Patent Application
20090273517
1. Field of the Invention
This invention relates generally to a vertically integrated electronically steered phased array and, more particularly, to a vertically integrated electronically steered phased array that is synchronized to a global clock signal applied to a local oscillator from a tunable phase-locked loop (PLL) in each channel of the array.
2. Discussion of the Related Art
Transceiver arrays are widely used in wireless communications, radar applications and sonar applications. A transceiver array is an array of transceiver channels each including an antenna where the channels combine to provide a directional beam for both transmitting and receiving purposes, including beam scanning. As the directivity of the array increases, the gain of the array also increases.
Various types of transceiver arrays are known in the art that provide beam steering. One known transceiver array type includes mounting individual transceiver front-end channels on a mechanical device that moves to provide beam steering or scanning.
Another known transceiver array type is a phased array. A typical phased array includes an antenna in each channel that is connected to a phase shifter, and a power combiner for adding the signals together from the antennas. The phase shifters control either the phase of the excitation current of the antenna for transmission or the phase of the receive signals. When the signals are combined, a beam is formed in a particular direction. Particularly, a transmit beam is formed in space, and a receive beam adds coherently if the signals are received from a particular region of space. The radiation pattern of the transceiver array is determined by the amplitude and phase of the current at each of the antennas. If only the phase of the signals is changed and the amplitude of the signals is fixed, the beam can be steered.
Another type of transceiver array employs digital beamforming systems have been developed in the art that eliminate the need for the phase shifters to provide beam steering. The digital beamforming systems digitally provide beam steering. One advantage of digital beamforming is that once the RF information from each channel is captured in the form of a digital stream, digital signal processing techniques and algorithms can be used to process the data in the spatial domain.
Digital beamforming is based on the conversion of the RF signal at each antenna into base-band signals. The beamforming is provided by weighting each digital signal from the channels, thereby adjusting their amplitude and phase so that when they are subsequently added together in post-processing they form the desired beam. Thus, the linear phase weight applied to the digitized signal at each channel can make the antenna beam appear as if it is steered to different angular directions.
In accordance with the teachings of the present invention, a vertically integrated electronically steered phased array is disclosed that employs beamsteering using a programmable phase locked loop including a local oscillator. The local oscillator provides an oscillator signal that is converted to an RF signal that can be either up-converted for a transmit operation or down-converted for a receive operation. The relative off-set between independently generated local oscillator signals forms the basis of the off-set phase required for a phased array. The absolute measure of off-set phase is referenced to a globally distributed clock signal that aligns the zero degree phase shift of the oscillator.
Additional features of the present invention will become apparent from the following description and appended claims, taken in conjunction with the accompanying drawings.
a) and 4(b) are a top perspective view and a bottom perspective view, respectively, of a filled antenna horn and RF circuitry applicable to be used in the vertically integrated electronically steered phased array shown in
a) and 5(b) are a top perspective view and a bottom perspective view, respectively, of an antenna and related circuit that can be used in the vertically integrated electronically steered phased array shown in
The following discussion of the embodiments of the invention directed towards a vertically integrated phased array is merely exemplary in nature, and is in no way intended to limit the invention or its applications or uses.
A controller 30 provides beamforming and signal modulation and demodulation for the signals transmitted by each transmit channel 12 and the signals received by each receive channel 14. The phased array 10 can operate as a local oscillator (LO) phase steered array or a digital beamformer where the phase control is provided in the digital beamforming controller 30, as is well understood by those skilled in the art. A signal to be transmitted is sent to an in-phase splitter 32 and a quadrature phase splitter 34 that provide signals 90° apart in phase. The splitters 32 and 34 provide the in-phase signals and the quadrature phase signals to the appropriate in-phase portion 16 and quadrature phase portion 18 in each of the transmit channels 12. Each of the in-phase portions 16 and the quadrature phase portions 18 in each of the transmit channels 12 includes a digital-to-analog converter 36 that converts the digital signals to analog signals for transmission. The analog signal is then sent to an optical attenuator 50 if taper or balance is required and a high-pass filter 38 that filters the analog signal to a certain frequency band. The filtered analog signal is then sent to a mixer 40 along with a local oscillator signal from a local oscillator in a phase locked loop (PLL) 42 to up-convert the signal to a high frequency signal for transmission. A detailed schematic of the PLL 42 is not necessary for a proper understanding of the invention in that such circuit diagrams are well known to those skilled in the art.
