JD
J Decker
Mon, Feb 16, 2026 1:37 PM
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment
to set the internal PWM; the GPS sync is close but it can have many
nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength
)... but anyway, these two devices I was wondering if they were sensitive
enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I
fiddled with them for quite a while then realized that mainly the
difference was, when I picked it up, it was no long exactly as level as it
was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the
front up and the back down they would run faster, and the back up and front
down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the
crystal and change the oscillation rate. (A new technique for
overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning,
and got those set close to the first set, and tested orientation; one was
just as bad as the first pair, the other barely experienced any change at
all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain
orientation, then placed them down in a slightly different orientation and
have an immediate drift.
a lot of specs don't mention the drift for accelerations in various
directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts <
time-nuts@lists.febo.com> wrote:
Hello Bob,
It’s difficult to find detailed documentation on the manufacturing process
because each manufacturer keeps its methods proprietary, and the techniques
have evolved significantly over the years.
For high-end crystal oscillators, there are multiple steps—from blank
cutting, adjustment, cleaning, to encapsulation—to ensure optimal
performance.
During lapping, the parts are not only measured but also optically
inspected to verify whether they are perfectly planar or have the required
convex profile.
After the electrodes are added, there are two main approaches for
frequency adjustment:
- Depositing material to fine-tune the frequency
- Plasma removal to trim the frequency
Both techniques have their advantages and disadvantages.
As you mentioned, the base, crystal, and cap are cleaned using specialized
detergents and handled in a cleanroom environment. Several high-temperature
cycles are then applied to stabilize atomic migration in the plating
material, which improves long-term aging performance.
The base and cap are handled with dedicated tools, and cold welding is
carried out by applying high pressure under vacuum in a controlled
low-temperature environment.
Another important consideration is avoiding any outgassing materials,
including the adhesive used to mount the crystal on the base.
There is actually a lot to say about the process and the many steps
involved in building a high-quality crystal. On our side alone, we have
more than 20 process steps, which gives you an idea of why the crystal ends
up being one of the main cost drivers in an OCXO design.
Best regards,
Jean-Charles
Best regards,
-----Message d'origine-----
De : Bob Bownes via time-nuts time-nuts@lists.febo.com
Envoyé : mardi 10 février 2026 16:48
À : mcleannb--- via time-nuts time-nuts@lists.febo.com
Cc : Bob Bownes bownes@gmail.com
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Should anyone want any, I think I still have a whole bunch of quartz slabs
from a military crystal/SAW mfr that I'm willing to share. No plating, some
may be optical beam splitters, but most are quartz blanks of a large
variety of sizes. I might have tossed them in the most recent Great Purge,
but if anyone is interested, lmk, and I'll check next time I'm in the
garage.
Bob
On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
Hi all,
I remember visiting a crystal manufacturing shop in the town where I
grew up as a teenager in the 1970's. The guy would make the crystals to
frequency or ever so slightly high. The ones that were slightly high he
would draw a tiny line of lead pencil to "pull" the frequency low.
Thank you
Nic McLean
President & Squad Captain
VRA Rescue NSW - WICEN
6 Judy Jakins Drive, Dubbo, NSW 2830
1300 872 777 or 0417 822 728
wicen.president@vrarescue.org
-----Original Message-----
From: Bob kb8tq via time-nutstime-nuts@lists.febo.com
Sent: Tuesday, 10 February 2026 9:53 AM
To: Discussion of precise time and frequency
measurementtime-nuts@lists.febo.com
Cc: Bob kb8tqkb8tq@n1k.org
Subject: [time-nuts] Re: Question on quartz crystal manufacturing
Hi
The problem is that you can’t measure the frequency of a bare blank very
well. You need those plated electrodes and the ultra clean package to get
the “other stuff” out of the measurement.
In order to see what’s going on, you need to turn it into a finished
crystal. This is what’s done on things like the USO’s. They build lots of
finished crystals and sort them out. Each one goes into a “dummy”
OCXO and is tested for various parameters. The ones that meet all the
requirements move on into the finished product.
On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
The "make some blanks and run them in an oscillator for a while, before
finalizing the assembly" is what APL does for Ultra Stable Oscillators for
space flight.
I can't recall if they actually assemble the physics package (the part
that goes in a dewar) and run them before choosing the ones to fully
integrate. It would make sense, because one thing you want to deal with the
the residual stresses of the crystal in the holder, etc. And to identify
those that were "on frequency and didn't drift too far" and "didn't have
oddities in the aging process". They might have put them in the dewar
before the long duration burn in. I think there is some pretest too, I mean
you have to make sure that you've ground/finished/plated them to hit the
desired frequency.
As I recall for one batch of USOs, there were several dozen fabricated
and aged, of which 2 got finally used. From roughly 1000 blanks.
At $1M a copy, that fairly fits Bob's high BoM cost description.
On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
Hi
Well, never say never :) :) :).
It’s certainly not been part of any process I have been involved with.
However it may well have been done by others. I am not aware of it
being done by anybody these days. Again that could just be my not
having talked to the right folks. It also may (as you suggest) be
something done “long long ago”.
One important note not mentioned earlier:
That assumes that “precision crystal” = stuff like OCXO crystals.
Even that term may have multiple meanings to various folks. A 3rd or
5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass
package is what comes quickly to mind.
What is indeed done when processing these crystals:
The crystal growth process for synthetic quartz gets more involved
when precision parts are involved. You go from a fairly quick growing
approach to one that takes much longer.
That does involve a range of steps past just the basic growth process.
Sealed crystals after fabrication are normally run through bake and
The objective is to reduce the “front end aging” (or whatever you
want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
Occasionally processes do include running a crystal in a circuit
prior to fabricating the final oscillator. The most common example is
with low volume OCXO’s that have a very high BOM. Sorting crystals
for aging makes sense if > 90% of the crystals will be tossed out. For
the vast majority of OCXO’s made, this is not needed / not useful.
On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
Bob kb8tq via time-nuts writes:
- How is the cap placed on top of the holder under vacuum?
There is a step before this where the blank is finish plated to put
The manipulation process is a “that depends” sort of thing. It
could simply be “folks” using their hands via big black “vacuum
gloves”. It could be something more automated.
Somewhere, possibly BSTJ, I read about quartz for precision crystals
being 'stress-tested' and 'annealed' a process which involved forced
oscillation while gradually lowering temperature and
voltage/(amplitude?)
Was that (only) a thing with natural quartz ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment
to set the internal PWM; the GPS sync is close but it can have many
nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength
)... but anyway, these two devices I was wondering if they were sensitive
enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I
fiddled with them for quite a while then realized that mainly the
difference was, when I picked it up, it was no long exactly as level as it
was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the
front up and the back down they would run faster, and the back up and front
down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the
crystal and change the oscillation rate. (A new technique for
overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning,
and got those set close to the first set, and tested orientation; one was
just as bad as the first pair, the other barely experienced any change at
all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain
orientation, then placed them down in a slightly different orientation and
have an immediate drift.
a lot of specs don't mention the drift for accelerations in various
directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts <
time-nuts@lists.febo.com> wrote:
> Hello Bob,
>
> It’s difficult to find detailed documentation on the manufacturing process
> because each manufacturer keeps its methods proprietary, and the techniques
> have evolved significantly over the years.
>
> For high-end crystal oscillators, there are multiple steps—from blank
> cutting, adjustment, cleaning, to encapsulation—to ensure optimal
> performance.
>
> During lapping, the parts are not only measured but also optically
> inspected to verify whether they are perfectly planar or have the required
> convex profile.
>
> After the electrodes are added, there are two main approaches for
> frequency adjustment:
>
> - Depositing material to fine-tune the frequency
> - Plasma removal to trim the frequency
>
> Both techniques have their advantages and disadvantages.
>
> As you mentioned, the base, crystal, and cap are cleaned using specialized
> detergents and handled in a cleanroom environment. Several high-temperature
> cycles are then applied to stabilize atomic migration in the plating
> material, which improves long-term aging performance.
>
> The base and cap are handled with dedicated tools, and cold welding is
> carried out by applying high pressure under vacuum in a controlled
> low-temperature environment.
>
> Another important consideration is avoiding any outgassing materials,
> including the adhesive used to mount the crystal on the base.
>
> There is actually a lot to say about the process and the many steps
> involved in building a high-quality crystal. On our side alone, we have
> more than 20 process steps, which gives you an idea of why the crystal ends
> up being one of the main cost drivers in an OCXO design.
>
> Best regards,
>
> Jean-Charles
>
> Best regards,
> -----Message d'origine-----
> De : Bob Bownes via time-nuts <time-nuts@lists.febo.com>
> Envoyé : mardi 10 février 2026 16:48
> À : mcleannb--- via time-nuts <time-nuts@lists.febo.com>
> Cc : Bob Bownes <bownes@gmail.com>
> Objet : [time-nuts] Re: Question on quartz crystal manufacturing
>
>
> Should anyone want any, I think I still have a whole bunch of quartz slabs
> from a military crystal/SAW mfr that I'm willing to share. No plating, some
> may be optical beam splitters, but most are quartz blanks of a large
> variety of sizes. I might have tossed them in the most recent Great Purge,
> but if anyone is interested, lmk, and I'll check next time I'm in the
> garage.
>
> Bob
>
>
> On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
> > Hi all,
> > I remember visiting a crystal manufacturing shop in the town where I
> grew up as a teenager in the 1970's. The guy would make the crystals to
> frequency or ever so slightly high. The ones that were slightly high he
> would draw a tiny line of lead pencil to "pull" the frequency low.
> >
> > Thank you
> >
> > Nic McLean
> > President & Squad Captain
> > VRA Rescue NSW - WICEN
> > 6 Judy Jakins Drive, Dubbo, NSW 2830
> > 1300 872 777 or 0417 822 728
> > wicen.president@vrarescue.org
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: Bob kb8tq via time-nuts<time-nuts@lists.febo.com>
> > Sent: Tuesday, 10 February 2026 9:53 AM
> > To: Discussion of precise time and frequency
> > measurement<time-nuts@lists.febo.com>
> > Cc: Bob kb8tq<kb8tq@n1k.org>
> > Subject: [time-nuts] Re: Question on quartz crystal manufacturing
> >
> > Hi
> >
> > The problem is that you can’t measure the frequency of a bare blank very
> well. You need those plated electrodes and the ultra clean package to get
> the “other stuff” out of the measurement.
> >
> > In order to see what’s going on, you need to turn it into a finished
> crystal. This is what’s done on things like the USO’s. They build *lots* of
> finished crystals and sort them out. Each one goes into a “dummy”
> > OCXO and is tested for various parameters. The ones that meet all the
> requirements move on into the finished product.
> >
> > Bob
> >
> >> On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
> time-nuts@lists.febo.com> wrote:
> >>
> >>
> >>
> >>
> >>
> >>
> >> The "make some blanks and run them in an oscillator for a while, before
> finalizing the assembly" is what APL does for Ultra Stable Oscillators for
> space flight.
> >> I can't recall if they actually assemble the physics package (the part
> that goes in a dewar) and run them before choosing the ones to fully
> integrate. It would make sense, because one thing you want to deal with the
> the residual stresses of the crystal in the holder, etc. And to identify
> those that were "on frequency and didn't drift too far" and "didn't have
> oddities in the aging process". They might have put them in the dewar
> before the long duration burn in. I think there is some pretest too, I mean
> you have to make sure that you've ground/finished/plated them to hit the
> desired frequency.
> >>
> >> As I recall for one batch of USOs, there were several dozen fabricated
> and aged, of which 2 got finally used. From roughly 1000 blanks.
