An Efficient and Linear SSPA With Embedded Power Flexibility for Ka-Band Downlink SatCom Applications

This article presents the design, realization, and experimental characterization of a solid-state power amplifier (SSPA) conceived for $Ka$ -band downlink (17.3–20.2 GHz) satellite communication (SatCom) applications. To the best of the author’s knowledge, this is the first space-borne SSPA realized for this peculiar application to achieve a technology readiness level of 5, entirely based on European technologies. All the activities carried out during the development phase will be discussed, starting from the power budget analysis, moving through the ad hoc design and characterization of the elementary 10-W microwave monolithic integrated circuits (MMICs) power amplifier (PA) in the 100-nm gate length gallium nitride (GaN) on silicon technology, and ending with the circuit solutions conceived to achieve state-of-the-art performance like saturated output power larger than 125 W with a gain and an overall efficiency better than 70 dB and 22%, respectively, while satisfying all the space-derating rules. The SSPA is composed of two subunits: the radio-frequency tray (RFT) to amplify the useful RF signal and the electronic power conditioner (EPC) to interface the module with the satellite primary bus, to actuate remote telecommand/telemetry services and to set different operating modes. The latter is based on a space-qualified microcontroller, whereas the former exploits a low-loss radial splitter/combiner structure in the waveguide to spatially combine the power provided by 16 elementary 10-W MMIC PA. The RFT also includes an input and output isolator, a gain control unit (GCU), a driver, and an analog linearizer. The latter helps to achieve a noise-to-power ratio (NPR) better than 18 dB when the SSPA is tested with a white-noise-like signal having a peak-to-average power ratio (PAPR) of 10 dB and an instantaneous bandwidth of 2.9 GHz. In this working condition, the SSPA provides an average output power and efficiency close to 100 W and 22%, respectively. Finally, it is worth mentioning that the SSPA is equipped with a power flexibility feature that allows it to reduce its power consumption up to 20%, depending on the actual RF power needed at the satellite level.


I. INTRODUCTION
S ATELLITE communication (SatCom) services have become essential in our interconnected world, bridging the connectivity gap in remote and underserved areas.These services are utilized across various industries, including telecommunications, broadcasting, maritime, aviation, emergency response, and scientific research [1], [2].Satellites provide reliable and wide-reaching coverage, enabling voice, data, and multimedia communication services.They play a vital role in enabling global connectivity, ensuring that people, businesses, and communities can stay connected regardless of their geographic location.As the ground network continues to advance with 5G and beyond [3], SatCom will continue to evolve, offering even faster data rates, higher capacity, and expanded capabilities, further enhancing the interoperability of our global communication infrastructure [4].
To support this evolution, the next step is the deployment of constellations of very-high-throughput satellites (vHTSs), which will have communication capacity even larger than 1 Tb/s per satellite, fully integrated with the terrestrial network and with high coverage and power flexibility [5], [6].To this aim, technology breakthroughs at the payload level are needed.For instance, the use of complex modulated signals with a large peak-to-average power ratio (PAPR) is envisaged to maximize the data rate, thus the linearity of the transmitted signal will be a crucial figure of merit.Additionally, active antenna arrays together with solutions to maximize the efficiency of the system while reducing the output power are expected, to guarantee both coverage and power flexibility [7], [8].These are relatively new concepts for space applications, which have to be added to the typical requirements such as high reliability, good thermal management, low size, and weight [9].This scenario clearly complicates the design of power amplifiers (PAs), whose performances, considering the typical requirement of hundreds of watts needed at the satellite level, dramatically affect those of the overall payload.Indeed, the PA is the most power-hungry element, thus affecting the efficiency and thermal management of the payload, as well as the main responsible for the transmitted signal quality due to its spectral regrowth.Also, mass and volume are significantly affected by the PA, since it is made redundant to face possible reliability issues [10].
High power at high frequency in space applications is traditionally linked to the adoption of traveling wave tube amplifiers (TWTAs), as power stage [11].However, advancements in solid-state technologies, such as gallium nitride (GaN), nowadays offer potential solutions to realize solid-state PAs (SSPAs) addressing space challenges while introducing significant benefits with respect to TWTAs, such as lower hardware cost, absence of heating time, graceful degradation, enhanced flexibility, and so on [10], [12].Efforts are continuously being made to develop high-performance SSPAs that try to strike the right balance among challenging and contrasting features such as power efficiency, thermal management, signal quality, space constraints, and many others [13], [14], [15], [16], [17], [18].
This article provides an additional contribution to this race, presenting the design, realization, and complete experimental characterization of an SSPA conceived for K a-band downlink (17.3-20.2GHz) SatCom applications.To the best of the author's knowledge, this is the first space-borne SSPA realized for the K a-band, entirely based on European technologies, to achieve a technology readiness level of 5, which means that the SSPA has been validated in the relevant environment [19].This article extends the content of the international microwave symposium (IMS) conference contribution [20], adding more insights into the SSPA design, above all stressing the conceived circuit solutions that have enabled the achievement of remarkable experimental results reported hereafter, and not included in [20].In particular, the outcomes of the verification tests campaign will be detailed, including thermal and vacuum tests as well as the linearity performance in terms of the noise-to-power ratio (NPR) with modulated signals having PAPR and instantaneous bandwidth of 10 dB and 2.9 GHz, respectively.
This SSPA is the main outcome of a research project named FLEXGAN [21], founded under the H2020 program of the European Union.The team is composed of five entities: TTI Norte, Tecnnalia, Macom European Semiconductor Center (MESC), AIRBUS, and the E.E.Dept. of the University of Rome Tor Vergata.
This article is organized as follows.The SSPA concept and architecture are presented in Section II, discussing the innovative circuit solutions conceived during its development.The results of the experimental characterization, ranging from small-to large-signal measurements, thermal-vacuum cycles, and linearity characterization, are reported and discussed in Section III. Conclusions are drawn in Section IV.