As will be discussed in detail below, the PLL 42 is a programmable PLL that receives a clock signal that aligns the zero phase of the local oscillator signal. Control data is provided that establishes the phase off-set of the local oscillator with the edge of the clock signal. These control data sets the relative phase of the local oscillator in the PLL 42, and thus, controls the phase of the signal transmitted by the particular transmit channel 12 to provide beam steering in the phased array 10. The PLL 42 generates a frequency in the VCO and the digital data conditions the negative feedback loop of the PLL 42 to provide the off-set. It is the off-set with respect to the edge of the clock signal that sets the relative phase in the PLL 42 that can be changed to time shift the local oscillator signal to provide phase control. The PLLs 42 can be any PLL suitable for the purposes described herein, such as the PLL disclosed in U.S. Patent application publication 2006/0164132 to Martin et al. The high frequency signal may then be band-pass filtered by a filter 44 and amplified by power amplifiers 46 and 48 to be transmitted by the antenna 24.
When the transmit/receive switch 26 is switched to the receive mode, signals received by the antenna 24 are amplified by a low noise amplifier (LNA) 52, and then separated into the in-phase portion 20 and the quadrature phase portion 22 of the receive channel 14. In both of the in-phase portion 20 and the quadrature phase portion 22, the received high frequency signal is sent to a mixer 54 that also receives a signal from a local oscillator in a PLL 56 to provide frequency down-conversion to an intermediate frequency analog signal. As with the transmit channel, the local oscillator in the PLL 56 receives a global clock signal for phase alignment and a programmable off-set is used for phase control. The intermediate frequency signals may be band-pass filtered by a band-pass filter 58, and may further be attenuated by an attenuator 64, and amplified by an amplifier 60 before being sent to an analog-to-digital converter 62 where the intermediate frequency analog signals are converted to digital signals. The digital signals in the in-phase portion 20 of the receive channel 14 are sent to an in-phase summer 66 where all of the in-phase signals from all of the receive channels 14 are added. Likewise, the receive signals in the quadrature phase portion 22 are sent to a quadrature phase summer 68. The added in-phase signals and the quadrature phase signals are then sent to the controller 30.
In another embodiment, the signals may be added at the IF level using a conventional corporate distribution network that are then used in any manner known to those skilled in the art.
As will be discussed in detail below, the phased array 10 includes various integrated components that accurately control the phase applied to the local oscillators and circuit components for providing accurate timing to synchronize the independent local oscillators used to create the phase shift.
For the cavity backed patch antenna embodiment, a metalized plane 86 including coupling slots 96 is formed on a top surface 88 of the interconnect layer 74 and acts as a ground plane for the patch antenna elements 78. A bottom surface 90 of the interconnect layer 74 includes a plurality of circuit components 92 fabricated thereon, and can include, for example, the PLLs 42 and 56, the mixers 40 and 54, the band-pass filters 44 and 58, the attenuators 50 and 64, the power amplifiers 46 and 48, the LNAs 52 and the amplifiers 60 in the phased array 10. Digital components 94 of the phased array 70 can be integrated or formed on the packaging layer 76, such as the digital-to-analog converters 36, the analog-to-digital converters 62, the band-pass filters 38, etc.
Each of the layers 72, 74 and 76 can be fabricated from a variety of different materials, such as silicon, organic substrates, such as liquid crystal polymers and/or parylenes, low-temperature co-fired ceramics (LTCC), and any microwave substrate that is suitable and capable of producing multi-layer structures. Further, conventional multi-layer printed circuit boards (PCB) can be used for the layers 72, 74 and 76. All of the necessary active components are mounted on the appropriate layer and are electrically connected to the circuits using any suitable technique, such as flip-chip connections or wire bonds. Further, the antennas, distribution networks and DC bias networks can be printed on the appropriate layer using a fabrication technology appropriate for each type of substrate. For example, screen printing can be used to fabricate metal lines on LTCC, photolithography can be used for fabricating metal structures on silicon, parylene and LCP. Vias may be formed through silicon layers using dry reactive ion etching (DRIE), vias can be formed in the PCB material either by mechanical or laser drilling, and vias can be formed in the green LTCC stack by drilling or die cutting. Such layering techniques can be used to form virtually any structure required to provide the entire RF system from the antenna to the active component interconnects and active component packaging.