> >>
> >> At $1M a copy, that fairly fits Bob's high BoM cost description.
> >>
> >>
> >>
> >> On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
> time-nuts@lists.febo.com> wrote:
> >>
> >> Hi
> >>
> >> Well, never say never :) :) :).
> >>
> >> It’s certainly not been part of any process I have been involved with.
> >> However it may well have been done by others. I am not aware of it
> >> being done by anybody these days. Again that could just be my not
> >> having talked to the right folks. It also may (as you suggest) be
> something done “long long ago”.
> >>
> >> One important note not mentioned earlier:
> >>
> >> That assumes that “precision crystal” = stuff like OCXO crystals.
> >> Even that term may have multiple meanings to various folks. A 3rd or
> >> 5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass
> package is what comes quickly to mind.
> >>
> >> What is indeed done when processing these crystals:
> >>
> >> The crystal growth process for synthetic quartz gets more involved
> >> when precision parts are involved. You go from a fairly quick growing
> approach to one that takes much longer.
> >> That does involve a range of steps past just the basic growth process.
> >>
> >> Sealed crystals after fabrication are normally run through bake and
> thermo-cycle processes.
> >> The objective is to reduce the “front end aging” (or whatever you
> >> want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
> weeks saves money.
> >>
> >> Occasionally processes do include running a crystal in a circuit
> >> prior to fabricating the final oscillator. The most common example is
> >> with low volume OCXO’s that have a very high BOM. Sorting crystals
> >> for aging makes sense if > 90% of the crystals will be tossed out. For
> the vast majority of OCXO’s made, this is not needed / not useful.
> >>
> >> Fun
> >>
> >> Bob
> >>
> >>> On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
> >>>
> >>> --------
> >>> Bob kb8tq via time-nuts writes:
> >>>
> >>>>> 3) How is the cap placed on top of the holder under vacuum?
> >>>> There is a step before this where the blank is finish plated to put
> it on frequency.
> >>>>
> >>>> The manipulation process is a “that depends” sort of thing. It
> >>>> could simply be “folks” using their hands via big black “vacuum
> gloves”. It could be something more automated.
> >>> Somewhere, possibly BSTJ, I read about quartz for precision crystals
> >>> being 'stress-tested' and 'annealed' a process which involved forced
> >>> oscillation while gradually lowering temperature and
> >>> voltage/(amplitude?)
> >>>
> >>> Was that (only) a thing with natural quartz ?
> >>>
> >>> --
> >>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
> >> _______________________________________________
> >> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> >> send an email totime-nuts-leave@lists.febo.com
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> >> send an email totime-nuts-leave@lists.febo.com
> > _______________________________________________
> > time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
> > an email totime-nuts-leave@lists.febo.com
> > _______________________________________________
> > time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
> > an email totime-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an
> email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
BK
Bob kb8tq
Mon, Feb 16, 2026 2:04 PM
Hi
Having worked for multiple companies that made precision crystals, no two of them “did it” quite the same way. Even different plants owned by the same people had different approaches. There is no “one way to do it”.
Bob
On Feb 16, 2026, at 5:04 AM, Jean-Charles BILLEBAULT via time-nuts time-nuts@lists.febo.com wrote:
Hello Bob,
It’s difficult to find detailed documentation on the manufacturing process because each manufacturer keeps its methods proprietary, and the techniques have evolved significantly over the years.
For high-end crystal oscillators, there are multiple steps—from blank cutting, adjustment, cleaning, to encapsulation—to ensure optimal performance.
During lapping, the parts are not only measured but also optically inspected to verify whether they are perfectly planar or have the required convex profile.
After the electrodes are added, there are two main approaches for frequency adjustment:
- Depositing material to fine-tune the frequency
- Plasma removal to trim the frequency
Both techniques have their advantages and disadvantages.
As you mentioned, the base, crystal, and cap are cleaned using specialized detergents and handled in a cleanroom environment. Several high-temperature cycles are then applied to stabilize atomic migration in the plating material, which improves long-term aging performance.
The base and cap are handled with dedicated tools, and cold welding is carried out by applying high pressure under vacuum in a controlled low-temperature environment.
Another important consideration is avoiding any outgassing materials, including the adhesive used to mount the crystal on the base.
There is actually a lot to say about the process and the many steps involved in building a high-quality crystal. On our side alone, we have more than 20 process steps, which gives you an idea of why the crystal ends up being one of the main cost drivers in an OCXO design.
Best regards,
Jean-Charles
Best regards,
-----Message d'origine-----
De : Bob Bownes via time-nuts time-nuts@lists.febo.com
Envoyé : mardi 10 février 2026 16:48
À : mcleannb--- via time-nuts time-nuts@lists.febo.com
Cc : Bob Bownes bownes@gmail.com
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Should anyone want any, I think I still have a whole bunch of quartz slabs from a military crystal/SAW mfr that I'm willing to share. No plating, some may be optical beam splitters, but most are quartz blanks of a large variety of sizes. I might have tossed them in the most recent Great Purge, but if anyone is interested, lmk, and I'll check next time I'm in the garage.
Bob
On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
Hi all,
I remember visiting a crystal manufacturing shop in the town where I grew up as a teenager in the 1970's. The guy would make the crystals to frequency or ever so slightly high. The ones that were slightly high he would draw a tiny line of lead pencil to "pull" the frequency low.
Thank you
Nic McLean
President & Squad Captain
VRA Rescue NSW - WICEN
6 Judy Jakins Drive, Dubbo, NSW 2830
1300 872 777 or 0417 822 728
wicen.president@vrarescue.org
-----Original Message-----
From: Bob kb8tq via time-nutstime-nuts@lists.febo.com
Sent: Tuesday, 10 February 2026 9:53 AM
To: Discussion of precise time and frequency
measurementtime-nuts@lists.febo.com
Cc: Bob kb8tqkb8tq@n1k.org
Subject: [time-nuts] Re: Question on quartz crystal manufacturing
Hi
The problem is that you can’t measure the frequency of a bare blank very well. You need those plated electrodes and the ultra clean package to get the “other stuff” out of the measurement.
In order to see what’s going on, you need to turn it into a finished crystal. This is what’s done on things like the USO’s. They build lots of finished crystals and sort them out. Each one goes into a “dummy”
OCXO and is tested for various parameters. The ones that meet all the requirements move on into the finished product.
Bob
On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nutstime-nuts@lists.febo.com wrote:
The "make some blanks and run them in an oscillator for a while, before finalizing the assembly" is what APL does for Ultra Stable Oscillators for space flight.
I can't recall if they actually assemble the physics package (the part that goes in a dewar) and run them before choosing the ones to fully integrate. It would make sense, because one thing you want to deal with the the residual stresses of the crystal in the holder, etc. And to identify those that were "on frequency and didn't drift too far" and "didn't have oddities in the aging process". They might have put them in the dewar before the long duration burn in. I think there is some pretest too, I mean you have to make sure that you've ground/finished/plated them to hit the desired frequency.
As I recall for one batch of USOs, there were several dozen fabricated and aged, of which 2 got finally used. From roughly 1000 blanks.
At $1M a copy, that fairly fits Bob's high BoM cost description.
On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nutstime-nuts@lists.febo.com wrote:
Hi
Well, never say never :) :) :).
It’s certainly not been part of any process I have been involved with.
However it may well have been done by others. I am not aware of it
being done by anybody these days. Again that could just be my not
having talked to the right folks. It also may (as you suggest) be something done “long long ago”.
One important note not mentioned earlier:
That assumes that “precision crystal” = stuff like OCXO crystals.
Even that term may have multiple meanings to various folks. A 3rd or
5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass package is what comes quickly to mind.
What is indeed done when processing these crystals:
The crystal growth process for synthetic quartz gets more involved
when precision parts are involved. You go from a fairly quick growing approach to one that takes much longer.
That does involve a range of steps past just the basic growth process.
Sealed crystals after fabrication are normally run through bake and thermo-cycle processes.
The objective is to reduce the “front end aging” (or whatever you
want to call it). Having a fabricated OCXO hit spec in 5 days vs 5 weeks saves money.
Occasionally processes do include running a crystal in a circuit
prior to fabricating the final oscillator. The most common example is
with low volume OCXO’s that have a very high BOM. Sorting crystals
for aging makes sense if > 90% of the crystals will be tossed out. For the vast majority of OCXO’s made, this is not needed / not useful.
Fun
Bob
On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
Bob kb8tq via time-nuts writes:
- How is the cap placed on top of the holder under vacuum?
There is a step before this where the blank is finish plated to put it on frequency.
The manipulation process is a “that depends” sort of thing. It
could simply be “folks” using their hands via big black “vacuum gloves”. It could be something more automated.
Somewhere, possibly BSTJ, I read about quartz for precision crystals
being 'stress-tested' and 'annealed' a process which involved forced
oscillation while gradually lowering temperature and
voltage/(amplitude?)
Was that (only) a thing with natural quartz ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
Hi
Having worked for multiple companies that made precision crystals, no two of them “did it” quite the same way. Even different plants owned by the same people had different approaches. There is no “one way to do it”.
Bob
> On Feb 16, 2026, at 5:04 AM, Jean-Charles BILLEBAULT via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Hello Bob,
>
> It’s difficult to find detailed documentation on the manufacturing process because each manufacturer keeps its methods proprietary, and the techniques have evolved significantly over the years.
>
> For high-end crystal oscillators, there are multiple steps—from blank cutting, adjustment, cleaning, to encapsulation—to ensure optimal performance.
>
> During lapping, the parts are not only measured but also optically inspected to verify whether they are perfectly planar or have the required convex profile.
>
> After the electrodes are added, there are two main approaches for frequency adjustment:
>
> - Depositing material to fine-tune the frequency
> - Plasma removal to trim the frequency
>
> Both techniques have their advantages and disadvantages.
>
> As you mentioned, the base, crystal, and cap are cleaned using specialized detergents and handled in a cleanroom environment. Several high-temperature cycles are then applied to stabilize atomic migration in the plating material, which improves long-term aging performance.
>
> The base and cap are handled with dedicated tools, and cold welding is carried out by applying high pressure under vacuum in a controlled low-temperature environment.
>
> Another important consideration is avoiding any outgassing materials, including the adhesive used to mount the crystal on the base.
>
> There is actually a lot to say about the process and the many steps involved in building a high-quality crystal. On our side alone, we have more than 20 process steps, which gives you an idea of why the crystal ends up being one of the main cost drivers in an OCXO design.
>
> Best regards,
>
> Jean-Charles
>
> Best regards,
> -----Message d'origine-----
> De : Bob Bownes via time-nuts <time-nuts@lists.febo.com>
> Envoyé : mardi 10 février 2026 16:48
> À : mcleannb--- via time-nuts <time-nuts@lists.febo.com>
> Cc : Bob Bownes <bownes@gmail.com>
> Objet : [time-nuts] Re: Question on quartz crystal manufacturing
>
>
> Should anyone want any, I think I still have a whole bunch of quartz slabs from a military crystal/SAW mfr that I'm willing to share. No plating, some may be optical beam splitters, but most are quartz blanks of a large variety of sizes. I might have tossed them in the most recent Great Purge, but if anyone is interested, lmk, and I'll check next time I'm in the garage.
>
> Bob
>
>
> On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
>> Hi all,
>> I remember visiting a crystal manufacturing shop in the town where I grew up as a teenager in the 1970's. The guy would make the crystals to frequency or ever so slightly high. The ones that were slightly high he would draw a tiny line of lead pencil to "pull" the frequency low.