II. SSPA CONCEPT AND ARCHITECTURE
The main requirements of the SSPA, listed in Table I, have been defined by AIRBUS accounting for the expected needs of vHTSs.Three nominal operating points (NOPs) are foreseen, depending on the peculiarity of the signal to be amplified (continuous wave (CW) for NOP1 and multicarrier having up to 2.9-GHz instantaneous bandwidth and 10-dB PAPR, for NOP2 and NOP3) and the linearity level, in terms of NPR, to be achieved.For the development of the flexible K a-Band SSPA, the architecture depicted in Fig. 1 was proposed.It is composed of two main units: the low-frequency unit, named electronic power conditioner (EPC), and the radio-frequency tray (RFT) based mainly on GaN technology.

A. Electronic Power Conditioner Implementation
The EPC is divided into two subsystems: the power supply unit (PSU) and the control module (CM).The PSU interfaces the SSPA with the satellite external bus (100 V), thus providing the dc bias voltages to the SSPA internal devices.It is composed of the cascade of two filtering stages, needed to comply with the EMI/EMC requirements, followed by three galvanic isolated dc/dc converters, as shown in Fig. 2(a).In particular, one converter is used to provide the regulated voltage, from 5 to −5 V, to the low-power components of the RFT, whereas the other two feed the GaN power stages with a secondary voltage from 8 to 9 V. Considering the high power demanded to these converters, asynchronous half-bridge topology based on GaN switching devices with hysteretic mode control has been adopted.Moreover, a switching circuitry is embedded in the supply path to ensure the adequate biasing sequence of the normally-ON devices of the RFT.The PSU components have been selected accounting for their operating conditions, that is, peak current, peak voltage, and maximum temperature, also applying the space-derating standard.Fig. 2(b) shows its measured efficiency as a function of the delivered dc power for different secondary voltages, from 8 to 9 V. Efficiency values are in the range of 87%-93% for delivered output power larger than 40 W. It is worth mentioning that 9 V is the nominal drain bias voltage of the ad hoc designed MMIC PAs, whereas the feature of varying remotely the PSU output voltage down to 8 V was conceived to enable the SSPA power flexibility feature.
The CM is implemented relying upon a space-qualified microcontroller (SAMV71Q21RT from Microchip).As shown in Fig. 3, it implements the telemetry/telecommand (TM/TC) services as well as the remote control through the CAN bus.It monitors the SSPA behavior in terms of gain, power delivered, temperature in GaN devices, and so on and sets the internal parameters (i.e., the control signals) to actuate either the fixed gain mode (FGM) or the automatic level control (ALC) operating condition.In particular, when the FGM is enabled, the SSPA has to change and maintain a fixed gain value defined by the onboard computer (user), whereas when the ALC is ON, it has to maintain the output power selected by the user despite input power variation, from −18 to −29 dBm (i.e., IBO + 1 dB to IBO − 10 dB).Finally, if the satellite operator needs less output power, for instance, because of better weather conditions or coverage reallocation, the CM can enable the power flexibility feature.