The phased arrays 10 and 70 synchronize each active antenna element to a single global clock to perform digital phase-shifting on a locally generated LO signal. This removes the necessity for precise high-frequency distribution networks. The phased arrays 10 and 70 do not need to perform power combining or distribution at high-frequency, thereby achieving high watt/watt efficiency by not dissipating costly RF signals. Unlike RF distributed systems that use multi-stage amplification to compensate for signal loss in phase-shifters and variable attenuators, the phased arrays 10 and 70 produce all RF signals directly at each element. Each element operates as an independent radio channel, combining with robust IF-signal combining and distribution networks, the array is highly tolerant of multi-element failure. Because the high frequency distribution networks have been removed, the phased arrays 10 and 70 are well-suited to compact multi-layer PCB fabrication, allowing smaller system size and weight with higher mechanical stability when compared with single-layer PCB or silicon-based arrays.
As the frequency of operation and the degree of scan angle goes up, the ability to use patch antennas in the PCB architecture of the phased arrays 10 and 70 is reduced. Normally, the RFICs are mounted in internal cavities directly behind a planar element, but at higher frequencies, the element spacing becomes prohibitively small and a sub-array architecture may not be able to be used for wide scan angles. For these applications, the antenna element architecture needs to provide millimeter-wave efficiency, low-directivity, low-coupling, appropriate polarization, be economic to produce and be easily integratable with a PC board.
a) and 4(b) are a top perspective view and a bottom perspective view, respectively, of a filled-waveguide antenna horn 320 that provides the properties discussed above. The horn 320 includes a horn element 322 including an expanded portion 324 and a base portion 326. The expanded portion 324 is filled with a suitable dielectric that operates to give the antenna horn 320 low-directivity and efficiency. The filled horn 320 provides the λ/2 spacing required for the array and maintains low directivity. The dielectric material should be castable, have low-moisture absorption and be impact resistant. The horn element 322 can be a metal plated material having desired impact properties. The horn element 322 can be cast independently and plated individually, or a plated mold can be provided that will be the surface of the element and the dielectric can be cast directly into the mold.
The filled-waveguide horn 320 also includes a printed circuit board (PCB) 328 having a connector end 330 that is configured to be coupled with the base portion 326. The connector portion 330 includes a plurality of connectors 332 that operate to secure the printed circuit board 328 to the horn element 322. The connectors 332 can be any suitable connector for this purpose, such as solder, pins, epoxy, etc. The PCB 328 also includes a feed line 334 that makes contact with the metalized portion of the horn element 322, and includes a flared portion to provide impedance matching. The feed 334 is electrically coupled to integrated circuits or RFICs 336 and 338 that provide some of the electronics for each channel in the phased array. For example, the RFICs 336 and 338 could include the mixer 40, the PLL 42, the filter 44 and the amplifiers 46 and 48 in the transmit channel 12 and/or the LNA 52, the mixer 54 and the PLL 56 in the receive channel 14. The horn element 322 would be the antenna 24, and could be only a receiver antenna, only a transmitter antenna or a transceiver antenna depending on the particular application. The RFICs 336 and 338 are coupled to connectors 340 that would be coupled to edge connectors, such as on the control layer 76 in the phased array 70. The filled-waveguide horn 320 will be configured vertically relative to the plane of the layer 76, and will provide the desired polarization.
a) and 5(b) are a perspective top view and a perspective bottom view, respectively, of an antenna circuit 350 that also provides the desired characteristics and properties at high frequencies for the phased arrays discussed above. The antenna circuit 350 includes a dielectric substrate or layer 352 on which is formed a metalized layer 354 that is patterned to define an antenna element 356, such as a tapered-slot type antenna. RFICs 360 and 362 are provided on a back side of the layer 352, and can include the same circuit components as the RFICs 336 and 338. An antenna feed 358 is electrically coupled to the RFIC 360 by a via 366 that extends through the layer 352 and is electrically coupled to a feed location of the antenna element 356. Connectors 364 are electrically coupled to the RFICs 360 and 362, and can be coupled to edge connectors, such as on the control layer 76, as discussed above. The antenna circuit 350 is also vertically oriented relative to the PCB layer.
