Development of High-Stability Miniaturized Oven Controlled Crystal

Development of High-Stability Miniaturized Oven
Controlled Crystal Oscillator
Wan-Lin Hsieh*, Chia-Wei Chen, Chen-Ya Weng, Che-Lung Hsu, Sheng-Hsiang Kao, and Chien-Wei Chiang
TXC Corporation
No. 4, Kung Yeh 6th Rd., Ping Cheng Industrial District, Pin Cheng District 32459, Tao Yuan City, Taiwan
*[email protected]
Abstract—This paper reports on the development of a highstability miniaturized Oven Controlled Crystal Oscillator
(OCXO) in the size of 9.7 mm x 7.5 mm based on the analog
oscillation circuit combined with the conventional temperature
sensing circuit using a thermistor and related discrete electrical
components. The finite element method (FEM) is implemented to
optimize the oven stability of the oven structure. As a result, a
highly stable oven performance less than ±1°C variation within
the ambient temperature ranging from -40 to 85°C is both
numerically analyzed and experimentally demonstrated.
Consequently, this result implies that the frequency stability of
the miniaturized OCXO can achieve less than ±20 ppb using an
AT-cut crystal.
Keywords—Miniaturized OCXO, finite element method (FEM)
I.
INTRODUCTION
Over the past few years, small cell solutions have shown
their ability to achieve higher radio density and capacity by
deploying as standalone networks or integrating with
conventional macro cell. In addition, the coverage of next
generation 4G-LTE telecommunication will be highly extended
by using small cell technology including indoor and outdoor
applications. However, for small cell application, the physical
size is the key factor affecting the small cell designers to select
the frequency control component due to the preferential
requirement of the use of a single small package with the cost
effective solutions. As a result, much attention has been paid to
developing the miniaturized Oven Controlled Crystal
Oscillator (OCXO) to meet the stringent system requirement of
small cell applications.
In 2012, Ishii et al reported on the development of the
digital signal processing-oven controlled crystal oscillator
(DSP-OCXO) using an AT cut crystal as temperature sensor,
showing the frequency stability is capable to achieve less ±20
4
Fig. 1. A photograph of the 9.7 mm x 7.5 mm miniaturized OCXO.
Fig. 2. Functional block diagram of the miniaturized OCXO. Thermistor is
used as the temperature sensor.
ppb within the ambient temperature ranging from -40 to 85°C
[1]. However, this digital oven control circuit still suffers from
the small fluctuation in temperature stability [2]. To deal with
this issue, we propose a high-stability miniaturized OCXO in
the size of 9.7 mm x 7.5 mm based on the analog oscillation
circuit combined with the conventional temperature sensing
circuit using a thermistor, as seen the real appearance of the
OCXO in Fig. 1. The functional block diagram of the
miniaturized OCXO is shown in Fig. 2. The conventional
temperature sensing circuit using thermistor as a temperature
sensor is implemented for the temperature control circuit. To
optimize the ovenized structure, thermal analysis simulation
according to the finite element method (FEM) is utilized [3].
Consequently, a highly stable oven performance less than
±1°C variation from -40 to 85°C is both numerically and
experimentally demonstrated.
I.
MODELING OF HEAT TRANSFER
OCXO is able to perform a very high degree of frequency
stability within a wide operating temperature range [4]. This
can be achieved by placing the crystal in a thermally insulated
oven structure with a thermostatically controlled heater
element. In order to model the ovenized structure of OCXO
during the operation, the tree-dimensional simulation of the
time dependent process is carried out.
In the simulation, the governing mechanism for the heat
transfer modeling in OCXO only considers the conduction
effect. Therefore, the general heat conduction equation can be
derived from the first law of thermodynamics as shown in
below:
∙
0,
(1)
Fig. 4. Illustration of the proposed OCXO using double-heater structure.
Fig. 3. The heat source for the thermostatically controlled heater is defined
as a function of time, including the warm-up and steady step.
where [kg/m ] denotes the density, and
[J/(kgK)] is the
heat capacity at constant pressure. Fourier’s Law of conduction
gives:
,
(2)
where k [W/(mK)] is the thermal conductivity. It should be
noted that the convective effect is ignored by setting the
velocity equals to zero in the numerical domain of air.
