Thermal analysis, simulation and benchtop testing of an X-Band transmitter RF power amplifier (RFPA) printed circuit board (PCB) for the University of Colorado, Boulder Earth Escape Explorer Deep Space CubeSat shows that FujiPoly XR-Um thermal interface material (TIM) provides a better solution compared to the alternatives considered. Results show a reduction in the operational temperature of the PCB from greater than 120°C to 77°C, which is below the maximum 85°C environmental operating temperature of the RFPA.

Dissipation of thermal power generated from RFPAs in CubeSats is a challenge for the effective operation of communications systems. Operation near thermal limits can impact output power and expected lifetime. While a challenge for all satellites, in this article we analyze an RFPA developed for the NASA Cube Quest Challenge by the University of Colorado, Boulder, considering the thermal dissipation problem in overall system design. The RFPA PCB includes an RFPA, bias controller and voltage regulator. The GaAs RFPA has an efficiency of approximately 25 percent and is the dominant system heat source. Low thermal conductivity between the RFPA and the CubeSat introduces a large temperature gradient, causing the RFPA to potentially operate at excessively high temperatures.

The ability to close a satellite communications link is a function of the received signal-to-noise ratio (SNR), which is typically from 2 to 10 dB for current modulation and coding schemes. The received SNR is a function of numerous factors, the most dramatic being space loss, which is proportional to the squared distance to the spacecraft. For deep space missions, space loss can exceed 240 dB (i.e., 24 orders of magnitude). This loss is typically offset by using large antennas with significant gain, by lowering the data rate to reduce the noise bandwidth and by increasing spacecraft transmitter power. While large deep space satellites have the mass, volume and budget to use higher power traveling wave tube amplifiers, most small satellites do not have this option and generally opt for lower power solid-state power amplifiers (SSPA). Some downlink analyses for deep space communication have used RFPA outputs from 2 to 10 W.1-2

SSPAs are categorized based on the underlying semiconductor technology, namely more mature GaAs and the more recently available GaN.3-4 While GaN technology is more efficient than GaAs, consider that GaN amplifiers typically run at drain voltages of 24 to 48 V compared with the 6 to 8 V for GaAs amplifiers. Currently, many CubeSats use 8 and 12 V battery busses, enabling a buck converter configuration for GaAs SSPAs, while GaN SSPAs use boost conversion for the same battery busses. While boost converters are typically less efficient than buck converters, operating a SSPA at a higher voltage requires lower current for the same DC power. These system trades and the differences in efficiency need to be considered in the system design.

The CU Earth Escape Explorer (CU-E3) communication system used a GaAs MMIC to achieve approximately 25 percent efficiency in its first design iteration. While improved efficiency is expected in future designs, 25 percent is used as the baseline for this study. With a 25 percent efficient RFPA MMIC powered by a 10 W (8 V at 1.25 A) power source, 7.5 W of thermal power is generated. Given the small MMIC size (approximately 25 mm2), thermal power dissipation is a challenge. The resulting power density is 300,000 W/m2, compared to the solar constant of 1370 W/m2. This illustrates the need to effectively address the thermal power handling in high-power, deep space transceivers.

The CU-E3 design uses a FujiPoly XR-Um TIM at the interface between the bottom copper layer of the PCB and the mounting face of the CubeSat structure. This TIM improves thermal conductivity from the RFPA PCB into the CubeSat, reducing the temperature gradient. Additionally, the TIM is electrically insulating, enabling the typical design goal of a single point ground. Modeling and analysis of the thermal system was aided using desktop simulation software.


The CU-E3 CubeSat is the University of Colorado entry into the NASA Cube Quest Challenge.5 The Cube Quest Challenge, sponsored by the Space Technology Mission Directorate’s Centennial Challenges program, was NASA’s first prize competition in space. Entrants competed for three available 6U CubeSat dispenser slots on the Exploration Mission-1 (EM-1), the first un-crewed lunar flyby of the Orion spacecraft, scheduled for launch by the Space Launch System (SLS) in December 2019. The University of Colorado Boulder’s CU-E3 6U CubeSat was one of the three finalists selected under the Cube Quest Challenge for launch on the SLS.6

Figure 1

Figure 1 Deep Space Derby mission trajectory for the CU-E3 CubeSat.

CU-E3 is designed to compete in the Deep Space Derby portion of the Cube Quest Challenge. To compete, the CU-E3 CubeSat will travel into deep space using the Earth escape trajectory from the Orion and SLS EM-1, with the duration of the mission one year from launch. Figure 1 shows the approximate Earth escape trajectory of the Orion, with the competition beginning at the 4 million km point, after the spacecraft has passed the moon. The Deep Space Derby competition has four main challenges:

1. Best burst data rate - Awards $225,000 for the spacecraft communicating the largest cumulative error-free data block within a 30-minute period. The minimum requirement is to communicate at least one 1024-bit error-free data block in the 30-minute window.

2. Largest aggregate data volume sustained over time - Awards $675,000 for producing the largest cumulative volume of error-free data in a 28-day period, with 1,000 1024-bit data blocks the minimum requirement.

3. Spacecraft longevity - Awards $225,000 for the longest period (i.e., number of days) between the first and last 1024-bit error-free data blocks, after passing the 4 million km distance and following the 28-day competition period.

