Introduction
A novel, broadband quasi-optical frequency multiplier system has been
designed and simulated for high efficiency, high power millimeter wave
applications. Initial proof-of-principle doubler and tripler systems have been
developed using a conventional 10 W Ka-Band TWT as the driver.
The eventual goal of this research is a high power source using two microwave
power module (MPM) drivers (200 W total) and three frequency multiplier arrays.
This is predicted to produce > 40 W cw output power at V-Band and > 30 W cw
output power at W-Band, and to offer dramatic savings in size, weight and cost
as compared to conventional coupled-cavity TWT sources. Eventual systems
could also employ the 18-40 GHz MMPM currently under development
[Led98].
Motivation
Major advances in miniature TWTs, advanced power supplies,
electronics materials technology, solid state electronics,
millimeter wave integrated circuit (MMIC) technology, and quasi-optics promise
a new generation of high-frequency devices and
systems to solve critical problems in radar, remote sensing,
communications, plasma diagnostics, radio astronomy, atmospheric
radiometry, and imaging. In the microwave region ( 2-18 GHz ),
a combination of new technologies, such as MMIC amplifiers and
equalizers together with miniature TWTs have resulted in the
microwave power module (MPM) which produces record power densities
at low cost thereby offering exciting opportunities in
communications, ECM and phased array radars. Although there are
efforts underway to directly extend this approach to the
millimeter wave region, it is worthwhile to consider alternative
solid state/ vacuum electronics hybrids. One such approach is to
frequency multiply the output of the MPM using nonlinear solid
state devices and quasi-optical power combining. The technique of
quasi-optical spatial power combining of the output of large
planar grid arrays of solid state devices avoids the Ohmic losses and
limitations associated with conventional power combining techniques and
has been shown to provide high output power levels (up to 5 W at 99 GHz
in initial proof-of-principle experiments)
[Jou88, Qin93, Liu93].
Our projections for the performance potential are shown in Fig. 1 where
the average power limitations of current solid state and vacuum beam
sources are also displayed, for purposes of comparison.
Fig. 1. State-of-the-art for solid state devices and vacuum
electronic beam tubes in the microwave to submillimeter wave
region [after R.Parker, NRL].
Design Overview
A
quasi-optical frequency multiplier grid array consists of thousands of
varactor devices and antennas on a single wafer. For the tripler doubler
grid array, the diodes are configured in a back-to-back layout. The frequency
doubler contains biased Schottky diodes. For both arrays vertical metal
strips serve as antennas. This grid array approach is extremely attractive
because of its low fabrication cost, small-size realization and graceful
degradation qualities. The concept of monolithic Watt-level diode-grid
frequency tripler arrays was implemented by our group in
initial proof-of-principle experiments with multi-quantum-barrier
varactor (MQBV) and Schottky quantum barrier varactor (SQBV) grids
fabricated on a GaAs substrate
[Liu93]
where an output power at 99 GHz of 5 W from the SQBV array have been
achieved. This provides a measure of the potential of this new technology
for high power, high frequency application.
Overmoded Waveguide Design
A new approach has been taken in the development of frequency multiplier grid
arrays. In previous work, the grid arrays were placed in the far field of an
antenna. The multiplied output power was sampled using a second antenna.
This technique is excellent for exploring the physics of the diode grid arrays,
however, it has significant drawbacks as a practical system. The finite grid
array size and multiple reflections introduce significant diffraction loss
[Liu93]. Diffraction losses
occur when the finite diameter incident wave strikes the array substrate and
begins to spread due to diffraction effects. Although this causes no
significant difficulties near the center of the array, near the edges of the
array the diffracted signal moves enough laterally with each reflection such
that it "walks off" the array losing appreciable amounts of power out the
sides of the array substrate.
If the frequency multiplier system is placed in the far field of the input antenna, in the absence of focusing optics, only a fraction of the transmitted power is received by the multiplier. This can cause losses of 20% or more to the input power. Although this can be corrected by a lens system, the lenses needed add significantly to the size and weight. Additionally, this technique under utilizes the array, leaving area at the edges that is unpumped and unused. Since the array is usually the most expensive component of the frequency multiplier grid array system, this can be a costly choice.
Fig. 2. Overmoded waveguide structure.
Many of the difficulties with free-space quasi-optical frequency multiplier grid array systems can be resolved by placing the quasi-optical system inside a waveguide as shown in Fig. 2. In such a system, all fundamental input power is delivered to the array and all harmonic power generated is delivered to the output waveguide without the need for focusing optics. This appreciably reduces the size and weight of the system. In addition, the waveguide walls eliminate the diffraction walk-off losses. Finally, the waveguide fixture provides for a convenient mounting structure for the elements in the nonlinear quasi-optical circuit.
Simulations
Extensive
system and device simulation studies have also been
performed. The HP HFSS and Ansoft Maxwell 3D EM simulation codes
have been employed to perform layout level optimization. An
accurate simulation model for quasi-optical diode grid arrays has
been developed using the actual device and antenna geometries
thereby properly accounting for parasitics. This approach has been shown
to provide an excellent description of other quasi-optical grid arrays
such as a high performance 60 GHz Schottky diode beam controller
[Jia98, Sjo95]. The HP MDS and EESOF
Libra software packages have been employed to perform harmonic
balance simulation. For a Ka-W band frequency tripler, > 35% efficiency
and a broadband (27 GHz - 37 GHz input at the 3 dB points) frequency
tripler is predicted with our optimized device and circuit parameters.
These simulations assume a uniform power per unit cell across
the entire surface of the array. However, since the frequency multiplier
array will actually be placed in an overmoded waveguide, the sine squared
dependence of rf power across the wide dimension of a waveguide will result
in a slightly lower power and efficiency than these first simulations predict.