The heat source for the thermostatically controlled heater
is defined as a function of time as shown in Fig. 3. The initial
temperature of the heat source is 25°C, and the heating process
includes the warm-up step with a heating rate 2.5 °C/s and the
steady step that keeps the temperature at 97.5°C.
Convective heat flux boundary condition is adopted to
represent the outside wall of the stainless steel cover, exposing
to the ambient air:
∙
,
(3)
where n denotes the normal to the wall, is the heat transfer
represents the temperature of the ambient
coefficient, and
is
air. In this paper, the operating temperature range of
from -40 to 85°C.
(around several millimeters) of the device. As a result, this
multi-scale issue may lead to a longer computational time due
to the finer mesh is required. To this end, the highly
conductively layer is implemented in the numerical model by
setting the copper trace as a boundary. It should be mentioned
that this assumption still satisfies the heat transfer formulation
in (1).
The materials and the physical properties used in the
numerical simulation are shown in Table 1, including the
ambient air, printed circuit board (PCB), ceramic for the crystal
and heater package, copper for the trace embedded in the PCB,
and stainless steel for the cover.
II.
DESIGN OF OVENIZED STRUCTURE
To optimize the ovenized structure, the thermal analysis
based on FEM simulation is utilized to investigate the oven
stability. First, we optimize the physical location of the
thermistor and show the effect of the relative position between
thermistor and heater on the oven controlled circuit. Second, a
double-heater structure is proposed to enhance the oven
stability.
The illustration of the proposed OCXO is shown in Fig. 4,
including two heaters (Heater 1 and Heater 2), three PCB
layers, a thermistor, and a TCXO. The passive components,
such as chip resistors and capacitors, are neglected in the
The thickness of the copper trace (around several micro
meters) in the PCB is much thinner than other components
TABLE I.
MATERIALS AND PHYSICAL PROPERTIES USED IN THE
NUMERICAL MODELING.
Material
Density
[kg/ ]
Thermal conductivity
[W/(mK)]
Heat capacity
[J/(kgK)]
Air
1200
0.0257
1003.5
PCB
1900
0.3
1369
Ceramic
3850
31
840
Copper
8700
400
385
Stainless steel
7837
16.9
486
Fig. 5. The physical location of the thermistor is analyzed to optimize the
oven structure. The simulation results show that the temperature variation of
the thermistor is less than 0.3°C in the ambient temperature from -40 to 85°C.
The insets represent the temperature distribution when ambient temperature is
-40°C before and after optimization, respectively.
simulation herein. When thermistor is used as a temperature
sensor, an accurately temperature detection is critical for the
feedback loop of the oven control circuit. For example, when
the thermistor does not completely contact with the heater, as
shown in the inset in Fig. 5, the temperature of the thermistor is
only 85°C when the heat source of heater is set at 97.5°C while
the ambient temperature is -40°C. As a result, over 8°C
variation of thermistor is obtained (solid line shown in Fig. 5)
in the ambient temperature from -40 to 85°C. This large
variation indicates that an overheating to the crystal may occur
at the low ambient temperature condition since the temperature
of thermistor is much lower than the reference temperature set
from the feedback loop of the oven control circuit. In other
words, this result implies that the physical distance between a
crystal and a temperature sensor should be carefully studied to
decide the most proper location. To this end, we parametrically
analyze the physical location of the thermistor. We found that
when the thermistor is completely contact with Heater 1, as
seen in Fig. 4 (dashed line), the temperature variation of the
thermistor less than 0.3°C variation could be achieved.
Fig. 6. (a) The thermal distribution of the proposed OCXO obtained by
simulation is presented when the ambient temperature is 85°C. The doubleheater structure is utilized to improve the oven stability. (b) Comparison of
crystal temperature between the one-heater and double-heater structure. (c)
Temperature increment (ΔT) using double-heater structure compared to the
one-heater structure.