The present invention employs components and techniques so that the phase shift for the local oscillators in the phased arrays 10 and 70 are provided by using a tunable or programmable phase locked loop (PLL) that can provide phase control. The combination of these two technologies can provide a superior phased array in as many bits of phase accuracy is possible. Also, the control of the phased array is greatly simplified because a simple digital data buffer can be used to control the state of the PLL. This differs from previous disclosures in that phase shifting is provided for a radio on the LO, the clock is not phase shifted and each channel forming the phase shift is an independent signal generator that is synchronized to a global clock, which provides an absolute reference. Thus, for instance, a phase off-set can be provided that is important in correcting for manufacturing variances. Likewise, the low-level phase control can be written to a data register so that the PLL can provide the phase off-set without the need for continual control.
As mentioned above, programmable PLLs can be used to control the clock signal applied to the local oscillators for phase control. These oscillators can be set with a specific phase off-set with respect to a global reference clock. This means that a phased array can be formed by providing the appropriate phase off-set between elements. The operation of the phase off-set programmable oscillator allows for the control through a data register. This provides control of any relative phase off-set between zero and 2π. This can be used to form any number of phased array beams by recognizing the fact that the sum of any two sine waves of the same frequency is a sine wave at the same frequency, but with a unique amplitude and phase off-set.
In the general case with varying amplitudes and phases, the resulting sum is a sine wave at a different phase and amplitude, but at the same frequency. Because steering a beam using this condition requires adjusting the relative phase between the antenna elements in the array, and that this compromises separate oscillators each with independently controlled phases that are the same frequency, then it is clear that two beams may be formed simultaneously by forming an oscillation with the correct amplitude and phase for the sum of the two required oscillations. In fact, N beams can be formed using the same principle and any desired amplitude tapering may also be included.
For this effect to be most useful, compensating amplification needs to be included. This is necessary because in both transmit and receive, power is divided between the channels formed by the beams. Thus, it is important to include amplitude compensation as part of the design. For example, if two simultaneous beams are to be formed, 3 dB of additional gain is required at the output of the transmitter and/or the input of the receiver. If ten simultaneous beams are to be formed, 10 dB of additional gain will be required. Such gain may be present in an adjustable form to optimize power consumption or may be fixed. Techniques for producing adjustable gain can be provided by voltage programmable amplifiers, switchable parallel amplifiers, where the switches may be addressed as much as required, a combination of these, etc.
It has been recognized that the distribution of accurate timing control, typically in the scale of picoseconds, and even femtoseconds for frequencies over 60 GHz, is often a limiting factor to system performance. This is particularly important in phased arrays that rely on the distribution of a global clock. There are several important features to all systems that require accurate distributed timing. Manufacturing tolerances in semiconductor devices and transmission lines lead to skew and jitter phase errors, known to those skilled in the art. The variations in PCB line dimensions, variations in integrated circuit device delays, phase noise in oscillators, and cross-talk between digital and reference timing circuits leads to overall timing distribution errors that can significantly degrade system performance.
The power associated with the distribution of clock signals is often significant in high speed clock networks, especially in CMOS based devices. The cost of accurate timing distribution can often preclude the use of extremely accurate timing distribution systems if the distribution must support fan-out levels in hundreds of devices. For these reasons it is important to identify a technique of precise reference timing that is both extremely accurate, and cost and power effective.
One method for providing accurate timing is to distribute a clock signal by providing phase accurate distribution, such as found in traditional driver networks, cabling networks, etc. However, such systems are limited by fan-out of the clock source, that is, the number of elements that can be driven directly by a frequency source. The present invention provides a system that can provide clock distribution to large numbers of circuit elements.
The resonant H-tree architecture provides a high precision timing network where a clock signal can be distributed over a large number of elements. The basic element is a programmable oscillator where some of the oscillators operate at low frequencies and some of the oscillators operate at high frequencies. This provides correct phase shifts across an array when many elements are placed along an array edge. For example, if a phased array beam is to be steered to an angle θ and the array elements are spaced at λ/2; then each element from the array has a progressive phase shift of nπ sin θ radians. However, a PLL will wrap around the phase shift every 2π radians. This limitation becomes an important consideration for phased arrays that are based on programmable PLLs that can be tuned to provide a reference phase shift with respect to the other PLLs.
