacitance specified for the
frequency calibration.

If you have an existing oscillator that you want a crystal
for, then the crystal has to be specified for the load
capacitance of that oscillator.

Having said that, overtone crystals are usually specified
at series resonance, especially at 5th and higher overtones.

The original text that you quoted is simply nonsense,
as you suspected, so something got messed up along the way.

Rick N6RK

All crystals are series resonant.  Some are merely cut for resonance
with an external capacitor such as 30 pFd.  The crystals in my designs
operate solely at series resonance and an external capacitor is not
specified. Q is a function of the care in manufacturing and affects
pullibility. Angle of cut determines the upper turning point
temperature.  Standard AT cuts have a 65° turning point.
That turning point is also affected by the overtone of operation and
is different for fundamental and 3rd overtone.  5th and higher
overtones have nearly the same turnover temperature compared to 3rd
overtone.  See https://gonascent.com/papers/crystals/xtlcurve.jpg

On Sun, Aug 14, 2022 at 10:27 AM Corby Dawson via time-nuts
time-nuts@lists.febo.com wrote:

Hi

If this is a more or less “from scratch” build:

1. You can get perfectly good 3rd overtone crystals at 60 MHz.
   They will be easier to work with than a 5th. You will get more pull
   range and the phase noise will be as good / better.

2. As Rick mentions, you want a series resonant ( no load cap )
   crystal for this sort of thing. The same “tune range” stuff gets into
   this part of it as well.

3. If you are really headed to 10’s of GHz, you might want to think
   through the entire chain. In some cases a 120 MHz crystal is a
   better bet than 60. Your far removed phase noise can / will be better
   with 2X ( or 3X or maybe even 4X) the “starting” frequency.

4. I would not want to do this sort of thing without a pretty good
   PCB layout being part of it. There are just to many ways for VHF
   oscillators to get a bit nutty ….

5. Typically aging is not a “big deal” in a multiplier chain sort of setup.
   Everything gets phase locked all the time. You simply need enough
   tune range to keep it running that way. I would not spend a lot of
   time “worrying” about aging.

6. Oddly enough crystals do impact phase noise at a 100Hz offset
   at 120 MHz. For a tidy sum, Croven can sell you some pretty good
   crystals. Not clear if 100 Hz matters or not though ….

7. At the typical ADEV sort of measurement Tau’s everything should
   be dependent on your PLL design rather than the OCVCXO’s. Yes,
   there can be crazy problems. If the phase noise is “good” then
   ADEV should be into the PLL arena.

No this isn’t easy to guess at. There are way to many variables in
terms of what one might want to do.

Of course, there’s also the “off the shelf” alternative of just going
with one of the many low noise VCXO’s out there:

https://www.crystek.com/crystal/spec-sheets/vcxo/CVSS-945.pdf https://www.crystek.com/crystal/spec-sheets/vcxo/CVSS-945.pdf

The unfortunate truth is that Digikey will sell you a 60 MHz example
for ~$20 ( if they had them in stock ….. ). Getting a “build to spec”
crystal for that price, not so much.

Bob

Hi

Crystals (of any sort) have whatever turnover the manufacturer decided
they wanted them to have. There is no “standard” target for any crystal
cut. It needs to be specified when the crystal is ordered.  That is true
of heated crystals as well as ones that go into un-heated applications.
Unless you have a spec, there is no way to “guess” what this or that
random crystal will have as a turn point.

Q at a specific frequency is a function of resistance and motional capacitance.
Pull ability is a a function of motional capacitance, but not resistance. You
can have a crystal with high Q and reasonable pull ability. It will have a
low resistance.

Indeed C0 also gets into the pull ability stuff to a degree. To what degree
depends more than a bit on the circuit used. You can ( and folks very much
do ) “resonate out” the C0 to get rid of the effect. The impact of C0 is
what makes the pull ability appear to change with finish point.

Bob

On 8/14/2022 11:47 AM, Bob kb8tq via time-nuts wrote:

A few years ago, I designed a ~6.84... MHz source
(you can guess what it was used for).  It used
an NEL 10 MHz VCXO multiplied to VHF and used
to phase lock an Abracon VCXO.  A microwave DRO
was offset by a synthesizer to make the output
frequency.  The DRO was divided down and
compared to the Abracon VCXO to phase lock it.
I looked at a lot of possible architectures and
this was the best one we could come up with.
It was not actually all that expensive except
for the NEC oscillator.

Rick N6Rk

On 8/14/2022 1:08 PM, Bob kb8tq via time-nuts wrote:

And indeed only about half of the 10811 crystals even exhibit a turnover
of any kind.  Instead, these crystals merely have a temperature range
with very low tempco.  For all of those crystals, they would set the
oven to 82 degrees.  Because of the double rotation, there isn't
actually any very strict definition of an "SC cut".  They can't
even agree on whether "SC" stands for "stress compensated" or
"Santa Clara"  :-)

The E1938 crystals were just repackaged 10811 crystals, but the cut
had to be adjusted to always have a turnover.  We actually used the
upper turnover because the E1938 had to operate at ambient up 85
degrees.  BTW, the E1938 crystals had to be calibrated for 10.000000 MHz
at series resonance, unlike the 10811 crystals which were calibrated
for 20 pF (IIRC).

Rick N6RK