>>
>> Thank you
>>
>> Nic McLean
>> President & Squad Captain
>> VRA Rescue NSW - WICEN
>> 6 Judy Jakins Drive, Dubbo, NSW 2830
>> 1300 872 777 or 0417 822 728
>> wicen.president@vrarescue.org
>>
>>
>>
>>
>>
>> -----Original Message-----
>> From: Bob kb8tq via time-nuts<time-nuts@lists.febo.com>
>> Sent: Tuesday, 10 February 2026 9:53 AM
>> To: Discussion of precise time and frequency
>> measurement<time-nuts@lists.febo.com>
>> Cc: Bob kb8tq<kb8tq@n1k.org>
>> Subject: [time-nuts] Re: Question on quartz crystal manufacturing
>>
>> Hi
>>
>> The problem is that you can’t measure the frequency of a bare blank very well. You need those plated electrodes and the ultra clean package to get the “other stuff” out of the measurement.
>>
>> In order to see what’s going on, you need to turn it into a finished crystal. This is what’s done on things like the USO’s. They build *lots* of finished crystals and sort them out. Each one goes into a “dummy”
>> OCXO and is tested for various parameters. The ones that meet all the requirements move on into the finished product.
>>
>> Bob
>>
>>> On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<time-nuts@lists.febo.com> wrote:
>>>
>>>
>>>
>>>
>>>
>>>
>>> The "make some blanks and run them in an oscillator for a while, before finalizing the assembly" is what APL does for Ultra Stable Oscillators for space flight.
>>> I can't recall if they actually assemble the physics package (the part that goes in a dewar) and run them before choosing the ones to fully integrate. It would make sense, because one thing you want to deal with the the residual stresses of the crystal in the holder, etc. And to identify those that were "on frequency and didn't drift too far" and "didn't have oddities in the aging process". They might have put them in the dewar before the long duration burn in. I think there is some pretest too, I mean you have to make sure that you've ground/finished/plated them to hit the desired frequency.
>>>
>>> As I recall for one batch of USOs, there were several dozen fabricated and aged, of which 2 got finally used. From roughly 1000 blanks.
>>>
>>> At $1M a copy, that fairly fits Bob's high BoM cost description.
>>>
>>>
>>>
>>> On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<time-nuts@lists.febo.com> wrote:
>>>
>>> Hi
>>>
>>> Well, never say never :) :) :).
>>>
>>> It’s certainly not been part of any process I have been involved with.
>>> However it may well have been done by others. I am not aware of it
>>> being done by anybody these days. Again that could just be my not
>>> having talked to the right folks. It also may (as you suggest) be something done “long long ago”.
>>>
>>> One important note not mentioned earlier:
>>>
>>> That assumes that “precision crystal” = stuff like OCXO crystals.
>>> Even that term may have multiple meanings to various folks. A 3rd or
>>> 5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass package is what comes quickly to mind.
>>>
>>> What is indeed done when processing these crystals:
>>>
>>> The crystal growth process for synthetic quartz gets more involved
>>> when precision parts are involved. You go from a fairly quick growing approach to one that takes much longer.
>>> That does involve a range of steps past just the basic growth process.
>>>
>>> Sealed crystals after fabrication are normally run through bake and thermo-cycle processes.
>>> The objective is to reduce the “front end aging” (or whatever you
>>> want to call it). Having a fabricated OCXO hit spec in 5 days vs 5 weeks saves money.
>>>
>>> Occasionally processes do include running a crystal in a circuit
>>> prior to fabricating the final oscillator. The most common example is
>>> with low volume OCXO’s that have a very high BOM. Sorting crystals
>>> for aging makes sense if > 90% of the crystals will be tossed out. For the vast majority of OCXO’s made, this is not needed / not useful.
>>>
>>> Fun
>>>
>>> Bob
>>>
>>>> On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
>>>>
>>>> --------
>>>> Bob kb8tq via time-nuts writes:
>>>>
>>>>>> 3) How is the cap placed on top of the holder under vacuum?
>>>>> There is a step before this where the blank is finish plated to put it on frequency.
>>>>>
>>>>> The manipulation process is a “that depends” sort of thing. It
>>>>> could simply be “folks” using their hands via big black “vacuum gloves”. It could be something more automated.
>>>> Somewhere, possibly BSTJ, I read about quartz for precision crystals
>>>> being 'stress-tested' and 'annealed' a process which involved forced
>>>> oscillation while gradually lowering temperature and
>>>> voltage/(amplitude?)
>>>>
>>>> Was that (only) a thing with natural quartz ?
>>>>
>>>> --
>>>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>> send an email totime-nuts-leave@lists.febo.com
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>> send an email totime-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
>> an email totime-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
>> an email totime-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
SH
Steve Hendrix, P.E.
Thu, Feb 19, 2026 11:13 AM
On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation.
Interesting experiments. I wonder if the earth's magnetic field could be
in play here. That is, does rotating them while keeping them "level"
change the game? Depending upon your latitude, there is a vertical
component to the magnetic field.
Steve Hendrix
On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
> Something I ran into about crystals, is the cut and package can affect the
> stability relative to gravitational orientation.
Interesting experiments. I wonder if the earth's magnetic field could be
in play here. That is, does rotating them while keeping them "level"
change the game? Depending upon your latitude, there is a vertical
component to the magnetic field.
Steve Hendrix
BK
Bob kb8tq
Thu, Feb 19, 2026 1:18 PM
Hi
OCXO’s have a “g” sensitivity in the 0.1 to 2 ppb / G range. Exotic ones might be well below
that. You are unlikely to run into one of those.
Simple test: Flip the unit over and see how many more ns / s it changes. One ppb is 1 ns / s.
That sounds easy. There are a few “gotcha’s”:
-
The sensitivity is always the “max” sensitivity. It may or may not align with the X, Y or Z
axis of the device. Even looking at X,Y, and Z, it’s a good bet one of them will have significantly
lower sensitivity.
-
OCXO’s are temperature sensitive. Transient changes in temperature drive them a bit nuts.
Drafts are one easy way create a temperature transient. You can easily see 10’s of ppb of
change for some pretty minor changes in air flow.
-
If you plot phase ( and thus frequency) on an OCXO in stable conditions it does move this
way and that. Half the time, it’s going to move in the “expected direction”. This stretches out
the needed measurement interval.
-
OCXO’s often take a long while to stabilize. How long depends on a lot of things. Some
very well known designs can take months to “calm down” after being off power / in storage
for a number of years. This very much does impact the level of data you can collect on them.
If you move an OCXO 1 meter in altitude you “change gravity” by about 0.3 ppm per meter.
https://walter.bislins.ch/bloge/index.asp?page=Earth+Gravity+Calculator
Net result is that an OCXO with 1 ppb sensitivity would change by about 0.3 x 10^-15 due to
this effect.
How accurate is your measurement? :)
Bob
On Feb 16, 2026, at 8:37 AM, J Decker via time-nuts time-nuts@lists.febo.com wrote:
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment
to set the internal PWM; the GPS sync is close but it can have many
nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength
)... but anyway, these two devices I was wondering if they were sensitive
enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I
fiddled with them for quite a while then realized that mainly the
difference was, when I picked it up, it was no long exactly as level as it
was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the
front up and the back down they would run faster, and the back up and front
down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the
crystal and change the oscillation rate. (A new technique for
overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning,
and got those set close to the first set, and tested orientation; one was
just as bad as the first pair, the other barely experienced any change at
all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain
orientation, then placed them down in a slightly different orientation and
have an immediate drift.
a lot of specs don't mention the drift for accelerations in various
directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts <
time-nuts@lists.febo.com> wrote:
Hello Bob,
It’s difficult to find detailed documentation on the manufacturing process
because each manufacturer keeps its methods proprietary, and the techniques
have evolved significantly over the years.
For high-end crystal oscillators, there are multiple steps—from blank
cutting, adjustment, cleaning, to encapsulation—to ensure optimal
performance.
During lapping, the parts are not only measured but also optically
inspected to verify whether they are perfectly planar or have the required
convex profile.
After the electrodes are added, there are two main approaches for
frequency adjustment:
- Depositing material to fine-tune the frequency
- Plasma removal to trim the frequency
Both techniques have their advantages and disadvantages.
As you mentioned, the base, crystal, and cap are cleaned using specialized
detergents and handled in a cleanroom environment. Several high-temperature
cycles are then applied to stabilize atomic migration in the plating
material, which improves long-term aging performance.
The base and cap are handled with dedicated tools, and cold welding is
carried out by applying high pressure under vacuum in a controlled
low-temperature environment.
Another important consideration is avoiding any outgassing materials,
including the adhesive used to mount the crystal on the base.
There is actually a lot to say about the process and the many steps
involved in building a high-quality crystal. On our side alone, we have
more than 20 process steps, which gives you an idea of why the crystal ends
up being one of the main cost drivers in an OCXO design.
Best regards,
Jean-Charles
Best regards,
-----Message d'origine-----
De : Bob Bownes via time-nuts time-nuts@lists.febo.com
Envoyé : mardi 10 février 2026 16:48
À : mcleannb--- via time-nuts time-nuts@lists.febo.com
Cc : Bob Bownes bownes@gmail.com
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Should anyone want any, I think I still have a whole bunch of quartz slabs
from a military crystal/SAW mfr that I'm willing to share. No plating, some
may be optical beam splitters, but most are quartz blanks of a large
variety of sizes. I might have tossed them in the most recent Great Purge,
but if anyone is interested, lmk, and I'll check next time I'm in the
garage.
Bob
On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
Hi all,
I remember visiting a crystal manufacturing shop in the town where I
grew up as a teenager in the 1970's. The guy would make the crystals to
frequency or ever so slightly high. The ones that were slightly high he
would draw a tiny line of lead pencil to "pull" the frequency low.
Thank you
Nic McLean
President & Squad Captain
VRA Rescue NSW - WICEN
6 Judy Jakins Drive, Dubbo, NSW 2830
1300 872 777 or 0417 822 728
wicen.president@vrarescue.org
-----Original Message-----
From: Bob kb8tq via time-nutstime-nuts@lists.febo.com
Sent: Tuesday, 10 February 2026 9:53 AM
To: Discussion of precise time and frequency
measurementtime-nuts@lists.febo.com
Cc: Bob kb8tqkb8tq@n1k.org
Subject: [time-nuts] Re: Question on quartz crystal manufacturing
Hi
The problem is that you can’t measure the frequency of a bare blank very
well. You need those plated electrodes and the ultra clean package to get
the “other stuff” out of the measurement.
In order to see what’s going on, you need to turn it into a finished
crystal. This is what’s done on things like the USO’s. They build lots of
finished crystals and sort them out. Each one goes into a “dummy”
OCXO and is tested for various parameters. The ones that meet all the
requirements move on into the finished product.
On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
The "make some blanks and run them in an oscillator for a while, before
finalizing the assembly" is what APL does for Ultra Stable Oscillators for
space flight.
I can't recall if they actually assemble the physics package (the part
that goes in a dewar) and run them before choosing the ones to fully
integrate. It would make sense, because one thing you want to deal with the
the residual stresses of the crystal in the holder, etc. And to identify
those that were "on frequency and didn't drift too far" and "didn't have
oddities in the aging process". They might have put them in the dewar
before the long duration burn in. I think there is some pretest too, I mean
you have to make sure that you've ground/finished/plated them to hit the
desired frequency.
As I recall for one batch of USOs, there were several dozen fabricated
and aged, of which 2 got finally used. From roughly 1000 blanks.
At $1M a copy, that fairly fits Bob's high BoM cost description.
On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
Hi
Well, never say never :) :) :).
It’s certainly not been part of any process I have been involved with.
However it may well have been done by others. I am not aware of it
being done by anybody these days. Again that could just be my not
having talked to the right folks. It also may (as you suggest) be
something done “long long ago”.