4. Farthest communication distance from Earth - Awards $225,000 to successfully communicate at least one 1024-bit error-free block of data from the spacecraft at the greatest distance from the Earth after passing the 4 million km point and before the end of the competition.


A link budget estimates the power and aperture required to close the communications link. Table 1 provides a summary of the CU-E3 link budget at 8447.6 MHz with an SSPA output power of approximately 3.5 W.7 Several parameters, such as the signal and noise powers at the receiver, bit rate, energy per bit over noise power density (Eb/No) and link margin are illustrated for distances of 4 million and 27 million km. The two values for each parameter represent the analyses for a reflector array (RFA) and horn antenna. The transmitted effective isotropic radiated power (EIRP) includes the losses in the spacecraft’s transmit chain, the antenna gain and the RFPA output. These values are shown in dBm and assume a 3.5 W SSPA output.


Novel two- or three-stage GaAs RFPA MMICs achieve 30 to 50 percent efficiency in a laboratory environment;8 however, most commercially available GaAs amplifiers have 25 to 35 percent efficiency. The area near the gate of the MMIC, where the heat is generated, is very small; consequently, the power density is high. Table 2 illustrates this by summarizing GaAs RFPA MMICs operating in the 8 to 8.5 GHz range and reported in the literature, showing output power, power-added efficiency (PAE), PHEMT gate periphery and power density. The power density of the final stage of the RFPA is given by

Pd(W/m) = Po(W) / [Gp(mm) x 103N]

where Pd(W/m) is the power density of the final stage, Po(W) the output power from the final stage, Gp(mm) the gate periphery of a single PHEMT device in the final stage and N the number of PHEMT devices in the final stage.


The link budget of Table 1 requires the transmitter output power to be approximately 3.5 W to achieve at least a 600 bps downlink bit rate. The power and heat budget in Table 3 shows the need to handle approximately 10 W of thermal power or about 10 J/s of thermal energy, with the RFPA being the major source. Limited RFPA efficiency with the small MMIC size and a large deep space communication distance emphasizes the need to address the thermal power dissipation.



The RFPA amplifies the −3.5 dBm output from the High Rate CubeSat Communication System (HRCCS) to the required transmit power level of approximately 3.5 W.13-15 The RFPA PCB contains a buck convertor IC, a power amplifier bias sequence IC, a preamplifier MMIC and a power amplifier MMIC. The PCB has four layers: the RF/signal (layer 1), RF ground (layer 2), power (layer 3) and a ground/heat sink (layer 4).

Figure 2

Figure 2 RFPA PCB layout showing the area for the preamplifier chip and peripherals (A), power amplifier chip and peripherals (B), bias controller/sequencer chip and peripherals (C), buck converter chip and peripherals (D) and DSUB connector (E).

Figure 2 shows the layout of layer 1. Three dielectrics are sandwiched between the four copper layers (Rogers RO4350B, FR4 and FR4, in the same order). The top copper layer is electroless nickel immersion gold plated to prevent oxidation or other chemical reactions in the harsh space environment. The dielectric specifications of the PCB layers (top to bottom) are:

  • Dielectric 1 - Rogers RO4350B: 6.6 mils thick, Dk = 3.48, Df = 0.0037 at 10 GHz and 23°C.
  • Dielectric 2 - Isola 370HR: 14 mils thick, Dk = 3.92, Df = 0.0250 at 10 GHz and 23°C.
  • Dielectric 3 - Isola 370HR: 6.6 mils thick, Dk = 3.92, Df = 0.0250 at 10 GHz and 23°C.

The total thickness, including the copper and solder-mask layers, is approximately 48.9 mils.

The dielectric layer thickness is uniform along the stack to avoid PCB warping from thermal expansion during manufacturing. Power traces are 60 mils wide for the inner layer, which is not exposed to the environment, and 25 mils wide for the top and bottom layers. Vias have a current rating of 0.5 A per via. The vias and power traces are designed to reduce current density and avoid temperature rises in the PCB.

Figure 3

Figure 3 Thermal model material stack.

Figure 4

Figure 4 Predicted RFPA PCB temperature vs. PCB area using thermal software.

Table 4



As the highest thermal power density is in the RFPA, additional vias are added around the RFPA MMIC to enable efficient thermal conduction from the top layer (layer 1) to the bottom layer (layer 4). A portion of layer 4, under the RFPA, is designed as a heat sink and electrically isolated from the rest of the layer. One reason for electrical isolation is to provide separation between the RF and DC grounds, which are electrically connected on layer 1. Separation of these ground nets on layer 4 provides distinct locations for the thermal current flow that originates from both the RF and non-RF portions of the circuitry.

Use of a TIM of limited area helps the thermal power dissipation as well as maintaining electrical isolation between layer 4 of the PCB and the aluminum enclosure. The primary reason for electrical isolation between the PCB and its enclosure is avoiding ground loops. The aluminum enclosure is designed to shield the circuitry from radiation in space. FujiPoly XR-Um TIM satisfies the total mass loss (TML) and volume resistivity requirements and is the interface material of choice. The TML specification ensures the outgassing performance acceptable for space. Because the FujiPoly XR-Um TIM offers a thermal conductivity of only 17 W/mK, improvements in thermal conductivity can be obtained using materials which have higher thermal conductivity, such as hexagonal boron nitride nanoribbons (~2000 W/mK) and industrial isotropically enriched 12C diamond (3000 W/mK).