Additional simulations have therefore been run to predict the output power and
efficiency across the overmoded waveguide taking into account this variation
in drive power and associated diode conversion efficiency (see Fig. 3).
Using these simulations, > 30% efficiency and > 3 W output power are
predicted for the overmoded waveguide 10 W Ka- to W-band frequency
tripler.
Fig. 3. Power and efficiency per unit cell across the overmoded
waveguide.
Experimental Results
The
overmoded waveguide approach has been explored with two proof-of-principle systems.
In these systems, a 10W Ka-Band TWT was used as a driver. Figures 4 and 5 show the results of these experiments. Small arrays were used to experimentally explore the physics of overmoded waveguide grid arrays. This resulted in lower output powers than previous far-field experiments. However, higher power levels per diode were achieved with the frequency doubler and higher efficiencies were achieved with the frequency tripler than in previous far-field experiments.
 |
 |
Fig. 4. SQBV based frequency tripler grid array results at 96 GHz. |
Fig. 5. Schottky diode based frequency doubler grid array at 66 GHz. |
- 76.5 mW output power at 96 GHz
- 2.36 % Efficiency
- 250 devices
- 0.3 mW per device
|
- 410 mW output power at 66 GHz
- 6.26 % efficiency
- 56 devices
- 7.3 mW per device
|
Work in Progress
Frequency Doubler Grid Arrays
A new fixture is being fabricated which will require only a single bias connection to the array, a significant reduction from present fixture which requires n+1 bias lines for an
n x m array. This is expected to improve efficiency by a factor of 5 to 8. A
collaboration with the Army Research Laboratory has recently been established wherein they
are fabricating a series of GaAs wafers using their advanced fabrication facilities.
Their processes will provide Schottky diodes with a capacitance ratio, Cmax/Cmin = 6 and a cutoff frequency, fc = 630 GHz.
This is a significant improvement over the Cmax/Cmin = 2.3 and fc = 282 GHz values of the BNN varactor devices. Simulations predict > 40% efficiency at V-band, an increase by a factor of five. The first of the University of Maryland wafers will provide seven different designs of frequency doubler grid arrays.
In addition to providing increased efficiency, these designs will be used to advance the understanding of grid array physics by studying the effects of grid dimensions, including varying width to height ratios. The next wafer will include grid arrays designed to be driven by several TWT amplifiers available to this research group, including two 80 W TWTs, at Ka- and W-bands, and a 44 GHz, 250 W Milstar CCTWT.
Frequency Tripler Grid Arrays
The current high power, high efficiency frequency tripler development
activity consists of two phases. In Phase One, a single stage
tripler array will be employed to triple the 80 W cw output from a
Ka-band TWTA into W-band with a 94 GHz center frequency, 1% instantaneous bandwidth,
and 10% tunable output bandwidth, > 15% efficiency, and > 5 W output power.
In Phase Two, a two stage frequency multiplier system with > 15% overall efficiency
will be driven by a X-band source to produce a 100 W cw W-band source with a
94 GHz center frequency, 1% instantaneous bandwidth and 10% tunable bandwidth.
We predict a 30 W W-band source weighing only 7 lb and occupying
only 70 in3 with a unit cost in quantities of 1000 of approximately $40 k.
It should further be noted that using recently published long term cost
estimates by Hamilton [Ham95], the above
number is reduced to $18 k. Table 1 provides a comparison of the
combined MPM/frequency tripler grid array approach with the conventional
coupled-cavity TWT approach at
W-band. It illustrates the strong gains that may be made in the
areas of weight, volume, efficiency and cost through the use of
frequency tripler grid arrays. We also note that a single stage W-Band
frequency multiplier could be made with the 18-40 GHz MPM currently under
development [Led98] at even greater cost
savings.
Table 1. Preliminary W-Band Source Comparison Chart.
| Approach | Conventional Coupled Cavity TWT | MPM ( x 2) | Tripler Grid Array ( x 2) | MPM/Tripler Grid Array Combination |
| Basis | Hughes 982H | Varian VZM-6192P1 | SSQBV Array (2000 devices) x2 | ----- |
| Frequency | 93-95 GHz | 9-16 GHz | 81-110 GHz | 81-110 GHz |
| Peak Power | 50 W | 100 W x 2 | 15 W x 2 | 30 W |
| Average Power | 50 W | 100 W x 2 | 15 W x 2 | 30 W |
| Gain | 50 dB | 50 dB | -8 dB | 42 dB |
| Total Efficiency | 11 % | 23 % | 16.5 % | 3.8 % |
| Weight (including P.S.) | 10 lb (70 lb) | 2-3 lb x 2 (same) | 1-2 lb x 2 (same) | 7 lb (same) |
| Volume (including P.S.) | 304 in3 (3000 in3) | 29 in3 x 2 (same) | 6 in3 x 2 (same) | 70 in3 (same) |
| Cost (1's) (including P.S.) | $350 k | $45 k x 2 | $17 k x 2 | $124 k |
| Cost (100's) (including P.S.) | $225 k | $25 k x 2 | $16 k x 2 | $82 k |
| Cost (1000's) (including P.S.) | ----- | $15 k x 2 | $15 k x 2 | $40 k |
| Availability | Now | Now | 1999 | 1999-2000 |
The links in the above document are listed below for your convenience:
For additional information on high power frequency multipliers or the
high power, high frequency source project, please refer to the publications
listed on the references page or contact
Steven Rosenau directly via email:
rosenau@ece.ucdavis.edu
UCD Plasma Diagnosics & Millimeter Wave Technology - ATRI Research Projects
UCD Plasma Diagnosics & Millimeter Wave Technology Home Page
Revised October 8, 1998 -
rosenau@ece.ucdavis.edu
