Revolutionizing radar design with electronically reconfigurable GaN power amplifiers
[Copy link]
This blog post originally appeared in Microwave Product Digest*
This paper demonstrates for the first time a reliable commercial high power amplifier based on a multi-band transmitter design that uses the patented reconfigurable PA technology demonstrated by Charles Campbell [2,3,4]. The reconfigurable PA uses a single input and single output matching network that is reconfigured based on the control bit settings for each relevant band. Each bit setting configures all matching networks for optimal performance for a specific band, allowing the PA to achieve optimal system-level performance in a compact package. This reduces overall size and weight. This new reconfigurable PA design approach overcomes several shortcomings of traditional multi-band transmit front-end designs. The most obvious advantage is the elimination of the band select switch at the PA output. This reduces output loss by 0.8-1.0 dB, giving it a significant advantage over traditional design approaches. If designed with optimal load impedance and intelligent switch placement, the reconfigurable PA can approach the performance levels achieved with specific, independently tuned band amplifiers.
Modern radar systems are configured as multi-band radars that use multiple frequency bands to resolve complex scenarios under a variety of environmental and target conditions. These systems provide unparalleled levels of performance and are capable of detecting and tracking hostile targets. For example, Dual Band Radar (DBR) is the first radar system used by the U.S. Navy fleet that is capable of operating in two frequency bands (S/X band) simultaneously, coordinated by a single resource manager [1]. S-band signals are less susceptible to adverse weather and atmospheric attenuation. On the other hand, X-band is often used for high-resolution target imaging applications. Most current power amplifiers (PAs) are not suitable for multi-band radar systems because the required frequency bands are too far apart and each PA is optimized for each target band. There are several approaches to achieve broadband or multi-band characteristics by switching between individual PA MMICs. These approaches use broadband non-uniform distributed PAs (NDPAs) covering both bands or dual passband PA designs.
The advantages of reconfigurable PAs are even more pronounced when compared to broadband amplifiers. In broadband amplifier design, the load impedance is typically designed to be lower than the optimal load target value to achieve high output network bandwidth. Broadband amplifiers reduce output power and power added efficiency (PAE). Therefore, the ability to synthesize the optimal load impedance is key to reconfigurable PA design. This ultimately enables the amplifier field effect transistor (FET) periphery to be increased, thereby maximizing output power within thermal constraints. These design principles have been implemented in Qorvo's new product, the QPA0007**. The QPA0007 is a reconfigurable 30 WS/X-band power amplifier that uses Qorvo's 150 nm gate length GaN HEMT process technology (QGanN15). A comparison between the reconfigurable amplifier and broadband and traditional multi-band approaches is shown in Figure 1.
Figure 1: Comparison of multi-band power amplifier front ends
Process and packaging technology
Qorvo's QGaN15 process technology is ideal for high-power PA designs in X-Band. It features fast transistors with high gate-drain breakdown voltages, making it ideal for high-power applications. QGaN15 offers multiple process options for different circuit applications. For the QPA0007, a patented process technology was used to improve device and circuit performance. The top metal layer enables the use of narrower output matching traces, which significantly reduces physical area while maintaining metal current density design rules. Output network losses are not very sensitive to metal thickness. At X-Band, the use of a passivation layer degrades circuit performance but enables the use of cost-effective packaging. A second benefit of using a passivation layer on the chip is that it helps improve FET and passive network modeling accuracy compared to using only an overmolded package without a passivation layer. Using a more expensive air cavity package eliminates the passivation layer, resulting in higher circuit performance.
The QPA0007 is packaged in a cost-effective, overmolded 7mm x 6mm electroplated heat sink (PHS) package. The PHS package is very flexible and provides designers with a good off-chip heat dissipation path for medium output power devices. A wide range of input and output connections and a relatively large pad spacing allow for high PCB attachment yields. On the evaluation board (EVB), the control pins and gate pins can be connected from the top or bottom. For reliable drain connections, connections need to be made from both sides. The QPA0007 package pinout and dimensions, as well as the evaluation board, are shown in Figure 2.
Figure 2: QPA0007 PHS package and evaluation board
Circuit Design
Fundamentally, the QPA0007 is a two-stage reactively matched power amplifier. Band switching is accomplished by switchable capacitors and inductors controlled by on-chip level shifters that adjust the switching FET bias levels. Each network is designed to maintain the optimum load for each band. This is a minor tradeoff compared to band-specific designs. Output network losses are one of the key design parameters and are affected by switching losses. Fortunately, these switch losses are small compared to using separate band selection switches at the PA output. The number of tuning switches is minimized, both from an overall loss perspective and from a complexity and size perspective. Typically, to achieve low switching losses, the switch periphery tends to be larger, so the off capacitance is high. The off capacitance plays an important role in ensuring the effectiveness of the switching element. This limits the availability of switchable shunt capacitors in the output network. Typically, the tuning inductor for S-band is much larger than for X-band. In the signal path, using a series switch to tune the series inductor does not make much sense because of the additional switching losses and a compromise should be made to achieve a close to optimum load target value.