One important note not mentioned earlier:
That assumes that “precision crystal” = stuff like OCXO crystals.
Even that term may have multiple meanings to various folks. A 3rd or
5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass
package is what comes quickly to mind.
What is indeed done when processing these crystals:
The crystal growth process for synthetic quartz gets more involved
when precision parts are involved. You go from a fairly quick growing
approach to one that takes much longer.
That does involve a range of steps past just the basic growth process.
Sealed crystals after fabrication are normally run through bake and
The objective is to reduce the “front end aging” (or whatever you
want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
Occasionally processes do include running a crystal in a circuit
prior to fabricating the final oscillator. The most common example is
with low volume OCXO’s that have a very high BOM. Sorting crystals
for aging makes sense if > 90% of the crystals will be tossed out. For
the vast majority of OCXO’s made, this is not needed / not useful.
On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
Bob kb8tq via time-nuts writes:
- How is the cap placed on top of the holder under vacuum?
There is a step before this where the blank is finish plated to put
The manipulation process is a “that depends” sort of thing. It
could simply be “folks” using their hands via big black “vacuum
gloves”. It could be something more automated.
Somewhere, possibly BSTJ, I read about quartz for precision crystals
being 'stress-tested' and 'annealed' a process which involved forced
oscillation while gradually lowering temperature and
voltage/(amplitude?)
Was that (only) a thing with natural quartz ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
Hi
OCXO’s have a “g” sensitivity in the 0.1 to 2 ppb / G range. Exotic ones might be well below
that. You are unlikely to run into one of those.
Simple test: Flip the unit over and see how many more ns / s it changes. One ppb is 1 ns / s.
That sounds easy. There are a few “gotcha’s”:
1) The sensitivity is always the “max” sensitivity. It may or may not align with the X, Y or Z
axis of the device. Even looking at X,Y, and Z, it’s a good bet one of them will have significantly
lower sensitivity.
2) OCXO’s are temperature sensitive. Transient changes in temperature drive them a bit nuts.
Drafts are one easy way create a temperature transient. You can easily see 10’s of ppb of
change for some pretty minor changes in air flow.
3) If you plot phase ( and thus frequency) on an OCXO in stable conditions it does move this
way and that. Half the time, it’s going to move in the “expected direction”. This stretches out
the needed measurement interval.
4) OCXO’s often take a long while to stabilize. How long depends on a lot of things. Some
very well known designs can take months to “calm down” after being off power / in storage
for a number of years. This very much does impact the level of data you can collect on them.
If you move an OCXO 1 meter in altitude you “change gravity” by about 0.3 ppm per meter.
https://walter.bislins.ch/bloge/index.asp?page=Earth+Gravity+Calculator
Net result is that an OCXO with 1 ppb sensitivity would change by about 0.3 x 10^-15 due to
this effect.
How accurate is your measurement? :)
Bob
> On Feb 16, 2026, at 8:37 AM, J Decker via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Something I ran into about crystals, is the cut and package can affect the
> stability relative to gravitational orientation. I have two GPS
> conditioned OCXO crystal based devices that even have a manual adjustment
> to set the internal PWM; the GPS sync is close but it can have many
> nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength
> )... but anyway, these two devices I was wondering if they were sensitive
> enough to measure a difference of time dilation within a few meters
> height. I picked up one, and moved it up, and it drifted faster than the
> bottom one... (maybe?) then reversed it, and got a different drift rate; I
> fiddled with them for quite a while then realized that mainly the
> difference was, when I picked it up, it was no long exactly as level as it
> was, and in my perspective 'this looks flat to the other' was grossly off.
> I then went to a more controlled experiment, and found that tipping the
> front up and the back down they would run faster, and the back up and front
> down, the moved slower, turning them didn't matter... and there was even a
> difference tilting left/right. I did some research and found that
> gravitational differences in orientation can put different stresses on the
> crystal and change the oscillation rate. (A new technique for
> overclockers? Turn the system? :) )
>
> I had a few other pairs of ocxo circuits which just had resistive tuning,
> and got those set close to the first set, and tested orientation; one was
> just as bad as the first pair, the other barely experienced any change at
> all in any orientation.
>
> So, some drift between pairs I found was because I tuned them in a certain
> orientation, then placed them down in a slightly different orientation and
> have an immediate drift.
>
> a lot of specs don't mention the drift for accelerations in various
> directions.
>
> On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts <
> time-nuts@lists.febo.com> wrote:
>
>> Hello Bob,
>>
>> It’s difficult to find detailed documentation on the manufacturing process
>> because each manufacturer keeps its methods proprietary, and the techniques
>> have evolved significantly over the years.
>>
>> For high-end crystal oscillators, there are multiple steps—from blank
>> cutting, adjustment, cleaning, to encapsulation—to ensure optimal
>> performance.
>>
>> During lapping, the parts are not only measured but also optically
>> inspected to verify whether they are perfectly planar or have the required
>> convex profile.
>>
>> After the electrodes are added, there are two main approaches for
>> frequency adjustment:
>>
>> - Depositing material to fine-tune the frequency
>> - Plasma removal to trim the frequency
>>
>> Both techniques have their advantages and disadvantages.
>>
>> As you mentioned, the base, crystal, and cap are cleaned using specialized
>> detergents and handled in a cleanroom environment. Several high-temperature
>> cycles are then applied to stabilize atomic migration in the plating
>> material, which improves long-term aging performance.
>>
>> The base and cap are handled with dedicated tools, and cold welding is
>> carried out by applying high pressure under vacuum in a controlled
>> low-temperature environment.
>>
>> Another important consideration is avoiding any outgassing materials,
>> including the adhesive used to mount the crystal on the base.
>>
>> There is actually a lot to say about the process and the many steps
>> involved in building a high-quality crystal. On our side alone, we have
>> more than 20 process steps, which gives you an idea of why the crystal ends
>> up being one of the main cost drivers in an OCXO design.
>>
>> Best regards,
>>
>> Jean-Charles
>>
>> Best regards,
>> -----Message d'origine-----
>> De : Bob Bownes via time-nuts <time-nuts@lists.febo.com>
>> Envoyé : mardi 10 février 2026 16:48
>> À : mcleannb--- via time-nuts <time-nuts@lists.febo.com>
>> Cc : Bob Bownes <bownes@gmail.com>
>> Objet : [time-nuts] Re: Question on quartz crystal manufacturing
>>
>>
>> Should anyone want any, I think I still have a whole bunch of quartz slabs
>> from a military crystal/SAW mfr that I'm willing to share. No plating, some
>> may be optical beam splitters, but most are quartz blanks of a large
>> variety of sizes. I might have tossed them in the most recent Great Purge,
>> but if anyone is interested, lmk, and I'll check next time I'm in the
>> garage.
>>
>> Bob
>>
>>
>> On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
>>> Hi all,
>>> I remember visiting a crystal manufacturing shop in the town where I
>> grew up as a teenager in the 1970's. The guy would make the crystals to
>> frequency or ever so slightly high. The ones that were slightly high he
>> would draw a tiny line of lead pencil to "pull" the frequency low.
>>>
>>> Thank you
>>>
>>> Nic McLean
>>> President & Squad Captain
>>> VRA Rescue NSW - WICEN
>>> 6 Judy Jakins Drive, Dubbo, NSW 2830
>>> 1300 872 777 or 0417 822 728
>>> wicen.president@vrarescue.org
>>>
>>>
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: Bob kb8tq via time-nuts<time-nuts@lists.febo.com>
>>> Sent: Tuesday, 10 February 2026 9:53 AM
>>> To: Discussion of precise time and frequency
>>> measurement<time-nuts@lists.febo.com>
>>> Cc: Bob kb8tq<kb8tq@n1k.org>
>>> Subject: [time-nuts] Re: Question on quartz crystal manufacturing
>>>
>>> Hi
>>>
>>> The problem is that you can’t measure the frequency of a bare blank very
>> well. You need those plated electrodes and the ultra clean package to get
>> the “other stuff” out of the measurement.
>>>
>>> In order to see what’s going on, you need to turn it into a finished
>> crystal. This is what’s done on things like the USO’s. They build *lots* of
>> finished crystals and sort them out. Each one goes into a “dummy”
>>> OCXO and is tested for various parameters. The ones that meet all the
>> requirements move on into the finished product.
>>>
>>> Bob
>>>
>>>> On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
>> time-nuts@lists.febo.com> wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> The "make some blanks and run them in an oscillator for a while, before
>> finalizing the assembly" is what APL does for Ultra Stable Oscillators for
>> space flight.
>>>> I can't recall if they actually assemble the physics package (the part
>> that goes in a dewar) and run them before choosing the ones to fully
>> integrate. It would make sense, because one thing you want to deal with the
>> the residual stresses of the crystal in the holder, etc. And to identify
>> those that were "on frequency and didn't drift too far" and "didn't have
>> oddities in the aging process". They might have put them in the dewar
>> before the long duration burn in. I think there is some pretest too, I mean
>> you have to make sure that you've ground/finished/plated them to hit the
>> desired frequency.
>>>>
>>>> As I recall for one batch of USOs, there were several dozen fabricated
>> and aged, of which 2 got finally used. From roughly 1000 blanks.
>>>>
>>>> At $1M a copy, that fairly fits Bob's high BoM cost description.
>>>>
>>>>
>>>>
>>>> On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
>> time-nuts@lists.febo.com> wrote:
>>>>
>>>> Hi
>>>>
>>>> Well, never say never :) :) :).
>>>>
>>>> It’s certainly not been part of any process I have been involved with.
>>>> However it may well have been done by others. I am not aware of it
>>>> being done by anybody these days. Again that could just be my not
>>>> having talked to the right folks. It also may (as you suggest) be
>> something done “long long ago”.
>>>>
>>>> One important note not mentioned earlier:
>>>>
>>>> That assumes that “precision crystal” = stuff like OCXO crystals.
>>>> Even that term may have multiple meanings to various folks. A 3rd or
>>>> 5th overtone crystal in the 4 to 12 MHz region in a cold weld or glass
>> package is what comes quickly to mind.
>>>>
>>>> What is indeed done when processing these crystals:
>>>>
>>>> The crystal growth process for synthetic quartz gets more involved
>>>> when precision parts are involved. You go from a fairly quick growing
>> approach to one that takes much longer.
>>>> That does involve a range of steps past just the basic growth process.
>>>>
>>>> Sealed crystals after fabrication are normally run through bake and
>> thermo-cycle processes.
>>>> The objective is to reduce the “front end aging” (or whatever you
>>>> want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
>> weeks saves money.
>>>>
>>>> Occasionally processes do include running a crystal in a circuit
>>>> prior to fabricating the final oscillator. The most common example is
>>>> with low volume OCXO’s that have a very high BOM. Sorting crystals
>>>> for aging makes sense if > 90% of the crystals will be tossed out. For
>> the vast majority of OCXO’s made, this is not needed / not useful.
>>>>
>>>> Fun
>>>>
>>>> Bob
>>>>
>>>>> On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
>>>>>
>>>>> --------
>>>>> Bob kb8tq via time-nuts writes:
>>>>>
>>>>>>> 3) How is the cap placed on top of the holder under vacuum?
>>>>>> There is a step before this where the blank is finish plated to put
>> it on frequency.
>>>>>>
>>>>>> The manipulation process is a “that depends” sort of thing. It
>>>>>> could simply be “folks” using their hands via big black “vacuum
>> gloves”. It could be something more automated.
>>>>> Somewhere, possibly BSTJ, I read about quartz for precision crystals
>>>>> being 'stress-tested' and 'annealed' a process which involved forced
>>>>> oscillation while gradually lowering temperature and
>>>>> voltage/(amplitude?)