B. RFT Design and Implementation
The RFT is conceived to properly amplify the input multicarrier signal from −19 dBm up to the required output power levels reported in Table I while assuring high linearity and efficiency performance.It was divided into two main units: gain control unit (GCU) and high power section (HPS), as graphically depicted in Fig. 4.
The GCU includes a signal conditioning part and an analog linearizer.The former, based on several commercial space-qualified MMIC low-level amplifiers (LLAs) and attenuators (ATTs), allows to set the different working modes of the SSPA (either FGM or ALC), as well as to compensate for thermal/aging variations of all electronic components present in the channel.The linearizer, composed of a 90 • hybrid coupler loaded by two parallel Schottky diodes, was introduced to compensate for the unavoidable phase and amplitude distortions of the downstream MMIC PAs.The final layout and the photograph of the realized GCU are shown in Fig. 5, whereas its performance has been already discussed in [20], thus they are not reported here.In summary, the GCU can compensate up to 4.5 dB and 32 • of gain compression and phase expansion, respectively.
The HPS represents the most challenging unit of this SSPA.Accounting for the requirements in Table I and the features of the selected MMIC technology, its architecture was derived through a power budget analysis.The idea is to combine 16 GaN MMIC PAs by using highly efficient spatial powercombining techniques.For its development, the research work was conducted following two main paths in parallel.The first one was related to the ad hoc design of a high-efficiency and linear 10-W MMIC PA based on the commercially available and space-evaluated 100-nm GaN-Si technology of MESC, to be used as an elementary brick.It is important to highlight that the choice of not using different monolithic technology, such as GaN on silicon carbide (GaN-SiC) was intentional since one of the primary objectives of FLEXGAN was also to showcase the capabilities of the MESC GaN-Si process for SatCom applications.
The second one focused on the identification of powerful methods to combine 16 of these elementary MMICs with maximum efficiency, to achieve more than 125-W output power, thus leading to an SSPA that is comparable with TWTAs.
To mitigate the risks associated with an MMIC implementation at this frequency, two different versions of PA were designed.Moreover, to be compliant with space-derating rules and to avoid possible reliability issues, the nominal drain bias voltage was reduced to 9 V, from the maximum nominal value of 15 V, whereas the current density was set to roughly 60 mA/mm, that is, class AB.Such bias point allows keeping the maximum junction temperature of the active devices lower than 160 • C in the worst-case condition, that is, when the base plate of the SSPA is at 65 • C. Both MMICs were based on a three-stage architecture but exploiting different circuit configuration, as shown in Fig. 6.The first MMIC named AKILOS2 exploits a standard corporate configuration [see Fig. 6(a)], whereas the other, named DIAKOS2, uses a balanced configuration with an input-output Lange coupler [see Fig. 6(b)].The photograph of the realized chips is reported in Fig. 7.The detailed design of both MMICs together with their performance are reported in [22] and [23] for AKILOS2 and DIAKOS2, respectively.The output power and power added efficiency (PAE) measured on-wafer for several samples of both architectures are reported in Fig. 8.Both PAs provide more than 10 W of saturated output power all over the specified bandwidth with a gain larger than 24 dB.In terms of efficiency, AKILOS2 outperforms DIAKOS2 achieving a PAE peak of 45%.Therefore, after a careful evaluation of their performance and the investigation of their stability over time [24], AKILOS2 was selected to implement the elementary module of the SSPA.It is worth mentioning that the performance achieved by these MMICs represents the actual state-of-the-art PAs in this frequency range, at least using European technologies.
For their combination in the last stage of the HPS (see Fig. 4), the research work was focused on the investigation of spatial power combining techniques, since planar ones at this frequency and for this power level are characterized by unacceptable losses [25].Specifically, two main technologies have been considered: the cavity-based radial combiner (RC) and the single-waveguide longitudinal probe spatial combiner (SC).Their 3-D view is reported in Fig. 9, for comparison.Both solutions have been deeply investigated, also implementing dummy prototypes to experimentally verify their performances in terms of insertion loss and combining efficiency, as well as other key aspects for space applications such as multipaction, size, weight, mechanical, and thermal behaviors.The design description of both structures is reported in [26], whereas Table II summarizes the main outcomes of the comparative study.