For the purposes of illustration, assume that there are sixty-four antenna elements on a side of the array. Then, the next four elements should start with an offset of 2π°radians of phase shift and end with 4π radians of off-set. This pattern continues until the last four element grouping would begin at 30π radians of off-set and end at 32π radians of phase shift. In this example, it would take approximately 16 cycles of a carrier for the phase front to become coherent, i.e., be at full power. The consequence of this is most noticeable on high bandwidth signals. Consider an encoded signal with a chip rate in the microsecond range where the signal has a carrier frequency in the 10 GHz range, and in this case for a 1 microsecond pulse, approximately 0.02% of the signal energy will be affected. On the other hand, for a 1 GHz bandwidth on the same carrier, 20% of the bandwidth is affected. It is important to note that in this example, the steer angle is 30°. For a 60° steer angle, approximately 35% of the bandwidth is affected with a substantial energy loss. Thus, for large steer angles, large arrays and high bandwidths it is important to be able to provide additional phase shift across the array so that modulation events are aligned with as nearly an ideal phase front as possible.
Given a programmable PLL that allows for a differential phase shift between PLLs as referenced to a common system clock, then additional phase shifts may be applied by forming clock layers. In this case, a 50 MHz clock signal can be distributed to the first layer of the programmable PLLs that up-converts the MHz clock to a 1 GHz signal. These PLLs can be programmed to account for skew effects during layout. The distributed clock arrangement is capable of additional phase shift to account for skew in the H-tree distribution network or in any clock distribution that worked in general providing improved array performance.
Use of a PLL allows more correction of phase errors in the clock distribution network 140. This feature can be important for high-bandwidth phased array antennas. In general, it should be noted that other resonant and even non-resonant structures could be used to form the clock distribution layers provided that suitable timing accuracy can be maintained.
The state machine 192 may be integrated into the RFIC circuit and controls the feedback loop of the programmable PLL 208 through the control register 210, which also may be integrated into the RFIC. The beamforming network 206 provides instructions to the state machine 192 in coordination to facilitate combining and processing of the individual array channels so that beams may be formed according to a method of operation. The state machine 192 can provide multiple functions including scanning, adaptive beamforming, switching, amplitude modulation, temperature compensation, off-set error calibration and tracking. Particularly, the state machine 192 can form a single beam by combining all of the elements of the array and steer it by adjusting the phases on each programmable PLL and the amplitude of the combined PLLs. One method of amplitude control is to provide an attenuator on the IF channel, which may also be controlled by the state machine 192. Moreover, the state machine 192 can be used for scanning multiple beams. In that case the scan time can be reduced as 1-N because each beam can scan less area.
The state machine 192 can also be used for providing adaptive beamforming operations. In the presence of a jammer, the state machine 192 can acquire the angle of arrival of the jamming signal after an initial scan by, for example, post processing of the phase delay added in each PLL, and by using any existing digital beamforming algorithm, it can create a null in that direction. The state machine 192 can also be used for switching different PLLs in the system for broadband operation. The state machine 192 can correct for any temperature drifts in the PLL 208 or other active circuits due to temperature variations along the entire system. The state machine 192 can provide error correction skew, for example, skew that arises from buffering a clock signal in a PLL. In such a case, calibration of the system after assembly can be used to provide off-set information. The use of a system, such as an electro-optical mapping system, can measure the individual phase and amplitude of each channel providing a set of temperature correlated deterministic corrections that can be stored in non-volatile memory in each state machine and that can be applied to provide correct phase and amplitude for each channel to improve beamforming. Alternately, such corrections can be loaded into the volatile memory of each state machine from the beamformer when the system is powered on or even during operation.
As discussed above, a phased array architecture can be fabricated using a variety of materials and a variety of fabrication techniques. The present invention provides a phased array architecture that is a fully vertically integrated realization of an entire phased array with all of the required art of passive components, as well as all of the active components required to form a system capable of digital beamforming, where all of the components can be formed on single layer substrates with local oscillator phase shifting capability. This is critical if the overall package is to be minimal in that the overall package will be of minimal thickness and/or planar dimensions no larger than that required to form the antenna aperture. It is important to recognize that unlike known systems, the RFICs may or may not be monolithically integrated. For example, one RFIC might be comprised of the PLLs, one RFIC may include data registers and a state machine, another RFIC may be a power amplifier or a low noise amplifier and another RFIC may include the mixers.
