A Low Cost, Cavity Stabilized 5.8 GHz Oscillator Realized in LTCC
A cost-effective, highly stable 5.8 GHz FET oscillator realized in LTCC with an integral cavity resonator
To achieve high stability and low cost, a 5.8 GHz FET oscillator was realized in the low temperature, cofired ceramic (LTCC) medium, with an integral cavity resonator. The oscillator exhibits an average 0.23 MHz/°C drift over a temperature range of 30° to 60°C and < 5 MHz/V pulling at 26°C.
The oscillator described in this article was developed for automotive radar. Such systems require good stability and environmental ruggedness yet must be produced at low cost. Manual tuning in manufacture is a costly process, and therefore is unacceptable for circuits used in automotive applications. However, simple approaches to oscillator design, such as the use of microstrip or stripline resonators, do not provide adequate stability to guarantee that the oscillator frequency remains within the required limits over the wide range of environmental conditions (especially temperature) that automotive circuits experience. These methods also require expensive housings for mechanical robustness. Dielectric resonators would provide adequate stability but usually require expensive housings and manual tuning, unacceptably increasing the cost. Mechanical resonators might provide good stability with minimal manual tuning, but their size and cost are still too great.
A solution to this dilemma is to realize the oscillator and a cavity resonator simultaneously in LTCC, a multilayer ceramic medium. Horizontal top and bottom plates of the resonator are realized as solid metal; the vertical walls must be realized as a row of closely spaced vias. In this way, a cavity resonator is realized simultaneously with the oscillator, and no manual tuning is required.
As well as offering good stability, this structure has a number of advantages. All the conductors (except for a few contact pads) are buried in the ceramic, so they are impervious to moisture and dirt, and cannot corrode. The circuit's RF ground plane is likewise inside the ceramic, so ground problems are minimized. Because the resonator is shielded, unlike a microstrip circuit, the housing has no significant effect on the oscillator's frequency. Finally, the structure is mechanically stable, so a rigid housing is not necessary.
The resonator is designed by means of a planar electromagnetic simulator EMSight, part of the Microwave Office suite.1 In the initial design, the resonator is modeled as a simple square box, with solid walls, operated in the TE110 mode, excited by a vertical probe. Although it would be possible to model the via walls with the EM simulator, it would require significantly more computation time, which would not be justified by the minor improvement in accuracy.
The EM simulator is used to adjust the location of the probe to optimize the coupling. If the probe is too close to the center, the loaded Q is low; if it is too near the edge, strange resonances can appear. The best location, in this case, was found to be approximately halfway from the center to one wall. The resonator can then be modeled accurately as a simple, parallel RLC circuit.
Because it is impossible to create solid vertical metal walls in LTCC, the resonator's walls are realized by a row of vias, which extend through several layers in the LTCC. In general, it is bad practice to stack vias on top of one another. This causes a build-up of the natural "dome" on top of each via, and degrades the layers' flatness. Thus, the locations of the vias must be staggered, and metallization strips used between the via rows. The cavity's probe is realized similarly, by staggered vias with catch pads at each layer; this structure is necessitated by the LTCC fabrication-process rules. Because it is impossible to show this structure in its final form, the three-dimensional layout visualization is shown in Figure 1. The ground plane layer is not shown. RF conductors are buried microstrip lines. A detail of the resonator's probe is shown in Figure 2.
The ceramic used for the circuit is DuPont 951.2 This material has a dielectric constant of 7.4, a loss tangent of 0.007 and a thermal expansion coefficient of approximately 5 106 . The dielectric constant is somewhat process-dependent; the dielectric loss is somewhat greater than that of alumina, which has a loss tangent of ~0.001. As with most ceramics, thermal conductivity is high, 3.0 WM1 K1 , allowing for good conductive cooling of the FET. The high frequency resistivity of the buried silver metallization is difficult to determine, as are the leakage and dissipation losses in the via walls. The metallization resistance was estimated from transmission-line loss measurements. Including the dissipation losses in both the dielectric and metallizations, an estimate of ~150 for the Q is obtained. This is obviously much lower than would be achieved with either a metal or dielectric resonator; however, it is much better than a microstrip stub.