In terms of electrical performance, the RC shows better combination efficiency with respect to the SC (89% versus 73%), thanks to the lower insertion losses.This is mainly because, in the RC, the power coming from the 16 parallel modules is sum together in a single step, whereas in the SC, this function is accomplished by cascading three different levels of recombination [26].On the other hand, thanks to a more compact and integrated structure, the SC outperforms Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.the RC in terms of size and weight.This is essential because, in the SC, the MMICs are not individually packaged, being the hermeticity achieved by sealing the shell of the structure with a laser.Together with the hermeticity, in a space environment, it is fundamental to guarantee that the adopted structure is multipactor-free and that the junction temperature (T j ) of the GaN MMIC components does not exceed the boundary limit of 160 • C. In this respect, both solutions have been simulated by using Spark3D software for verifying the multipaction compliance, whereas the thermal-mechanical behavior has been evaluated by using dedicated multiphysics tools.With respect to multipaction, a margin higher than 10 dB, when compared to the maximum power expected at the SSPA level (e.g., 51 dBm), has been obtained in both cases, which is sufficient to satisfy the European Cooperation for Space Standardization (ECSS) rules without carrying out dedicated tests.In terms of thermal assembly, the RC assures a better distribution of the heat generated by the MMICs, and then a lower T j of the devices, essentially thanks to its bigger shape with respect to the SC, even if the latter, with the adoption of advanced materials with very high thermal conductivity, also complies with the T j limit.Finally, the graceful degradation, that is, the variation of the performance of the combining structure when a certain number of MMIC PAs is assumed to fail, has been evaluated.Simulation results have shown that the RC, due to the absence of isolated networks, is more sensible to MMIC failures with respect to SC. Fig. 10 shows the gain of both solutions for a different number of died modules normalized to the reference state, that is, when all PAs are working perfectly.For instance, with four MMICs OFF, the insertion loss of the SC degrades by about 2 dB, whereas, for the same condition, that of the RC degrades up to 4 dB with a larger ripple.
At the end of the comparative study, considering the pros and cons of both solutions, as well as the boundary conditions to be respected at the payload level, the RC realized in WR-42 and made in aluminum was selected to implement the SSPA.Each individual path of such a structure was also experimentally characterized by using a two-port vector network analyzer (VNA).In particular, the measurements of the splitter were carried out by keeping port one of the VNA always connected to its input port, whereas port two was connected to one of the 16 output ports, with the other 15 parallel ports connected to WR-42 50-terminations.Similarly, for the combiner, port two of the VNA was always connected to its output port, whereas port one was connected to one of the 16 input ports, with the other 15 parallel ports terminated with 50-loads.Fig. 11 shows the results of this characterization in terms of magnitude and phase of the S 21 parameter of each path.Notably, variations within ±0.3 dB and ±5 • in terms of magnitude and phase, respectively, have been observed, mainly due to manufacturing tolerances.Clearly, being 1-to-16 and 16-to-1 structures, the average gain is around −12 dB.
To be compatible with the designed radial structure, each MMIC has been individually packaged and equipped with input-output microstrip-to-waveguide hermetic transitions showing negligible insertion loss.Fig. 12 shows the most relevant steps in the assembly of the MMICs, from the designed envelope to the final high-power module (HPM).The package has been realized in copper, whereas hermetic feed-throughs have been used to bring the bias.Each HPM has been subject to an X-ray inspection to evaluate the presence of voids underneath the MMIC.Clearly, the higher the number of voids, the worse the thermal dissipation is.In our case, the eutectic process conceived for this assembly led to a very limited number of voids, as also visible in Fig. 12.
The RFT is completed by adding a gain stage between the GCU and the last stage of the HPS, namely the medium-power module (MPM) in Fig. 4. Also, this amplifier is based on AKILOS2 MMIC.The choice has been taken to optimize the cost and to have a driver working in almost linear conditions when the parallelized HPMs are pushed into compression or are subject to a gain reduction process, for instance, due to a temperature increase.The gain of the final stage is about 22 dB, thus the driver should provide a minimum output power of 51-22 = 29 dBm, which is well below the saturated output power of the developed MMIC.All 17 packaged MMICs (16 in the last stage plus the MPM) have been individually tested  before their integration into the HPS.Fig. 13 shows the linear performance of the assembled HPMs.The uniformity is quite good with an S 21 variation among samples of about ±1 dB and ±20 • , in terms of magnitude and phase, respectively.Moreover, to maximize the recombination efficiency in the HPS, the position of each HPM was chosen accounting for their individual phase and amplitude responses, as well as those of each path of the radial splitter/combiner structures reported in Fig. 11.This was done by performing an optimization process in ADS Keysight software, in which the optimized variable was the position of each HPM, and the goal was the minimization of both the phase misalignment and amplitude variation among the 16 paths.
Finally, Fig. 14(a) shows the measured output power in CW of all packaged MMICs as a function of the frequency, whereas Fig. 14(b) reports their PAE as a function of the output power at center frequency.The performance is well in line with those measured on-wafer [20] with the modules achieving an output power larger than 40 dBm all over the bandwidth and a PAE larger than 35%.