>>>>>
>>>>> Was that (only) a thing with natural quartz ?
>>>>>
>>>>> --
>>>>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
>>>> _______________________________________________
>>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>>> send an email totime-nuts-leave@lists.febo.com
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>>> send an email totime-nuts-leave@lists.febo.com
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
>>> an email totime-nuts-leave@lists.febo.com
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe send
>>> an email totime-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an
>> email to time-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com
>> To unsubscribe send an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
JB
Jean-Charles BILLEBAULT
Thu, Feb 19, 2026 3:03 PM
Hello J,
You’re absolutely on the right track — what you observed is a well-known effect called acceleration (or g) sensitivity of quartz oscillators.
A quartz crystal’s oscillation frequency depends on its mechanical strain.
When you tilt the oscillator, gravity applies a different static force vector. That slightly deforms the crystal blank.The resonant frequency shifts. It is often expressed in ppb/g and the sensitivity is not the same in all direction.
New resonator structure allow sensitivy 10 to 100 time lower but those are not standard products.
So your trials were dominated by mechanical stress mainly.
SC-cut are usually much better than AT-cut. But also the mounting technology has an impact, as an example 2-point or “tuning-fork” style mounts have higher sensitivity along certain axes while 3-point mount have better cancellation or limitation of stress.
Regards,
JC
-----Message d'origine-----
De : J Decker via time-nuts time-nuts@lists.febo.com
Envoyé : lundi 16 février 2026 14:37
À : Discussion of precise time and frequency measurement time-nuts@lists.febo.com
Cc : J Decker d3ck0r@gmail.com
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment to set the internal PWM; the GPS sync is close but it can have many nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength )... but anyway, these two devices I was wondering if they were sensitive enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I fiddled with them for quite a while then realized that mainly the difference was, when I picked it up, it was no long exactly as level as it was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the front up and the back down they would run faster, and the back up and front down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the crystal and change the oscillation rate. (A new technique for overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning, and got those set close to the first set, and tested orientation; one was just as bad as the first pair, the other barely experienced any change at all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain orientation, then placed them down in a slightly different orientation and have an immediate drift.
a lot of specs don't mention the drift for accelerations in various directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts < time-nuts@lists.febo.com> wrote:
Hello Bob,
It’s difficult to find detailed documentation on the manufacturing
process because each manufacturer keeps its methods proprietary, and
the techniques have evolved significantly over the years.
For high-end crystal oscillators, there are multiple steps—from blank
cutting, adjustment, cleaning, to encapsulation—to ensure optimal
performance.
During lapping, the parts are not only measured but also optically
inspected to verify whether they are perfectly planar or have the
required convex profile.
After the electrodes are added, there are two main approaches for
frequency adjustment:
- Depositing material to fine-tune the frequency
- Plasma removal to trim the frequency
Both techniques have their advantages and disadvantages.
As you mentioned, the base, crystal, and cap are cleaned using
specialized detergents and handled in a cleanroom environment. Several
high-temperature cycles are then applied to stabilize atomic migration
in the plating material, which improves long-term aging performance.
The base and cap are handled with dedicated tools, and cold welding is
carried out by applying high pressure under vacuum in a controlled
low-temperature environment.
Another important consideration is avoiding any outgassing materials,
including the adhesive used to mount the crystal on the base.
There is actually a lot to say about the process and the many steps
involved in building a high-quality crystal. On our side alone, we
have more than 20 process steps, which gives you an idea of why the
crystal ends up being one of the main cost drivers in an OCXO design.
Best regards,
Jean-Charles
Best regards,
-----Message d'origine-----
De : Bob Bownes via time-nuts time-nuts@lists.febo.com Envoyé :
mardi 10 février 2026 16:48 À : mcleannb--- via time-nuts
time-nuts@lists.febo.com Cc : Bob Bownes bownes@gmail.com Objet :
[time-nuts] Re: Question on quartz crystal manufacturing
Should anyone want any, I think I still have a whole bunch of quartz
slabs from a military crystal/SAW mfr that I'm willing to share. No
plating, some may be optical beam splitters, but most are quartz
blanks of a large variety of sizes. I might have tossed them in the
most recent Great Purge, but if anyone is interested, lmk, and I'll
check next time I'm in the garage.
Bob
On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
Hi all,
I remember visiting a crystal manufacturing shop in the town where I
grew up as a teenager in the 1970's. The guy would make the crystals
to frequency or ever so slightly high. The ones that were slightly
high he would draw a tiny line of lead pencil to "pull" the frequency low.
Thank you
Nic McLean
President & Squad Captain
VRA Rescue NSW - WICEN
6 Judy Jakins Drive, Dubbo, NSW 2830
1300 872 777 or 0417 822 728
wicen.president@vrarescue.org
-----Original Message-----
From: Bob kb8tq via time-nutstime-nuts@lists.febo.com
Sent: Tuesday, 10 February 2026 9:53 AM
To: Discussion of precise time and frequency
measurementtime-nuts@lists.febo.com
Cc: Bob kb8tqkb8tq@n1k.org
Subject: [time-nuts] Re: Question on quartz crystal manufacturing
Hi
The problem is that you can’t measure the frequency of a bare blank
very
well. You need those plated electrodes and the ultra clean package to
get the “other stuff” out of the measurement.
In order to see what’s going on, you need to turn it into a finished
crystal. This is what’s done on things like the USO’s. They build
lots of finished crystals and sort them out. Each one goes into a “dummy”
OCXO and is tested for various parameters. The ones that meet all
the
requirements move on into the finished product.
On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
The "make some blanks and run them in an oscillator for a while,
before
finalizing the assembly" is what APL does for Ultra Stable Oscillators
for space flight.
I can't recall if they actually assemble the physics package (the
part
that goes in a dewar) and run them before choosing the ones to fully
integrate. It would make sense, because one thing you want to deal
with the the residual stresses of the crystal in the holder, etc. And
to identify those that were "on frequency and didn't drift too far"
and "didn't have oddities in the aging process". They might have put
them in the dewar before the long duration burn in. I think there is
some pretest too, I mean you have to make sure that you've
ground/finished/plated them to hit the desired frequency.
As I recall for one batch of USOs, there were several dozen
fabricated
and aged, of which 2 got finally used. From roughly 1000 blanks.
At $1M a copy, that fairly fits Bob's high BoM cost description.
On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
Hi
Well, never say never :) :) :).
It’s certainly not been part of any process I have been involved with.
However it may well have been done by others. I am not aware of it
being done by anybody these days. Again that could just be my not
having talked to the right folks. It also may (as you suggest) be
something done “long long ago”.
One important note not mentioned earlier:
That assumes that “precision crystal” = stuff like OCXO crystals.
Even that term may have multiple meanings to various folks. A 3rd
or 5th overtone crystal in the 4 to 12 MHz region in a cold weld or
glass
package is what comes quickly to mind.
What is indeed done when processing these crystals:
The crystal growth process for synthetic quartz gets more involved
when precision parts are involved. You go from a fairly quick
growing
approach to one that takes much longer.
That does involve a range of steps past just the basic growth process.
Sealed crystals after fabrication are normally run through bake and
The objective is to reduce the “front end aging” (or whatever you
want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
Occasionally processes do include running a crystal in a circuit
prior to fabricating the final oscillator. The most common example
is with low volume OCXO’s that have a very high BOM. Sorting
crystals for aging makes sense if > 90% of the crystals will be
tossed out. For
the vast majority of OCXO’s made, this is not needed / not useful.
On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
Bob kb8tq via time-nuts writes:
- How is the cap placed on top of the holder under vacuum?
There is a step before this where the blank is finish plated to
put
The manipulation process is a “that depends” sort of thing. It
could simply be “folks” using their hands via big black “vacuum
gloves”. It could be something more automated.
Somewhere, possibly BSTJ, I read about quartz for precision
crystals being 'stress-tested' and 'annealed' a process which
involved forced oscillation while gradually lowering temperature
and
voltage/(amplitude?)
Was that (only) a thing with natural quartz ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
Hello J,
You’re absolutely on the right track — what you observed is a well-known effect called acceleration (or g) sensitivity of quartz oscillators.
A quartz crystal’s oscillation frequency depends on its mechanical strain.
When you tilt the oscillator, gravity applies a different static force vector. That slightly deforms the crystal blank.The resonant frequency shifts. It is often expressed in ppb/g and the sensitivity is not the same in all direction.
New resonator structure allow sensitivy 10 to 100 time lower but those are not standard products.
So your trials were dominated by mechanical stress mainly.
SC-cut are usually much better than AT-cut. But also the mounting technology has an impact, as an example 2-point or “tuning-fork” style mounts have higher sensitivity along certain axes while 3-point mount have better cancellation or limitation of stress.
Regards,
JC
-----Message d'origine-----
De : J Decker via time-nuts <time-nuts@lists.febo.com>
Envoyé : lundi 16 février 2026 14:37
À : Discussion of precise time and frequency measurement <time-nuts@lists.febo.com>
Cc : J Decker <d3ck0r@gmail.com>
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment to set the internal PWM; the GPS sync is close but it can have many nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength )... but anyway, these two devices I was wondering if they were sensitive enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I fiddled with them for quite a while then realized that mainly the difference was, when I picked it up, it was no long exactly as level as it was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the front up and the back down they would run faster, and the back up and front down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the crystal and change the oscillation rate. (A new technique for overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning, and got those set close to the first set, and tested orientation; one was just as bad as the first pair, the other barely experienced any change at all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain orientation, then placed them down in a slightly different orientation and have an immediate drift.
a lot of specs don't mention the drift for accelerations in various directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts < time-nuts@lists.febo.com> wrote:
> Hello Bob,
>
> It’s difficult to find detailed documentation on the manufacturing
> process because each manufacturer keeps its methods proprietary, and
> the techniques have evolved significantly over the years.
>
> For high-end crystal oscillators, there are multiple steps—from blank
> cutting, adjustment, cleaning, to encapsulation—to ensure optimal
> performance.
>
> During lapping, the parts are not only measured but also optically
> inspected to verify whether they are perfectly planar or have the
> required convex profile.
>
> After the electrodes are added, there are two main approaches for
> frequency adjustment:
>
> - Depositing material to fine-tune the frequency
> - Plasma removal to trim the frequency
>
> Both techniques have their advantages and disadvantages.
>
> As you mentioned, the base, crystal, and cap are cleaned using
> specialized detergents and handled in a cleanroom environment. Several
> high-temperature cycles are then applied to stabilize atomic migration
> in the plating material, which improves long-term aging performance.
>
> The base and cap are handled with dedicated tools, and cold welding is
> carried out by applying high pressure under vacuum in a controlled
> low-temperature environment.
>
> Another important consideration is avoiding any outgassing materials,
> including the adhesive used to mount the crystal on the base.
>
> There is actually a lot to say about the process and the many steps
> involved in building a high-quality crystal. On our side alone, we
> have more than 20 process steps, which gives you an idea of why the
> crystal ends up being one of the main cost drivers in an OCXO design.
>
> Best regards,
>
> Jean-Charles
>
> Best regards,
> -----Message d'origine-----
> De : Bob Bownes via time-nuts <time-nuts@lists.febo.com> Envoyé :
> mardi 10 février 2026 16:48 À : mcleannb--- via time-nuts
> <time-nuts@lists.febo.com> Cc : Bob Bownes <bownes@gmail.com> Objet :
> [time-nuts] Re: Question on quartz crystal manufacturing
>
>
> Should anyone want any, I think I still have a whole bunch of quartz
> slabs from a military crystal/SAW mfr that I'm willing to share. No
> plating, some may be optical beam splitters, but most are quartz
> blanks of a large variety of sizes. I might have tossed them in the
> most recent Great Purge, but if anyone is interested, lmk, and I'll
> check next time I'm in the garage.
