Super stable BVA Quartz resonators... BVA??
Hi all,
For explanation how the BVA works, please see the attached slide which
shows schematically its internal construction. (The explaning text is in
German, as it is from my periodically given crystal seminar.)
The key points which yield the BVA's improved aging are, that
- the whole resonator package is made from quartz, consisting of the
resonationg quartz plate in the middle section and the two mounting &
sealing plaates on top and bottem - you may call it a Hamburger style ;-) - the resonating plate is held through quartz bridges rather than metal
suspensions, thus reducing mounting stress. - the BVA is "electrodeless", as the electrodes are evaporated on the
inner side of the upper and lower mounting plates. Therefore no stress
between the quartz surface and the metal eletrode. - Therefore the energizing electrical field is applied through an
airgap, which represents two small load capacitors in series to the
resonator, thus making the resonator electrically "stiffer" and less
sensitive to circuit influences.
On the other side you may imagine the main problems associated with such
a construction:
- The difficulty to manufacture the convex and concave shaped parts with
such a precision, that the curvature yields a constant and very small
airgap. - To realize the fine adjustment to frequency, because the unit cannot
be tuned in the conventional way, i.e. by plating some metal on the
electrode. - The frequency accuracy to which the resonator has to manufactured,
because the resonator frequency cannot be pulled with the external
circuit elements by more than some hundred ppb.
It may be interesting to note, that there was a company "BVA
Industries", which wanted to generate their income solely from making
BVA - which failed. Maybe because of the cost could not be covered by
the revenues from the rather limited market.
Bernd DK1AG
AXTAL GmbH & Co. KG
Facility MOS
Wasemweg 5
D-74821 Mosbach / Germany
fon: +49 (6261) 939834
fax: +49 (6261) 939836
www.axtal.com
Thanks for the description, it is very interesting. I have a follow up
question if you don't mind.
How does crystal aging look on a graph of temp versus frequency. This
graph has some temperature point where the slope of frequency variation
goes to zero and the crystal is quite stable around this temperature.
What does this graph look like as a crystal ages? Does the optimal
operating temperature change over time, that is, does the graph tend to
move left and right, or does aging tend to move the graph vertically
(the optimal freq stays the same, but the optimal operating temperature
changes as a result of aging).
I hope this makes sense...
jeff
Bernd T-Online wrote:
Hi all,
For explanation how the BVA works, please see the attached slide which
shows schematically its internal construction. (The explaning text is in
German, as it is from my periodically given crystal seminar.)
The key points which yield the BVA's improved aging are, that
- the whole resonator package is made from quartz, consisting of the
resonationg quartz plate in the middle section and the two mounting &
sealing plaates on top and bottem - you may call it a Hamburger style ;-) - the resonating plate is held through quartz bridges rather than metal
suspensions, thus reducing mounting stress. - the BVA is "electrodeless", as the electrodes are evaporated on the
inner side of the upper and lower mounting plates. Therefore no stress
between the quartz surface and the metal eletrode. - Therefore the energizing electrical field is applied through an
airgap, which represents two small load capacitors in series to the
resonator, thus making the resonator electrically "stiffer" and less
sensitive to circuit influences.
On the other side you may imagine the main problems associated with such
a construction:
- The difficulty to manufacture the convex and concave shaped parts with
such a precision, that the curvature yields a constant and very small
airgap. - To realize the fine adjustment to frequency, because the unit cannot
be tuned in the conventional way, i.e. by plating some metal on the
electrode. - The frequency accuracy to which the resonator has to manufactured,
because the resonator frequency cannot be pulled with the external
circuit elements by more than some hundred ppb.
It may be interesting to note, that there was a company "BVA
Industries", which wanted to generate their income solely from making
BVA - which failed. Maybe because of the cost could not be covered by
the revenues from the rather limited market.
Bernd DK1AG
AXTAL GmbH & Co. KG
Facility MOS
Wasemweg 5
D-74821 Mosbach / Germany
fon: +49 (6261) 939834
fax: +49 (6261) 939836
www.axtal.com
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
We did a lot of characterization of crystal temp
vs freq in the E1938 development and never observed
any "aging" of these curves. Even on "green" crystals
with zero run time.
Rick Karlquist N6RK
Jeff Mock wrote:
Thanks for the description, it is very interesting. I have a follow up
question if you don't mind.
How does crystal aging look on a graph of temp versus frequency. This
graph has some temperature point where the slope of frequency variation
goes to zero and the crystal is quite stable around this temperature.
What does this graph look like as a crystal ages? Does the optimal
operating temperature change over time, that is, does the graph tend to
move left and right, or does aging tend to move the graph vertically
(the optimal freq stays the same, but the optimal operating temperature
changes as a result of aging).