C. SSPA Assembly
Once the realization and tests of every single subunit within the SSPA were completed, a structural box to accommodate all the electronic components in the smallest volume and footprint possible was designed.The SSPA envelope was specifically conceived to maximize the heat transfer toward the satellite chassis, also accounting for the forces and mechanical stress Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply. of a launch.Moreover, it was built by using lightweight composite materials, such as carbon fiber-reinforced plastic, to minimize the overall weight.The RFT was completed by adding a waveguide coupler at the output to sample the power and two isolators, one at the input and one at the output, to comply with the return losses requirement and, above all, to protect the SSPA against output load mismatch.Fig. 15 shows some photographs taken during the different phases of the assembly together with their final dimensions and weight.

III. EXPERIMENTAL RESULTS
A detailed and complete test campaign has been conducted to verify the SSPA performance and reliability accounting for space environment requirements.Therefore, the main electrical performance, mechanical, environmental (vacuum and temperature), and EMI/EMC characteristics have been experimentally checked.
A. Small-Signal Characterization Fig. 16 shows the scattering parameters measured for different attenuation levels, from 0 to 30 dB, and fixing the base plate temperature of the SSPA at T BP = 25 • C. The specified gain of 70 dB is achieved with 10 dB of the attenuation set in the GCU.This was properly chosen to ensure a larger dynamic range for compensating thermal/aging variations all over the SSPA lifetime.Such measurements have been carried out without enabling the ALC or FGM operating mode, thus there is no compensation for the gain ripple.Anyway, the latter is within ±1 dB all over the specified bandwidth.The input and output return losses are always better than 20 dB, thanks to the presence of the isolators.
The small-signal parameters have also been measured enabling the power flexibility feature (i.e., setting the drain bias voltage of the HPMs from the nominal value of 9-8 V), and setting the attenuation to 10 dB.Fig. 17 compares the small-signal gain with and without power flexibility.The important aspect is that the shape of the gain does not change, since the average gain reduction of about 2 dB can be easily compensated by setting a different attenuation level in the GCU.
To verify the intrinsic gain stability over temperature (i.e., the SSPA is in an open loop with both ALC and FGM disabled), the scattering parameters were measured by varying the base-plate temperature (T BP ) from 25 • C to 65 • C with 20 • C step.Fig. 18 shows the behavior of the small-signal gain as a function of the frequency for such T BP .Notably, the SSPA exhibited a quite stable gain with temperature from 17.3 to 18 GHz, with a negligible variation.From 18 GHz, the gain variation increases with frequency, achieving a peak of 3.5 dB at 20.2 GHz, whereas the average value in the rest   of the bandwidth is about 1.7 dB.For the same temperature conditions, the measured group delay and noise figure are reported in Figs.19 and 20, respectively.For these figures of merits, requirements were set to 0.5 ns over any 36-MHz bandwidth and to 10 dB, respectively.Also, in this case, the final SSPA results to be compliant with the specs.