>
> Bob
>
>
> On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
> > Hi all,
> > I remember visiting a crystal manufacturing shop in the town where I
> grew up as a teenager in the 1970's. The guy would make the crystals
> to frequency or ever so slightly high. The ones that were slightly
> high he would draw a tiny line of lead pencil to "pull" the frequency low.
> >
> > Thank you
> >
> > Nic McLean
> > President & Squad Captain
> > VRA Rescue NSW - WICEN
> > 6 Judy Jakins Drive, Dubbo, NSW 2830
> > 1300 872 777 or 0417 822 728
> > wicen.president@vrarescue.org
> >
> >
> >
> >
> >
> > -----Original Message-----
> > From: Bob kb8tq via time-nuts<time-nuts@lists.febo.com>
> > Sent: Tuesday, 10 February 2026 9:53 AM
> > To: Discussion of precise time and frequency
> > measurement<time-nuts@lists.febo.com>
> > Cc: Bob kb8tq<kb8tq@n1k.org>
> > Subject: [time-nuts] Re: Question on quartz crystal manufacturing
> >
> > Hi
> >
> > The problem is that you can’t measure the frequency of a bare blank
> > very
> well. You need those plated electrodes and the ultra clean package to
> get the “other stuff” out of the measurement.
> >
> > In order to see what’s going on, you need to turn it into a finished
> crystal. This is what’s done on things like the USO’s. They build
> *lots* of finished crystals and sort them out. Each one goes into a “dummy”
> > OCXO and is tested for various parameters. The ones that meet all
> > the
> requirements move on into the finished product.
> >
> > Bob
> >
> >> On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
> time-nuts@lists.febo.com> wrote:
> >>
> >>
> >>
> >>
> >>
> >>
> >> The "make some blanks and run them in an oscillator for a while,
> >> before
> finalizing the assembly" is what APL does for Ultra Stable Oscillators
> for space flight.
> >> I can't recall if they actually assemble the physics package (the
> >> part
> that goes in a dewar) and run them before choosing the ones to fully
> integrate. It would make sense, because one thing you want to deal
> with the the residual stresses of the crystal in the holder, etc. And
> to identify those that were "on frequency and didn't drift too far"
> and "didn't have oddities in the aging process". They might have put
> them in the dewar before the long duration burn in. I think there is
> some pretest too, I mean you have to make sure that you've
> ground/finished/plated them to hit the desired frequency.
> >>
> >> As I recall for one batch of USOs, there were several dozen
> >> fabricated
> and aged, of which 2 got finally used. From roughly 1000 blanks.
> >>
> >> At $1M a copy, that fairly fits Bob's high BoM cost description.
> >>
> >>
> >>
> >> On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
> time-nuts@lists.febo.com> wrote:
> >>
> >> Hi
> >>
> >> Well, never say never :) :) :).
> >>
> >> It’s certainly not been part of any process I have been involved with.
> >> However it may well have been done by others. I am not aware of it
> >> being done by anybody these days. Again that could just be my not
> >> having talked to the right folks. It also may (as you suggest) be
> something done “long long ago”.
> >>
> >> One important note not mentioned earlier:
> >>
> >> That assumes that “precision crystal” = stuff like OCXO crystals.
> >> Even that term may have multiple meanings to various folks. A 3rd
> >> or 5th overtone crystal in the 4 to 12 MHz region in a cold weld or
> >> glass
> package is what comes quickly to mind.
> >>
> >> What is indeed done when processing these crystals:
> >>
> >> The crystal growth process for synthetic quartz gets more involved
> >> when precision parts are involved. You go from a fairly quick
> >> growing
> approach to one that takes much longer.
> >> That does involve a range of steps past just the basic growth process.
> >>
> >> Sealed crystals after fabrication are normally run through bake and
> thermo-cycle processes.
> >> The objective is to reduce the “front end aging” (or whatever you
> >> want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
> weeks saves money.
> >>
> >> Occasionally processes do include running a crystal in a circuit
> >> prior to fabricating the final oscillator. The most common example
> >> is with low volume OCXO’s that have a very high BOM. Sorting
> >> crystals for aging makes sense if > 90% of the crystals will be
> >> tossed out. For
> the vast majority of OCXO’s made, this is not needed / not useful.
> >>
> >> Fun
> >>
> >> Bob
> >>
> >>> On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
> >>>
> >>> --------
> >>> Bob kb8tq via time-nuts writes:
> >>>
> >>>>> 3) How is the cap placed on top of the holder under vacuum?
> >>>> There is a step before this where the blank is finish plated to
> >>>> put
> it on frequency.
> >>>>
> >>>> The manipulation process is a “that depends” sort of thing. It
> >>>> could simply be “folks” using their hands via big black “vacuum
> gloves”. It could be something more automated.
> >>> Somewhere, possibly BSTJ, I read about quartz for precision
> >>> crystals being 'stress-tested' and 'annealed' a process which
> >>> involved forced oscillation while gradually lowering temperature
> >>> and
> >>> voltage/(amplitude?)
> >>>
> >>> Was that (only) a thing with natural quartz ?
> >>>
> >>> --
> >>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
> >> _______________________________________________
> >> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> >> send an email totime-nuts-leave@lists.febo.com
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> >> send an email totime-nuts-leave@lists.febo.com
> > _______________________________________________
> > time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> > send an email totime-nuts-leave@lists.febo.com
> > _______________________________________________
> > time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
> > send an email totime-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send
> an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send
> an email to time-nuts-leave@lists.febo.com
_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an email to time-nuts-leave@lists.febo.com
JD
J Decker
Fri, Feb 20, 2026 8:18 AM
Yes I thought magnetic field also, but I have many magnets, and placing
magnets in various position didn't have much effect.
To be more specific; These GPSDO GPS Tamed Thermostatic Crystal Oscillator
GPS Tamed Clock 10MHz Signal Source Global Positioning System Disciplined
Oscillator When flat, and oriented in any direction tangent to the
ground(yaw) ticked the same; if tilted there was a small variation, and
again yaw didn't matter. tipping it face down it ran faster, and face up
it ran slower, and still changing the yaw in that direction didn't matter
as much; tilting to the side would approach the side.
I personally wondered if it was a matter of the curvature of the space the
crystal is in; but it is any acceleration.
Another OCXO that's just a simple resistor tuned board changed slightly
under tilts - but the yaw didn't matter. I think its fast/slow direction
though was normal and upside down.
On Thu, Feb 19, 2026 at 11:15 PM Steve Hendrix, P.E. via time-nuts <
time-nuts@lists.febo.com> wrote:
On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
Something I ran into about crystals, is the cut and package can affect
stability relative to gravitational orientation.
Interesting experiments. I wonder if the earth's magnetic field could be
in play here. That is, does rotating them while keeping them "level"
change the game? Depending upon your latitude, there is a vertical
component to the magnetic field.
Steve Hendrix
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Yes I thought magnetic field also, but I have many magnets, and placing
magnets in various position didn't have much effect.
To be more specific; These GPSDO GPS Tamed Thermostatic Crystal Oscillator
GPS Tamed Clock 10MHz Signal Source Global Positioning System Disciplined
Oscillator When flat, and oriented in any direction tangent to the
ground(yaw) ticked the same; if tilted there was a small variation, and
again yaw didn't matter. tipping it face down it ran faster, and face up
it ran slower, and still changing the yaw in that direction didn't matter
as much; tilting to the side would approach the side.
I personally wondered if it was a matter of the curvature of the space the
crystal is in; but it is any acceleration.
Another OCXO that's just a simple resistor tuned board changed slightly
under tilts - but the yaw didn't matter. I think its fast/slow direction
though was normal and upside down.
On Thu, Feb 19, 2026 at 11:15 PM Steve Hendrix, P.E. via time-nuts <
time-nuts@lists.febo.com> wrote:
> On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
> > Something I ran into about crystals, is the cut and package can affect
> the
> > stability relative to gravitational orientation.
>
> Interesting experiments. I wonder if the earth's magnetic field could be
> in play here. That is, does rotating them while keeping them "level"
> change the game? Depending upon your latitude, there is a vertical
> component to the magnetic field.
>
> Steve Hendrix
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
>
MD
Magnus Danielson
Fri, Feb 20, 2026 8:52 AM
Hi,
On 2026-02-19 12:13, Steve Hendrix, P.E. via time-nuts wrote:
On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
Something I ran into about crystals, is the cut and package can
affect the
stability relative to gravitational orientation.
Interesting experiments. I wonder if the earth's magnetic field could
be in play here. That is, does rotating them while keeping them
"level" change the game? Depending upon your latitude, there is a
vertical component to the magnetic field.
These can be separated out and the gravitational effect is very clearly
there.
For environmental effects, and both gravitational and magnetic field
effects is included, has a separate standard called IEEE 1193. It has a
rich set of references to research material.
The gravitational effect is because the mass and graviational vector
creates a stress in the crystal blank, and that shifts the frequency.
The magnetic effect depends on how the resonator package is built and
the resonator is held, and how that creates mechanical stress on the
crystal blank. For instance, using steel clips to hold the resonator
makes it more sensitive to magnetic fields.
I know that electrostatic fields also affects, and there is research on
that.
Cheers,
Magnus
Hi,
On 2026-02-19 12:13, Steve Hendrix, P.E. via time-nuts wrote:
> On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
>> Something I ran into about crystals, is the cut and package can
>> affect the
>> stability relative to gravitational orientation.
>
> Interesting experiments. I wonder if the earth's magnetic field could
> be in play here. That is, does rotating them while keeping them
> "level" change the game? Depending upon your latitude, there is a
> vertical component to the magnetic field.
These can be separated out and the gravitational effect is very clearly
there.
For environmental effects, and both gravitational and magnetic field
effects is included, has a separate standard called IEEE 1193. It has a
*rich* set of references to research material.
The gravitational effect is because the mass and graviational vector
creates a stress in the crystal blank, and that shifts the frequency.
The magnetic effect depends on how the resonator package is built and
the resonator is held, and how that creates mechanical stress on the
crystal blank. For instance, using steel clips to hold the resonator
makes it more sensitive to magnetic fields.
I know that electrostatic fields also affects, and there is research on
that.
Cheers,
Magnus
BK
Bob kb8tq
Fri, Feb 20, 2026 1:12 PM
Hi
SC vs AT depends a lot on what you look at. Again, assuming you are looking at precision parts:
The “typical” numbers for an AT are in the 1 to 2 ppb range. For an SC you would expect 0.25 to 0.5 ppb.
Is that “a lot”? I guess it depends on what your needs are.
The gotcha is that both sets of numbers are typical rather than “you always get this”. They also are for the
worst case vector. You can indeed find AT’s that are below 1 ppb.
You also can find devices with custom fabrication processes (some patented) that will significantly reduce the
sensitivity. They are expensive and very uncommon.
Further confusing things: The typical test is to rotate the part 180 degrees. This takes you from +1 G to -1 G. Net
result is a “2G tip test”. Some folks report the data from that test directly. Thus you get data that is 2X larger than
you might expect.
Bob
On Feb 19, 2026, at 10:03 AM, Jean-Charles BILLEBAULT via time-nuts time-nuts@lists.febo.com wrote:
Hello J,
You’re absolutely on the right track — what you observed is a well-known effect called acceleration (or g) sensitivity of quartz oscillators.
A quartz crystal’s oscillation frequency depends on its mechanical strain.