Next, the effect of the additional heater (Heater 2) on the
thermal efficiency in the ovenized structure is investigated. It is
observed that a huge heat loss from Heater 1 to PCB 1 could be
found when Heater 2 is removed as seen in the thermal
distribution obtained by the simulation in Fig. 6(a). As seen in
Fig. 6 (b), a comparison of crystal temperature between the
one-heater and double-heater structure is presented. When
Heater 2 provides an additional heat source, the thermal
distribution in the oven is more uniform, where the temperature
in the steady step of Heater 1 and Heater 2 is set at 97.5°C
while the ambient temperature is 85°C. To quantify the
improvement of using double-heater structure, the
parametrically analysis by changing the temperature of Heater
2 while keeping Heater 1 at 97.5°C is studied. The result in Fig.
6(c) shows the temperature increment (ΔT) using double-heater
structure compared to the one-heater structure, where over
1.6°C is improved when Heater 2 is set at 97.5 ̊C.
In consequence, considering the double-heater structure
the oven stability of the crystal less than ±0.1°C is successfully
achieved as seen in the simulation result in Fig. 7 while the
external ambient temperature ranging from -40 to 85°C.
III.
Fig. 7. Temperature stability of the crystal referred to 25°C in the ambient
temperature ranging from -40 to 85°C obtained by simulation and experiment.
EXPERIMENTAL RESULTS
The ovenized structure has been parametrically optimized
by analyzing the physical location of thermistor related to the
heater and considering the double-heater concept used in the
device, the next step is to realize the miniaturized OCXO. As
shown in Fig. 7, the oven stability less than ±1°C variation
through experimental measurement is obtained. It should be
noted that the temperature of crystal is measured using a AC
cut blank as a temperature sensor, which has a great linearity
between frequency and temperature. Consequently, this result
implies that the frequency stability is possible to achieve less
than ±20 ppb using an AT-cut crystal. Fig. 8 shows the
corresponding output current when the ambient temperature is
25°C, from which the current of warm-up step and steady step
is 320 mA and 150 mA, respectively. In addition, the
corresponding thermal image captured by the infrared thermal
Fig. 9. Thermal distribution captured by the infrared thermal imaging
equipment. The temperature of Heater 2 is around 95°C.
Fig. 8. The output current when the ambient temperature is 25°C, where the
current of warm-up step and steady step is 320 mA and 150 mA, respectively.
REFERENCES
[1]
imaging equipment is shown in Fig. 9. It is observed that the
temperature of Heater 2 is around 95°C.
IV.
CONCLUSTION
In this paper, we develop a high-stability miniaturized
OCXO in the size of 9.7 mm x 7.5 mm. Thermal analysis
according to the FEM simulation is utilized to optimize the
ovenized structure. It has shown that the proposed doubleheater structure could enhance the oven stability. Therefore,
the experimental result shows a highly stable oven
performance less than ±1°C variation from -40 to 85°C,
indicating that a high frequency stability less than ± 20 ppb
can be ahieved if an AT-cut crystal is used.
[2]
[3]
[4]
Yasuto Ishii, Kaoru Kobayashi, Tsukasa Kobata, Manabu Ito, Shigenon
Watanabe, Shinichi Sato, Kazuo Akaike “A New Generation DSPOCXO Using Crystal Temperature Sensor,” International Frequency
Control Symposium, IEEE, pp.341-344, 2012.
Kaoru Kobayashi, Yoshiaki Mori, Tsukasa Kobata, Manabu Ito,
Shigenori Watanabe, Shinichi Sato, Kazuo Akaike, “High-Performance
DSP-TCXO Using Twin-Crystal Oscillator,” International Frequency
Control Symposium, IEEE, pp.1-4, 2014.
COMSOL Multiphysics Version 4.3a User Guide, October, 2012.
Vig, John R. “Introduction to quartz frequency standards,“ No. SLCETTR-92-1. ARMY LAB COMMAND FORT MONMOUTH NJ
ELECTRONICS TECHNOLOGY AND DEVICES LAB, 1992.