B. Large-Signal Characterization and Power Flexibility
Once completed the tests in linear conditions, the SSPA was subject to an extensive characterization campaign in a nonlinear regime to verify the most challenging requirements in terms of efficiency, linearity, and power levels.
Fig. 21 shows the measured output power and PAE of the overall SSPA when the input power is fixed to −19 dBm  and T BP = 25 • C. The output power is even larger than 51 dBm with a PAE better than 24% over a large portion of the bandwidth.The same features as functions of both the frequency and the input power are shown in Fig. 22, for completeness.
The output power and PAE at NOP1 have also been measured by varying T BP from 25 • C to 65 • C, obtaining the results reported in Fig 23 .The behaviors of such features look quite similar over temperatures, with an average reduction of the output power of 0.004 dB/ • C only.This demonstrates the very good thermal management of the mechanical box designed, and the capability of the SSPA to efficiently provide an almost constant and high power all over the specified temperature range.
To verify the ALC functionality, the SSPA output power was fixed to NOP3 (50 dBm) at the center frequency, and its input power was swept from −18 to −29 dBm with a 1-dB step.Fig. 24 reports the measured output power as a function of the input power, showing a very stable behavior.Similar trends have been measured also fixing the output power to different thresholds.
As mentioned in Section II-B, the SSPA implements a power flexibility feature.Therefore, the satellite central control can reduce the output power, depending on the actual needs, by sending a telecommand signal.When this feature is exploited, the EPC varies the dc biasing voltage of the HPMs accordingly with the requested power value, thus minimizing the SSPA power consumption with a minimal degradation in terms of linearity.The calibration law between the output  power and bias voltage was prederived by measuring the SSPA output power for different drain voltages, from 9 V, which corresponds to the nominal biasing point, to 8 V, which is the minimum level that the PSU can deliver.To appreciate the benefits introduced by such a quite innovative functionality, Table III reports the measured dc power consumption of the SSPA at center frequency for different biasing voltages, once the output power was fixed to a certain value.As can be noted, for NOP3 (50 dBm), the variation of the bias voltage from 9 to 8 V allows for a saving of nearly 100 W (22%) of dc power, whereas for NOP2 (48 dBm), the saving is of around 72 W (21%).Such results highlight the usefulness of the power flexibility feature in terms of power even if its usability has to be confirmed also evaluating the associated linearity.As it will be shown, reducing the biasing voltage causes roughly 2 dB of reduction in terms of NPR for a given output power level.The linearity of the SSPA has been evaluated at NOP2 and NOP3 by carrying out NPR measurements adopting a whitenoise-like signal (29 001 carriers spaced 100 kHz) having a PAPR of 10 dB with an instantaneous bandwidth of 2.9 GHz (i.e., covering the entire 17.3-20.2-GHzband) and by varying the notch bandwidth from 125 to 250 MHz.Such a characterization was carried out by varying the base-plate temperature and with and without the power flexibility, to correctly address the linearity-efficiency tradeoff.Notably, the NPR is defined and evaluated as the ratio of the output power spectral density (PSD) of the analyzed system with the notch filter presented and measured within the filter band, to the output PSD without the notch filter and measured within the same notch filter bandwidth [27].Fig. 25 shows one of the captured output spectra at NOP3 (100 W) with a 125-MHz notch filter and the power flexibility OFF at T BP = 25 • C. In this condition, the measured NPR is 18.94 dB with a power consumption of 446 W and a corresponding PAE of 22.4%.From such a figure, simply comparing the level of the channel power inside and outside the notch filter bandwidth, the NPR could seem higher with respect to the reported value.However, it is worth highlighting that this is just a graphical result since the NPR has to be evaluated considering the PSD (dBm/Hz) inside and outside the notch filter.
The most relevant results of this peculiar and fundamental characterization are summarized in Tables IV-VII.Notably, when the drain bias voltage of the HPMs is set to 8 V, the maximum output power does not fully comply with the NOP3 level, probably due to the reduced voltage swing available.On the other hand, it is always possible to satisfy the output power at NOP2 with a significant reduction of the dc consumption, while providing almost the same NPR value.

C. Thermal Vacuum Test
A thermal vacuum test is a fundamental step for spaceborne SSPAs.The aim is to verify whether the component can survive, without loss of integrity and functionality, to thermal and pressure conditions experienced in the space environment.Additionally, it provides a first experimental assessment of the design robustness against those destructive phenomena,  such as multipaction and corona breakdowns, that are hard to accurately predict during the development phase.Before reaching the vacuum conditions, a bake-out process of the system has been carried out to minimize the outgassing during the tests.The tests were carried out at the European High-Power Laboratory in Valencia (Spain).Table VIII summarizes the temperature limits adopted for the tests, whereas Table IX lists the adopted criteria.
Finally, Fig. 26 shows the evolution of the temperature in the chamber during the first thermal cycle together with the time frames allowed to carry out the functional tests, to verify the SSPA integrity.In particular, the functional test consists of a power sweep at the center frequency performed every 10 • C, between +5 • C and 65 • C.
As can be noted from Fig. 27, where the PAE and gain at T BP = 25 • C before and after the thermal-vacuum cycles are reported, no significant degradation has been observed.This, once again, confirms the robustness and reliability of the realized SSPA.