When you tilt the oscillator, gravity applies a different static force vector. That slightly deforms the crystal blank.The resonant frequency shifts. It is often expressed in ppb/g and the sensitivity is not the same in all direction.
New resonator structure allow sensitivy 10 to 100 time lower but those are not standard products.
So your trials were dominated by mechanical stress mainly.
SC-cut are usually much better than AT-cut. But also the mounting technology has an impact, as an example 2-point or “tuning-fork” style mounts have higher sensitivity along certain axes while 3-point mount have better cancellation or limitation of stress.
Regards,
JC
-----Message d'origine-----
De : J Decker via time-nuts time-nuts@lists.febo.com
Envoyé : lundi 16 février 2026 14:37
À : Discussion of precise time and frequency measurement time-nuts@lists.febo.com
Cc : J Decker d3ck0r@gmail.com
Objet : [time-nuts] Re: Question on quartz crystal manufacturing
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation. I have two GPS
conditioned OCXO crystal based devices that even have a manual adjustment to set the internal PWM; the GPS sync is close but it can have many nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength )... but anyway, these two devices I was wondering if they were sensitive enough to measure a difference of time dilation within a few meters
height. I picked up one, and moved it up, and it drifted faster than the
bottom one... (maybe?) then reversed it, and got a different drift rate; I fiddled with them for quite a while then realized that mainly the difference was, when I picked it up, it was no long exactly as level as it was, and in my perspective 'this looks flat to the other' was grossly off.
I then went to a more controlled experiment, and found that tipping the front up and the back down they would run faster, and the back up and front down, the moved slower, turning them didn't matter... and there was even a
difference tilting left/right. I did some research and found that
gravitational differences in orientation can put different stresses on the crystal and change the oscillation rate. (A new technique for overclockers? Turn the system? :) )
I had a few other pairs of ocxo circuits which just had resistive tuning, and got those set close to the first set, and tested orientation; one was just as bad as the first pair, the other barely experienced any change at all in any orientation.
So, some drift between pairs I found was because I tuned them in a certain orientation, then placed them down in a slightly different orientation and have an immediate drift.
a lot of specs don't mention the drift for accelerations in various directions.
On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts < time-nuts@lists.febo.com> wrote:
Hello Bob,
It’s difficult to find detailed documentation on the manufacturing
process because each manufacturer keeps its methods proprietary, and
the techniques have evolved significantly over the years.
For high-end crystal oscillators, there are multiple steps—from blank
cutting, adjustment, cleaning, to encapsulation—to ensure optimal
performance.
During lapping, the parts are not only measured but also optically
inspected to verify whether they are perfectly planar or have the
required convex profile.
After the electrodes are added, there are two main approaches for
frequency adjustment:
- Depositing material to fine-tune the frequency
- Plasma removal to trim the frequency
Both techniques have their advantages and disadvantages.
As you mentioned, the base, crystal, and cap are cleaned using
specialized detergents and handled in a cleanroom environment. Several
high-temperature cycles are then applied to stabilize atomic migration
in the plating material, which improves long-term aging performance.
The base and cap are handled with dedicated tools, and cold welding is
carried out by applying high pressure under vacuum in a controlled
low-temperature environment.
Another important consideration is avoiding any outgassing materials,
including the adhesive used to mount the crystal on the base.
There is actually a lot to say about the process and the many steps
involved in building a high-quality crystal. On our side alone, we
have more than 20 process steps, which gives you an idea of why the
crystal ends up being one of the main cost drivers in an OCXO design.
Best regards,
Jean-Charles
Best regards,
-----Message d'origine-----
De : Bob Bownes via time-nuts time-nuts@lists.febo.com Envoyé :
mardi 10 février 2026 16:48 À : mcleannb--- via time-nuts
time-nuts@lists.febo.com Cc : Bob Bownes bownes@gmail.com Objet :
[time-nuts] Re: Question on quartz crystal manufacturing
Should anyone want any, I think I still have a whole bunch of quartz
slabs from a military crystal/SAW mfr that I'm willing to share. No
plating, some may be optical beam splitters, but most are quartz
blanks of a large variety of sizes. I might have tossed them in the
most recent Great Purge, but if anyone is interested, lmk, and I'll
check next time I'm in the garage.
Bob
On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
Hi all,
I remember visiting a crystal manufacturing shop in the town where I
grew up as a teenager in the 1970's. The guy would make the crystals
to frequency or ever so slightly high. The ones that were slightly
high he would draw a tiny line of lead pencil to "pull" the frequency low.
Thank you
Nic McLean
President & Squad Captain
VRA Rescue NSW - WICEN
6 Judy Jakins Drive, Dubbo, NSW 2830
1300 872 777 or 0417 822 728
wicen.president@vrarescue.org
-----Original Message-----
From: Bob kb8tq via time-nutstime-nuts@lists.febo.com
Sent: Tuesday, 10 February 2026 9:53 AM
To: Discussion of precise time and frequency
measurementtime-nuts@lists.febo.com
Cc: Bob kb8tqkb8tq@n1k.org
Subject: [time-nuts] Re: Question on quartz crystal manufacturing
Hi
The problem is that you can’t measure the frequency of a bare blank
very
well. You need those plated electrodes and the ultra clean package to
get the “other stuff” out of the measurement.
In order to see what’s going on, you need to turn it into a finished
crystal. This is what’s done on things like the USO’s. They build
lots of finished crystals and sort them out. Each one goes into a “dummy”
OCXO and is tested for various parameters. The ones that meet all
the
requirements move on into the finished product.
On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
The "make some blanks and run them in an oscillator for a while,
before
finalizing the assembly" is what APL does for Ultra Stable Oscillators
for space flight.
I can't recall if they actually assemble the physics package (the
part
that goes in a dewar) and run them before choosing the ones to fully
integrate. It would make sense, because one thing you want to deal
with the the residual stresses of the crystal in the holder, etc. And
to identify those that were "on frequency and didn't drift too far"
and "didn't have oddities in the aging process". They might have put
them in the dewar before the long duration burn in. I think there is
some pretest too, I mean you have to make sure that you've
ground/finished/plated them to hit the desired frequency.
As I recall for one batch of USOs, there were several dozen
fabricated
and aged, of which 2 got finally used. From roughly 1000 blanks.
At $1M a copy, that fairly fits Bob's high BoM cost description.
On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
Hi
Well, never say never :) :) :).
It’s certainly not been part of any process I have been involved with.
However it may well have been done by others. I am not aware of it
being done by anybody these days. Again that could just be my not
having talked to the right folks. It also may (as you suggest) be
something done “long long ago”.
One important note not mentioned earlier:
That assumes that “precision crystal” = stuff like OCXO crystals.
Even that term may have multiple meanings to various folks. A 3rd
or 5th overtone crystal in the 4 to 12 MHz region in a cold weld or
glass
package is what comes quickly to mind.
What is indeed done when processing these crystals:
The crystal growth process for synthetic quartz gets more involved
when precision parts are involved. You go from a fairly quick
growing
approach to one that takes much longer.
That does involve a range of steps past just the basic growth process.
Sealed crystals after fabrication are normally run through bake and
The objective is to reduce the “front end aging” (or whatever you
want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
Occasionally processes do include running a crystal in a circuit
prior to fabricating the final oscillator. The most common example
is with low volume OCXO’s that have a very high BOM. Sorting
crystals for aging makes sense if > 90% of the crystals will be
tossed out. For
the vast majority of OCXO’s made, this is not needed / not useful.
On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
Bob kb8tq via time-nuts writes:
- How is the cap placed on top of the holder under vacuum?
There is a step before this where the blank is finish plated to
put
The manipulation process is a “that depends” sort of thing. It
could simply be “folks” using their hands via big black “vacuum
gloves”. It could be something more automated.
Somewhere, possibly BSTJ, I read about quartz for precision
crystals being 'stress-tested' and 'annealed' a process which
involved forced oscillation while gradually lowering temperature
and
voltage/(amplitude?)
Was that (only) a thing with natural quartz ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
Hi
SC vs AT depends a lot on what you look at. Again, assuming you are looking at precision parts:
The “typical” numbers for an AT are in the 1 to 2 ppb range. For an SC you would expect 0.25 to 0.5 ppb.
Is that “a lot”? I guess it depends on what your needs are.
The gotcha is that both sets of numbers are typical rather than “you always get this”. They also are for the
worst case vector. You can indeed find AT’s that are below 1 ppb.
You also can find devices with custom fabrication processes (some patented) that will significantly reduce the
sensitivity. They are expensive and very uncommon.
Further confusing things: The typical test is to rotate the part 180 degrees. This takes you from +1 G to -1 G. Net
result is a “2G tip test”. Some folks report the data from that test directly. Thus you get data that is 2X larger than
you might expect.
Bob
> On Feb 19, 2026, at 10:03 AM, Jean-Charles BILLEBAULT via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Hello J,
>
> You’re absolutely on the right track — what you observed is a well-known effect called acceleration (or g) sensitivity of quartz oscillators.
> A quartz crystal’s oscillation frequency depends on its mechanical strain.
> When you tilt the oscillator, gravity applies a different static force vector. That slightly deforms the crystal blank.The resonant frequency shifts. It is often expressed in ppb/g and the sensitivity is not the same in all direction.
> New resonator structure allow sensitivy 10 to 100 time lower but those are not standard products.
> So your trials were dominated by mechanical stress mainly.
>
> SC-cut are usually much better than AT-cut. But also the mounting technology has an impact, as an example 2-point or “tuning-fork” style mounts have higher sensitivity along certain axes while 3-point mount have better cancellation or limitation of stress.
>
> Regards,
>
> JC
>
> -----Message d'origine-----
> De : J Decker via time-nuts <time-nuts@lists.febo.com>
> Envoyé : lundi 16 février 2026 14:37
> À : Discussion of precise time and frequency measurement <time-nuts@lists.febo.com>
> Cc : J Decker <d3ck0r@gmail.com>
> Objet : [time-nuts] Re: Question on quartz crystal manufacturing
>
> Something I ran into about crystals, is the cut and package can affect the
> stability relative to gravitational orientation. I have two GPS
> conditioned OCXO crystal based devices that even have a manual adjustment to set the internal PWM; the GPS sync is close but it can have many nanoseconds of drift from each other ( 6Mhz is a pretty long wavelength )... but anyway, these two devices I was wondering if they were sensitive enough to measure a difference of time dilation within a few meters
> height. I picked up one, and moved it up, and it drifted faster than the
> bottom one... (maybe?) then reversed it, and got a different drift rate; I fiddled with them for quite a while then realized that mainly the difference was, when I picked it up, it was no long exactly as level as it was, and in my perspective 'this looks flat to the other' was grossly off.
> I then went to a more controlled experiment, and found that tipping the front up and the back down they would run faster, and the back up and front down, the moved slower, turning them didn't matter... and there was even a
> difference tilting left/right. I did some research and found that
> gravitational differences in orientation can put different stresses on the crystal and change the oscillation rate. (A new technique for overclockers? Turn the system? :) )
>
> I had a few other pairs of ocxo circuits which just had resistive tuning, and got those set close to the first set, and tested orientation; one was just as bad as the first pair, the other barely experienced any change at all in any orientation.
>
> So, some drift between pairs I found was because I tuned them in a certain orientation, then placed them down in a slightly different orientation and have an immediate drift.
>
> a lot of specs don't mention the drift for accelerations in various directions.
>
> On Mon, Feb 16, 2026 at 3:00 AM Jean-Charles BILLEBAULT via time-nuts < time-nuts@lists.febo.com> wrote:
>
>> Hello Bob,
>>
>> It’s difficult to find detailed documentation on the manufacturing
>> process because each manufacturer keeps its methods proprietary, and
>> the techniques have evolved significantly over the years.