I hope this makes sense...
jeff
Jeff Mock wrote:
How does crystal aging look on a graph of temp versus frequency....
What does this graph look like as a crystal ages? Does the optimal
operating temperature change over time, that is, does the graph tend to
move left and right, or does aging tend to move the graph vertically
(the optimal freq stays the same, but the optimal operating temperature
changes as a result of aging).
I can confirm Rick's statement, that there is no noticable effect of
aging on the turnover temperature (TOT).
Looking at it from physics standpoint, frequncy aging is mainly caused
by minor changes of vibrating mass and/or by changes in elastic
properties (both due to various mechanisms). See my earlier thrad on
aging mechanisms.
On the other hand, the turn-over temperature is primarily governed by
the cut angle (in connection with the resonator shape etc.). Mass
loading, i.e. changes of electrode mass, has also an infuence, but this
is much weaker. Roughly stated: a change of the TOT by one degC may be
caused by a variation of the mass loading by an amount eqivalent to a
frequency change of n*1000 ppm or so. This means that usual long term
aging of precision crystals of a few 100 ppb or even one or two ppm may
cause changes of TOT in the range of mK or less, which is
- much smaller than the accuracy of the determination of TOT
- neglictable compared to the impact of the aging of thermistors and of
the frequency determining capacitors and inductors of the sustaining
oscillator stage
Regards
Bernd
DK1AG
In message 475BC4C4.1030107@t-online.de, Bernd T-Online writes:
Jeff Mock wrote:
I can confirm Rick's statement, that there is no noticable effect of
aging on the turnover temperature (TOT).
You got me wondering: How is the TOT determined ? Is it per unit
or is it per batch ? How precisely does an OCXO hold the temperature
on the TOT ?
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Poul-Henning Kamp wrote:
You got me wondering: How is the TOT determined ? Is it per unit
or is it per batch ? How precisely does an OCXO hold the temperature
on the TOT ?
For crystals used in precision OCXO the TOT is measured and recorded per
unit.
There are two main methods: passive and active
In the passive method the crystal is inserted in a test fixture and its
frequency and other parameters are measured using a network analyzer
(state of the art). The crystals with its test fixture is in a precision
temperature chamber, whose temperature is varied in fine steps in the
vicinity of the expected TOT. The TOT is then calculated by curve
fitting of the f(T) data by using a 2nd or 3rd order curve fitting
algorithm, depending on the temperature span. Accuracy and repeatability
are in the range of a (few) tenth of a degC.
In the active method the crystal is operating in its oven, whose
temperature is varied over a certain interval. Evaluation of the TOT by
similar curve fitting as above. Accuracy and reproducibility is about
1/100 degC or slightly better.
The temperature accuracy and stability of an oven dpends strongly on its
construction, and there are several orders of magnitude difference
betewen a simple low-cost DIL14 size OCXO and a high-end OCXO with
sub-ppb stability. THe electronic part of the temperature control is not
the difficult task, the main issue is to have a UNIFORM temperature over
all frequency determining components, which is accurate and stable - and
uniformity needs physical size. Regarding stability over time, the most
critical components are the temperature sensors, mostly thermistors.
Regards
Bernd
DK1AG
Bernd T-Online BNeubig@t-online.de wrote:
I can confirm Rick's statement, that there is no noticable effect of
aging on the turnover temperature (TOT).
Looking at it from physics standpoint, frequncy aging is mainly caused
by minor changes of vibrating mass and/or by changes in elastic
properties (both due to various mechanisms). See my earlier thrad on
aging mechanisms.
On the other hand, the turn-over temperature is primarily governed by
the cut angle (in connection with the resonator shape etc.). Mass
loading, i.e. changes of electrode mass, has also an infuence, but this
is much weaker. Roughly stated: a change of the TOT by one degC may be
caused by a variation of the mass loading by an amount eqivalent to a
frequency change of n*1000 ppm or so. This means that usual long term
aging of precision crystals of a few 100 ppb or even one or two ppm may
cause changes of TOT in the range of mK or less, which is
- much smaller than the accuracy of the determination of TOT
- neglictable compared to the impact of the aging of thermistors and of
the frequency determining capacitors and inductors of the sustaining
oscillator stage
For a while, didn't HP sell temperature probes which were in fact
quartz crystals? Oscillation frequency was converted by some simple
electronics to a temperature, and at the time (60's?) they were
exquisitely convenient for measuring way better than a tenth of a
degree.
Either the frequency drift was negligible or it
was so slow that I don't remember any manual removal of frequency
drift effects.