D. EMI, EMC, and Mechanical Tests
The SSPA was also subject to both EMI and EMC tests as well as several mechanical stress tests.Before and after each test, the functionality of the SSPA was verified by performing Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.a power sweep at center frequency.EMI and EMC results were in line with the expectations, whereas mechanical tests have highlighted that the unit has maintained structural integrity, and no failures have been observed.
IV. CONCLUSION An SSPA conceived for the K a-band downlink SatCom applications has been presented and deeply discussed.To the best of the author's knowledge, this is the first space-borne SSPA realized for this peculiar application to achieve a technology readiness level of 5, entirely based on European technologies.The SSPA has been subject to a very extensive and complete measurements campaign, demonstrating state-of-the-art results without evidencing reliability issues.Saturated output power larger than 125 W with a gain and an overall efficiency better than 70 dB and 22%, respectively, have been measured over different base-plate temperatures.The utility of the power flexibility feature was demonstrated, achieving a reduction in the power consumption up to 20%, without significantly worsening the linearity.Indeed, an NPR better than 18 dB was measured when the SSPA outputs an average power up to 100 W, being driven with a white-noise-like signal having a PAPR of 10 dB and an instantaneous bandwidth of 2.9 GHz.Such remarkable results pave the road toward vHTSs fully based on solid-state technology.

Fig. 2 .
Fig. 2. (a) Schematic of the PSU and (b) its measured efficiency as a function of the delivered dc power for different secondary voltages.

Fig. 10 .
Fig. 10.Simulation of the gain reduction of both RC and SC when a certain number of PAs fails, that is, graceful degradation.

Fig. 11 .
Fig. 11.Magnitude and phase of the measured paths of the input (left) and output (right) section of the RC.

Fig. 12 .
Fig. 12. Assembly of the MMIC in the developed package.

Fig. 13 .
Fig. 13.Small signal gain and phase responses of the 17 HPMs.

Fig. 14 .
Fig. 14.(a) Measured output power as a function of the frequency and (b) PAE as a function of the output power at a center frequency of all HPMs.

Fig. 15 .
Fig. 15.Photographs of the SSPA during the assembly.

Fig. 16 .
Fig. 16.Scattering parameters of the SSPA at T BP = 25 • C setting different attenuation levels.

Fig. 17 .
Fig. 17.Small-signal gain of the SSPA with and without the power flexibility enabled at T BP = 25 • C.

Fig. 18 .
Fig. 18.Small-signal gain of the SSPA for different base-plate temperatures as a function of the frequency.

Fig. 19 .
Fig. 19.Group delay of the SSPA at 25 • C, 45 • C, and 65 • C of the base plate temperature.

Fig. 20 .
Fig. 20.Noise figure of the SSPA at 25 • C, 45 • C, and 65 • C of the base plate temperature.

Fig. 21 .
Fig. 21.SSPA output power and PAE at NOP1 and T BP = 25 • C as functions of the frequency for an input power of −19 dBm.

Fig. 22 .
Fig. 22.(a) SSPA output power and (b) PAE as functions of the frequency and input power at T BP = 25 • C.

Fig. 23 .
Fig. 23.SSPA output power and PAE at NOP1 as functions of the frequency for different T BP values.

Fig. 27 .
Fig.27.PAE and gain of the SSPA before and after the thermal-vacuum cycles as functions of the output power at the center frequency.

TABLE I REQUIREMENTS
OF THE SSPA Fig. 1.Block diagram of the SSPA.

TABLE II COMPARISON
BETWEEN RC AND SC

TABLE III SSPA
DC POWER CONSUMPTION FOR DIFFERENT BIAS VOLTAGES AND OUTPUT POWER LEVELS

TABLE IV NPR
RESULTS FOR THE T BP = 25 • C AND 125-MHz NOTCH FILTER

TABLE V NPR
RESULTS FOR THE T BP = 25 • C AND 250-MHz NOTCH FILTER

TABLE VI NPR
RESULTS FOR THE T BP = 65 • C AND 125-MHz NOTCH FILTER

TABLE VII NPR
RESULTS FOR THE T BP = 65 • C AND 250-MHz NOTCH FILTER

TABLE VIII TEMPERATURES
FOR THE THERMAL VACUUM TEST

TABLE IX THERMAL
VACUUM CYCLING TEST PARAMETERS Fig. 26.Thermal vacuum test profile.