One characteristic of LTCC is that, to prevent corrosion, the silver RF conductors are not located on the top surface of the substrate but on a layer below the surface. Thus, all RF conductors are either buried microstrip lines or, if they are located between ground planes, striplines. Striplines can be modeled easily by conventional models; however, most circuit simulators do not include buried microstrip models. If the circuit simulator does not include such models, the transmission lines must be treated as general TEM lines, and their impedances must be calculated by some other means. Unfortunately, most transmission-line calculators do not include buried microstrip; one that does is a versatile transmission-line analysis program, LINPAR3 , which can analyze strip conductors on arbitrary, layered dielectric structures. LTCC designers should also note that the effective dielectric constant of buried microstrip is much lower than in conventional microstrip; in fact, even with a thin top layer, it is nearly equal to the bulk dielectric constant of the LTCC ceramic.
The oscillator was designed in the conventional manner, using small-signal methods. These are well known and previously published.47 It was simulated by the harmonic-balance circuit simulator in Microwave Office. The FET used in the circuit is an NEC72218 low cost plastic-packaged MESFET; the device model was supplied by the manufacturer. External capacitors provide DC bias bypassing and blocking, and a chip resistor is used for gate bias; however, it could be integrated into the LTCC circuit if desired. The finished circuit is shown in Figure 3.
The predicted output power is 17.9 dBm at 3 V DC drain bias, which agrees reasonably well with the 16.5 dBm measured. Predicted and measured output power at 4 V are 20.2 and 18.2 dBm, respectively. The FET model from NEC, used in the simulations, provided adequate accuracy in predicting the oscillator's performance. The design frequency was 5.8 GHz; at 30°C the frequency was 5.755 GHz. No manual tuning was required.
Figure 4 shows the oscillator frequency as a function of temperature from 30° to 60°C. The frequency varies approximately linearly with temperature, averaging 0.23 MHz/°C over this range.
The frequency, as a function of DC drain bias, is shown in Figure 5. The shift in frequency of approximately 5 MHz compared to the frequency vs. temperature performance is caused by differences in load impedance. The maximum pulling is approximately 5 MHz/V. DC drain current, at 4 V drain bias, is 41.4 mA.
The use of an LTCC rectangular resonator is a practical way to realize a low cost oscillator. Good stability is achieved. No manual tuning or adjustment is needed. *
1. Applied Wave Research Inc., 1960 E. Grand Ave. Suite 500, El Segundo CA 90245, USA.
2. E. I. du Pont de Nemours and Co.; see http://www.dupont.com/mcm/gtapesys.
3. A. Djordjevic, LINPAR32: Parameters for Multiconductor Transmission Lines, Artech House, Norwood, MA, 1998.
4. E. Holzman and R.S. Robertson, Solid-state Microwave Power Oscillator Design, Artech House, Norwood, MA, 1992.
5. R.W. Rhea, Oscillator Design and Computer Simulation, McGraw-Hill, New York, 1995.
6. A.P.S. Khanna, "Oscillators," (Ch. 10), in D. Kajfez and P. Guilon, Dielectric Resonators, Artech House, Norwood, MA, 1986.
7. S. Maas, Nonlinear Microwave Circuits, IEEE Press, New York, 1997.
Stephen Maas received BSEE and MSEE degrees in electrical engineering from the University of Pennsylvania in 1971 and 1972, respectively, and a PhD in electrical engineering from UCLA in 1984. Since then, he has been involved in research, design and development of low noise and nonlinear microwave circuits and systems at the National Radio Astronomy Laboratory Observatory (where he designed the receivers for the Very Large Array), Hughes Aircraft Co., TRW, the Aerospace Corp. and the UCLA department of electrical engineering. He is now president and principal consultant of Nonlinear Technologies Inc.; recently he became Chief Technology Officer of Applied Wave Research Inc. Dr. Maas is the author of Microwave Mixers, Nonlinear Microwave Circuits and The RF and Microwave Circuit Design Cookbook. From 1990 until 1992 he was the editor of IEEE Transactions on Microwave Theory and Techniques and from 19901993 was an Adcom member and Publications Chairman of the IEEE MTT Society. He received the Microwave Prize in 1989 for his work on distortion in diode mixers. He is a fellow of the IEEE.