The interstage matching network is designed in a different environment than the output network. The interstage matching network is limited by bandwidth and space rather than loss. Therefore, smaller switches can be used in multiple locations to achieve the optimal load target value.
When comparing the output and interstage networks, the input network has more relaxed loss requirements and more room for switching and control signals, so it has the most flexibility. Both the input and interstage networks have some impact on the stability performance of the amplifier. Adding additional losses ensures stability under a wide range of operating conditions, especially in extreme cold conditions. The tuning capacitors and FET terminals are designed to withstand high voltage standing wave radio (VSWR) load conditions at maximum input drive conditions to avoid breakdown.
Finally, the overall design challenge is to limit the S-band small signal gain without destroying the X-band gain. Lower frequency FET performance helps improve S-band performance, but the challenge of extending the low-end bandwidth without degrading X-band performance limits S-band performance. Even with switchable tuning elements, this is challenging.
performance
The QPA0007 is tuned to cover the S-band 3.1-3.5 GHz and the X-band 9-11 GHz. The two band switching signals are complementary, 0 V and -10 V for the S-band and -10 V and 0 V for the X-band. The control signals result in a 5 mA source or sink current, depending on the band selection.
The quiescent bias current of the QPA0007 is 700 mA at 26 V or above. Since the input power forces the drain current to rise, the output power and PAE are completely unaffected by the quiescent bias current. Therefore, the quiescent bias current can be set based on other performance parameters such as small signal gain and switching time.
All reported measurements were obtained from a production EVB and calibrated using the QPA0007 input and output pins. The measured small signal gain at 25°C was 27 dB for the S-band and 23 dB for the X-band. This difference in small signal gain reflects the performance variation of the FET across the frequency band. The input return loss is above 20 dB for the S-band and 10 dB for the X-band. The measured S-parameters are shown in Figure 3.
Figure 3: QPA0007 S/X-band S-parameters
In S-band, the output power of QPA0007 is 45 dBm and PAE is 48%. The large signal gain at the optimum operating point is 21 dB and the current consumption is 2.6 A. In X-band, the output power is 44.5 dBm and the PAE is 32%. The large signal gain is 18.5 dB and the current consumption is 3.6 A. These results are measured under the condition of drain pulse of 100 s/1 ms. Figure 4 shows the large signal performance curve of S/X band.
Figure 4: S/X band output power and PAE
Harmonics are measured into a 50 Ω load. The second harmonic of the S-band is below -25 dBc, and the third harmonic is -25 dBc. The second and third harmonics of the X-band are -35 dBc and -55 dBc, respectively.
The small signal and drive stability of the QPA0007 has been tested at -40°C with a VSWR of 10:1 load. Device reliability has been tested at 85°C with a VSWR of 3:1 load using extreme input drive with no performance degradation.
Switching times can be divided into two categories: No band switching when the RF signal is on and band switching with the RF on. In real applications, it may not be necessary to turn on the RF at the same time as the band switching, but it illustrates the capabilities of the device. Switching times within the band range are less than 100 ns. Band switching with the RF on is less than 1 s. In both use cases, RF shutoff is almost instantaneous.
The power dissipation is 40 W in S-band and 70 W in X-band during 100 s pulse width and 1 ms pulse period. Pulsing can be achieved by using either drain pulses or RF pulses. This pulse characteristic keeps the device junction temperature below the long-term stability limit at 85°C substrate temperature. A comprehensive thermal analysis was completed to validate the thermal analysis conclusions based on the measured data. The QPA0007 fully meets the MSL 3 and HBM 250V production rating requirements. Table 1 summarizes the measured EVB results.
Table 1: Summary of QPA0007 measured data performance
Summarize
The reconfigurable multi-band PA approach presented in this article offers significant advantages over traditional band-switching PA front ends. Qorvo’s QPA0007 uses patented technology and is the industry’s first product to improve both output power and efficiency performance in both S/X bands using the same device. In addition, the QPA0007 offers customers cost-effective, high-volume packaging with a competitive form factor.
| 
提升卡
变色卡
千斤顶