I'm guessing the probe crystals were some special cut (don't know which!)
which was fairly linear or at least monotonic over the measurement
temperatures. I'm guessing that HP chose a cut which had a very large
tempco such that tempco dominated over any frequency drift.
Tim.
Tim Shoppa escribió:
For a while, didn't HP sell temperature probes which were in fact
quartz crystals? Oscillation frequency was converted by some simple
electronics to a temperature, and at the time (60's?) they were
exquisitely convenient for measuring way better than a tenth of a
degree.
Either the frequency drift was negligible or it
was so slow that I don't remember any manual removal of frequency
drift effects.
At least one model is the 2804A. Not much info about it in the Agilent
web site, but according to the 1986 catalog 'the temperature sensor is a
quartz crystal whose precise angle of cut gives an stable and repeatable
relationship between the resonant frequency and temperature'. But also
is mentioned there that 'The only adjustment necessary to remove effects
of thermal history on the sensor is a simple ice point or triple point
calibration adjustment using the front panel thumbwheel switches'.
Since the ice-point calibration would only be able to remove an offset,
I understand that this is the manual removal of frequency drift effects.
Of course, I suppose that the dritft would be small compared with the
quartz temperature coefficient. Anyway, a 10544 oscillator has a cold
offset that can easily be of 1000Hz, so if at 80 deg. C the offset is
zero, and at 25 deg. C the offset is 1000Hz, you easily have a rough
15Hz/deg C average tempco in that range - and the aging drift for this
oscillator is quite less than that.
Best regards,
Javier, EA1CRB
Javier javier@nebulosa.org wrote:
Tim Shoppa escribi?:
For a while, didn't HP sell temperature probes which were in fact
quartz crystals? Oscillation frequency was converted by some simple
electronics to a temperature, and at the time (60's?) they were
exquisitely convenient for measuring way better than a tenth of a
degree.
Either the frequency drift was negligible or it
was so slow that I don't remember any manual removal of frequency
drift effects.
At least one model is the 2804A. Not much info about it in the Agilent
web site, but according to the 1986 catalog 'the temperature sensor is a
quartz crystal whose precise angle of cut gives an stable and repeatable
relationship between the resonant frequency and temperature'. But also
is mentioned there that 'The only adjustment necessary to remove effects
of thermal history on the sensor is a simple ice point or triple point
calibration adjustment using the front panel thumbwheel switches'.
Since the ice-point calibration would only be able to remove an offset,
I understand that this is the manual removal of frequency drift effects.
Of course, I suppose that the dritft would be small compared with the
quartz temperature coefficient. Anyway, a 10544 oscillator has a cold
offset that can easily be of 1000Hz, so if at 80 deg. C the offset is
zero, and at 25 deg. C the offset is 1000Hz, you easily have a rough
15Hz/deg C average tempco in that range - and the aging drift for this
oscillator is quite less than that.
Very interesting, Javier. I'm guessing the 2804A was a 70's
implementation if it had thumbwheels. Thanks for posting the details
you found!
The unit I remember was not digital in the readout sense - it
worked like a Fluke differential voltmeter, where you dial in some
big rotary switches until you get a null on an analog meter. I
may be confusing a 60's era Fluke temperature probe with the HP
probes though!
Who else would've been building quartz temperature probes in the
60's? Fluke, Beckman, ??? And what cut crystal matches the
need for a huge and mostly linear tempco?
Tim.
Tim Shoppa wrote:
For a while, didn't HP sell temperature probes which were in fact
quartz crystals? Oscillation frequency was converted by some simple
electronics to a temperature, and at the time (60's?) they were
exquisitely convenient for measuring way better than a tenth of a
degree.
Either the frequency drift was negligible or it
was so slow that I don't remember any manual removal of frequency
drift effects.
I'm guessing the probe crystals were some special cut (don't know which!)
which was fairly linear or at least monotonic over the measurement
temperatures. I'm guessing that HP chose a cut which had a very large
tempco such that tempco dominated over any frequency drift.
They did indeed. This was the HP2801A, later followed by the HP2804A.
It uses a so-called LC-cut (Linear Coefficient) quartz crystal sensor,
which is a doubly rotated cut with ultra-linear frequency vs.
temperature characteristic with a slope of 35.4 ppm/K.
For more information see http://hparchive.com/Journals/HPJ-1965-03.pdf
BTW: I proudly own a HP2801A plus two crystal sensor elements. However I
cannot connect them to the instrument, because the 2801A has a special
connector for it. It looks like a smaller version of a BNC connector,
but the bayonet has three "nipples" instead of two. Does anyone on the
list know what kind of connector that is and where to get the
counterpart (plug)?
Regards
Bernd
DK1AG