>>
>> For high-end crystal oscillators, there are multiple steps—from blank
>> cutting, adjustment, cleaning, to encapsulation—to ensure optimal
>> performance.
>>
>> During lapping, the parts are not only measured but also optically
>> inspected to verify whether they are perfectly planar or have the
>> required convex profile.
>>
>> After the electrodes are added, there are two main approaches for
>> frequency adjustment:
>>
>> - Depositing material to fine-tune the frequency
>> - Plasma removal to trim the frequency
>>
>> Both techniques have their advantages and disadvantages.
>>
>> As you mentioned, the base, crystal, and cap are cleaned using
>> specialized detergents and handled in a cleanroom environment. Several
>> high-temperature cycles are then applied to stabilize atomic migration
>> in the plating material, which improves long-term aging performance.
>>
>> The base and cap are handled with dedicated tools, and cold welding is
>> carried out by applying high pressure under vacuum in a controlled
>> low-temperature environment.
>>
>> Another important consideration is avoiding any outgassing materials,
>> including the adhesive used to mount the crystal on the base.
>>
>> There is actually a lot to say about the process and the many steps
>> involved in building a high-quality crystal. On our side alone, we
>> have more than 20 process steps, which gives you an idea of why the
>> crystal ends up being one of the main cost drivers in an OCXO design.
>>
>> Best regards,
>>
>> Jean-Charles
>>
>> Best regards,
>> -----Message d'origine-----
>> De : Bob Bownes via time-nuts <time-nuts@lists.febo.com> Envoyé :
>> mardi 10 février 2026 16:48 À : mcleannb--- via time-nuts
>> <time-nuts@lists.febo.com> Cc : Bob Bownes <bownes@gmail.com> Objet :
>> [time-nuts] Re: Question on quartz crystal manufacturing
>>
>>
>> Should anyone want any, I think I still have a whole bunch of quartz
>> slabs from a military crystal/SAW mfr that I'm willing to share. No
>> plating, some may be optical beam splitters, but most are quartz
>> blanks of a large variety of sizes. I might have tossed them in the
>> most recent Great Purge, but if anyone is interested, lmk, and I'll
>> check next time I'm in the garage.
>>
>> Bob
>>
>>
>> On 2/9/26 10:21 PM, mcleannb--- via time-nuts wrote:
>>> Hi all,
>>> I remember visiting a crystal manufacturing shop in the town where I
>> grew up as a teenager in the 1970's. The guy would make the crystals
>> to frequency or ever so slightly high. The ones that were slightly
>> high he would draw a tiny line of lead pencil to "pull" the frequency low.
>>>
>>> Thank you
>>>
>>> Nic McLean
>>> President & Squad Captain
>>> VRA Rescue NSW - WICEN
>>> 6 Judy Jakins Drive, Dubbo, NSW 2830
>>> 1300 872 777 or 0417 822 728
>>> wicen.president@vrarescue.org
>>>
>>>
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: Bob kb8tq via time-nuts<time-nuts@lists.febo.com>
>>> Sent: Tuesday, 10 February 2026 9:53 AM
>>> To: Discussion of precise time and frequency
>>> measurement<time-nuts@lists.febo.com>
>>> Cc: Bob kb8tq<kb8tq@n1k.org>
>>> Subject: [time-nuts] Re: Question on quartz crystal manufacturing
>>>
>>> Hi
>>>
>>> The problem is that you can’t measure the frequency of a bare blank
>>> very
>> well. You need those plated electrodes and the ultra clean package to
>> get the “other stuff” out of the measurement.
>>>
>>> In order to see what’s going on, you need to turn it into a finished
>> crystal. This is what’s done on things like the USO’s. They build
>> *lots* of finished crystals and sort them out. Each one goes into a “dummy”
>>> OCXO and is tested for various parameters. The ones that meet all
>>> the
>> requirements move on into the finished product.
>>>
>>> Bob
>>>
>>>> On Feb 9, 2026, at 2:56 PM, Jim Lux via time-nuts<
>> time-nuts@lists.febo.com> wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> The "make some blanks and run them in an oscillator for a while,
>>>> before
>> finalizing the assembly" is what APL does for Ultra Stable Oscillators
>> for space flight.
>>>> I can't recall if they actually assemble the physics package (the
>>>> part
>> that goes in a dewar) and run them before choosing the ones to fully
>> integrate. It would make sense, because one thing you want to deal
>> with the the residual stresses of the crystal in the holder, etc. And
>> to identify those that were "on frequency and didn't drift too far"
>> and "didn't have oddities in the aging process". They might have put
>> them in the dewar before the long duration burn in. I think there is
>> some pretest too, I mean you have to make sure that you've
>> ground/finished/plated them to hit the desired frequency.
>>>>
>>>> As I recall for one batch of USOs, there were several dozen
>>>> fabricated
>> and aged, of which 2 got finally used. From roughly 1000 blanks.
>>>>
>>>> At $1M a copy, that fairly fits Bob's high BoM cost description.
>>>>
>>>>
>>>>
>>>> On Mon, 9 Feb 2026 11:05:30 -0500, Bob kb8tq via time-nuts<
>> time-nuts@lists.febo.com> wrote:
>>>>
>>>> Hi
>>>>
>>>> Well, never say never :) :) :).
>>>>
>>>> It’s certainly not been part of any process I have been involved with.
>>>> However it may well have been done by others. I am not aware of it
>>>> being done by anybody these days. Again that could just be my not
>>>> having talked to the right folks. It also may (as you suggest) be
>> something done “long long ago”.
>>>>
>>>> One important note not mentioned earlier:
>>>>
>>>> That assumes that “precision crystal” = stuff like OCXO crystals.
>>>> Even that term may have multiple meanings to various folks. A 3rd
>>>> or 5th overtone crystal in the 4 to 12 MHz region in a cold weld or
>>>> glass
>> package is what comes quickly to mind.
>>>>
>>>> What is indeed done when processing these crystals:
>>>>
>>>> The crystal growth process for synthetic quartz gets more involved
>>>> when precision parts are involved. You go from a fairly quick
>>>> growing
>> approach to one that takes much longer.
>>>> That does involve a range of steps past just the basic growth process.
>>>>
>>>> Sealed crystals after fabrication are normally run through bake and
>> thermo-cycle processes.
>>>> The objective is to reduce the “front end aging” (or whatever you
>>>> want to call it). Having a fabricated OCXO hit spec in 5 days vs 5
>> weeks saves money.
>>>>
>>>> Occasionally processes do include running a crystal in a circuit
>>>> prior to fabricating the final oscillator. The most common example
>>>> is with low volume OCXO’s that have a very high BOM. Sorting
>>>> crystals for aging makes sense if > 90% of the crystals will be
>>>> tossed out. For
>> the vast majority of OCXO’s made, this is not needed / not useful.
>>>>
>>>> Fun
>>>>
>>>> Bob
>>>>
>>>>> On Feb 9, 2026, at 4:40 AM, Poul-Henning Kamp wrote:
>>>>>
>>>>> --------
>>>>> Bob kb8tq via time-nuts writes:
>>>>>
>>>>>>> 3) How is the cap placed on top of the holder under vacuum?
>>>>>> There is a step before this where the blank is finish plated to
>>>>>> put
>> it on frequency.
>>>>>>
>>>>>> The manipulation process is a “that depends” sort of thing. It
>>>>>> could simply be “folks” using their hands via big black “vacuum
>> gloves”. It could be something more automated.
>>>>> Somewhere, possibly BSTJ, I read about quartz for precision
>>>>> crystals being 'stress-tested' and 'annealed' a process which
>>>>> involved forced oscillation while gradually lowering temperature
>>>>> and
>>>>> voltage/(amplitude?)
>>>>>
>>>>> Was that (only) a thing with natural quartz ?
>>>>>
>>>>> --
>>>>> Poul-Henning Kamp | UNIX since Zilog Zeus 3.20phk@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.
>>>> _______________________________________________
>>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>>> send an email totime-nuts-leave@lists.febo.com
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>>> send an email totime-nuts-leave@lists.febo.com
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>> send an email totime-nuts-leave@lists.febo.com
>>> _______________________________________________
>>> time-nuts mailing list --time-nuts@lists.febo.com To unsubscribe
>>> send an email totime-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send
>> an email to time-nuts-leave@lists.febo.com
>> _______________________________________________
>> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send
>> an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com To unsubscribe send an email to time-nuts-leave@lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
BK
Bob kb8tq
Fri, Feb 20, 2026 1:42 PM
Hi
Unlike things like Cs and Rb standards, quartz crystals have no intrinsic magnetic field sensitivity.
Bob
On Feb 19, 2026, at 6:13 AM, Steve Hendrix, P.E. via time-nuts time-nuts@lists.febo.com wrote:
On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
Something I ran into about crystals, is the cut and package can affect the
stability relative to gravitational orientation.
Interesting experiments. I wonder if the earth's magnetic field could be in play here. That is, does rotating them while keeping them "level" change the game? Depending upon your latitude, there is a vertical component to the magnetic field.
Steve Hendrix
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi
Unlike things like Cs and Rb standards, quartz crystals have no intrinsic magnetic field sensitivity.
Bob
> On Feb 19, 2026, at 6:13 AM, Steve Hendrix, P.E. via time-nuts <time-nuts@lists.febo.com> wrote:
>
> On 2026-02-16 8:37 AM, J Decker via time-nuts wrote:
>> Something I ran into about crystals, is the cut and package can affect the
>> stability relative to gravitational orientation.
>
> Interesting experiments. I wonder if the earth's magnetic field could be in play here. That is, does rotating them while keeping them "level" change the game? Depending upon your latitude, there is a vertical component to the magnetic field.
>
> Steve Hendrix
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
TV
Tom Van Baak
Fri, Feb 20, 2026 3:00 PM
On this page I show a quartz oscillator being rotated 180° while being
measured with a high resolution frequency counter:
http://leapsecond.com/pages/bva-rotate/
The first plot shows the change in frequency for each of the 6 axis
orientations. This was done by hand. The delta-f is dramatic and each
axis has its own unique offset and range, like a hidden fingerprint.
Note after each one minute measurement the oscillator was returned to
its resting position. That was to make sure I was seeing mostly instant
acceleration (of gravity) effects, not gradual thermal effects in the OCXO.
The remaining plots show the effect of the oscillator mounted on an
automated rotary table. This is a test of the resolution of the counter
as much as a test of the acceleration sensitivity of the oscillator.
There is a 10-second video of the test in progress.
I am hoping one day to automate this demonstration with a robotic arm in
order to create a full sphere map of Δf/f as a function of x,y,z. If any
of you have expertise in 3D actuators, let me know.
/tvb
On this page I show a quartz oscillator being rotated 180° while being
measured with a high resolution frequency counter:
http://leapsecond.com/pages/bva-rotate/
The first plot shows the change in frequency for each of the 6 axis
orientations. This was done by hand. The delta-f is dramatic and each
axis has its own unique offset and range, like a hidden fingerprint.
Note after each one minute measurement the oscillator was returned to
its resting position. That was to make sure I was seeing mostly instant
acceleration (of gravity) effects, not gradual thermal effects in the OCXO.
The remaining plots show the effect of the oscillator mounted on an
automated rotary table. This is a test of the resolution of the counter
as much as a test of the acceleration sensitivity of the oscillator.
There is a 10-second video of the test in progress.
I am hoping one day to automate this demonstration with a robotic arm in
order to create a full sphere map of Δf/f as a function of x,y,z. If any
of you have expertise in 3D actuators, let me know.
/tvb