The RFICs can also be fashioned from appropriate technologies, such as SiGe, GaAs, InGa, As, silicon or any suitable semiconductor. The RFICs likewise might be fully integrated into single monolithic RFICs or any suitable combinations of these. The techniques of connection between the vertical integrated RF circuits might contain the antennas, filters and RF interconnects, and the RFICs might be flip-chip transitions that connect the output/input of the RFICs to the antennas and intermediate frequency lines. Similarly, the connection might be comprised of wire bonds that can be used to realize such connections.
It is important to recognize that the ADC/DACs may or may not be present on each channel and that the IF signals may be combined completely or with various levels of hybridization between complete integration and digitization at each channel. For example, one option has the I/Q channel separated, digitized and then summed. A second option includes the channel summed first and then digitized. A third option is to use a single IF signal not separated into in-phase and quadrature phase paths and digitized. A fourth option is to choose a hybrid approach where some channels are added and some are not in order to form multiple independent beams. Likewise, in the case of a transmit/receive switch for the receiver and transmitter, only this switch may or may not be present. As previously stated, the power amplifiers and/or the low noise amplifiers may or may not be integrated with the mixer, and the PLL may or may not be integrated into a single RFIC containing all of the active components. The clock distribution lines may or may not be combined into single distribution points as may the digital control mines. Further, the antennas may be integrated into the stack or may be attached separately.
The control circuitry includes multiple control signals. The most important signal is from the clock distribution network. The network contains a resonant H-tree network terminated in the programmable PLLs or as previously described may be provided by any other suitable device. Planar inductors (not shown) may be added in each leaf of the H-tree resonate network. These inductors combine with the distributed capacitance of the H-tree resonate network at a desired frequency. A power clock generator is used to periodically replenish the energy losses of the system and maintain the amplitude of the oscillation. The outcome of this is a low skew and jitter clock distribution network. Therefore, the number of via transitions needed to transfer the timing to each PLL is minimal. A resonate H-tree network can be used for distributing a low-jitter clock signal of the required frequency to the programmable PLLs. In order to reduce cross-talk between the clock traces, guard traces grounded with closely spaced vias may be used. In order to increase the oscillation between layers, all remaining distribution networks can be implemented with the strip lines. The thickness of each layer and the position of shielding layers should be optimized in order to reduce cross-talk and also maintain impedance matching for the distribution networks. Multiple vias can be used for transitioning the signals between layers all the way to the top of the array. Further, multiple vias can be used for ground equalization and suppression of parasitic modes. Further, thermal vias can be used to transfer heat out of the structure and into a cooling system attached at the back side of the array architecture.
One concern for the phased array architecture is thermal management, particularly the removal of heat from the active components, mainly the power amplifiers, which might be integrated within the multi-layered structure. According to the invention, one option is to use the heat pipes 306 for heat removal. As is known in the art, heat pipes of this type include some fluid medium that can undergo a phase transition in a contained environment, such as water or alcohol, which is wicked up along the sides of the pipe. Heat generated at the contact point then causes the phase transition of the fluid medium and heat is removed by the vapor phase from a particular location and transferred to a heat rejection location as the vapor circulates within the pipe. The heat pipes 306 can be inserted inside the structure through etched holes formed by laser, DRIE, chemical etch, etc, and then soldered on a metal surface that is in contact with a heat generator. The removed heat can then be transferred to a heat exchanger located at the back side of the architecture.
A second option is to use thermal vias to transfer the heat on the back side of the architecture. In this case, multiple thermal vias can be placed in close proximately to the amplifiers in order to transfer the heat to an appropriate heat exchange point. In either case, heat generated by components on the digital layer may also be attached to the heat sink by soldered connection, by mechanical connection to the package or any other suitable technique.
Another option is to use a low-thermal resistance layer in the formation of the stack, such as thick metal layers, high conductivity substrates that can allow heat to move with minimum spreading from generating source to heat rejection points, such as thermal vias or the like.
The foregoing discussion discloses and describes merely exemplary embodiments of the present invention. One skilled in the art will readily recognize from such discussion, and from the accompanying drawings and claims, that various changes, modifications and variations can be made therein without departing from the spirit and scope of the invention as defined in the following claims.
