In the world of BUCs (Block Up Converters) and SSPAs that form the transmit side of a satellite link, the power amplifier has undergone a generational change from GaAs to the GaN HEMT. Terminals in the LEO-constellation era demand higher output power, higher efficiency, and a smaller enclosure — all at the same time. GaN is the practical technology that satisfies all three at once. This article digs into “why GaN” starting from the material physics, covers the thermal and linearity considerations a designer needs to keep in mind, and closes with the structural shift in the market.
Material Physics — Why the Wide Bandgap Matters
GaN’s advantage is largely decided at the level of material constants. Its bandgap is 3.4 eV — two to three times that of Si (1.1 eV) or GaAs (1.4 eV). What this feeds into directly is the breakdown field: roughly 3.3 MV/cm for GaN, about eight times that of GaAs (roughly 0.4 MV/cm). A high breakdown field means you can apply a high drain voltage. Whereas GaAs power devices operate at around 5–12 V, GaN HEMTs run at 28–50 V.
The figure above compares the key material properties of Si, GaAs, and GaN normalized to Si. The gap in breakdown field stands out. On top of that, the AlGaN/GaN hetero-interface forms a two-dimensional electron gas (2DEG) with a sheet density on the order of 1013 cm-2 without any doping, carrying very high current density while maintaining a mobility of 1500–2000 cm²/Vs. High voltage times high current — that is the real identity behind the extraordinary figure of 5–10 W/mm of power density (a GaAs pHEMT manages roughly 0.5–1 W/mm).
Looking at the Johnson figure of merit — a yardstick for high-frequency capability, proportional to breakdown field × saturation velocity — GaN comes out 15–20× ahead of GaAs. As a microwave power device, it is simply in a different league at the material level.
The practical meaning of an order-of-magnitude higher power density is that “for the same output power, the device is an order of magnitude smaller.” A smaller device has less parasitic capacitance and keeps its input and output impedances higher, which makes broadband matching easier and reduces the loss in the matching networks. The structure inherently favors achieving high power, wide bandwidth, and high efficiency simultaneously.
GaN-on-SiC — Thermal Design Rules Everything
Once power density goes up, heat removal becomes the bottleneck. Tens of watts of dissipation concentrate on a die a few millimeters square, so the substrate’s thermal conductivity directly governs performance. This is where the dominance of GaN-on-SiC (GaN on a SiC substrate) becomes clear. SiC’s thermal conductivity is about 4.9 W/cm·K — more than three times that of Si and nine times that of GaAs (0.55 W/cm·K). The game is how fast you can pull heat away from directly under the channel, and it is why high-power GaN for satellite communications and base stations almost without exception uses SiC substrates.
GaN also has the edge in reliability terms. Whereas the channel-temperature ceiling of GaAs is around 150 °C, many GaN devices are rated for operation up to the 200 °C class. That said, lifetime follows the Arrhenius law and shrinks exponentially with temperature — the correct design posture is not “it still works,” but managing junction temperature backwards from an MTTF target. Outdoor-mounted BUCs are routinely required to operate at ambient temperatures of +60 to +75 °C, so this difference in temperature margin is decisive.
What Actually Changes in a BUC — GaN vs GaAs
Translated into a BUC as a product, going GaN pays off in three ways.
- Miniaturization: Higher power density means a smaller PA for the same output and fewer combining stages. That flows straight into enclosure size and weight — a big deal for portable, vehicle-mount, and maritime antennas.
- Efficiency: Higher operating voltage means lower current, which cuts losses in the wiring and the power-supply chain. Together with the reduction in combining loss, system power consumption drops.
- Linearity, completed by a linearizer: GaN by itself exhibits memory effects rooted in electron traps (current collapse), so its distortion behavior can be less well-behaved than GaAs. This is why practical products assume the combination with a linearizer (or DPD). By canceling distortion with a linearizer, the amplifier can run at a shallower back-off, in the high-efficiency region near saturation.
As a concrete example, the Nisshinbo Micro Devices NJT8336 series (16 W Ku-band BUC) that we handle combines a GaN HEMT with a linearizer to deliver +42 dBm (minimum) output at a power consumption of 110–115 W, with guaranteed operation up to +75 °C. It is a textbook case of the GaN-plus-linearizer design philosophy.
Worked Examples — Pinning It Down with Numbers
Example 1: Power density, device size, and combining loss
Consider a PA delivering 16 W (+42 dBm). With GaN at 5 W/mm, the required gate periphery is about 3.2 mm — a scale that fits on a single chip. With a 0.7 W/mm GaAs pHEMT you need roughly 23 mm, which forces multi-chip power combining. Assuming a combiner loss of 0.5 dB:
Required device power = 16 W × 10^(0.5/10) ≈ 17.9 WIn other words, combining loss alone makes you generate — and throw away — about 12% extra power, with the corresponding increase in DC consumption and heat. Reducing the number of combining stages is itself an efficiency improvement.
Example 2: Estimating junction temperature
Suppose the final-stage PA operates at 16 W output with a PAE of 30%.
DC input = 16 / 0.30 ≈ 53 W
Dissipation = 53 − 16 = 37 W
Tj = Tcase + Rth(j-c) × dissipation
= 85 °C + 2.0 °C/W × 37 W = 159 °CEven under conditions where the case reaches 85 °C in a +75 °C outdoor environment, margin remains against GaN’s channel-temperature ceiling (200 °C class). The same conditions do not close for GaAs, whose ceiling is around 150 °C. As noted above, though, lifetime is an exponential function of temperature: every 10 °C you shave off Tj through heatsink design roughly doubles the MTTF (depending on the activation energy) — worth keeping in mind.
The Structural Shift the Market Is Showing
The RF GaN device market is estimated at roughly US$2 billion as of 2025, with growth in the 10%-plus CAGR range expected toward 2030. The drivers are defense radar, 5G base stations (massive MIMO), and satellite communications. In the satellite field, demand from LEO-constellation ground terminals and gateways is expanding rapidly, and in the tens-to-hundred-watt transmit stages once served by TWTAs (traveling-wave tube amplifiers), replacement by GaN SSPAs is steadily advancing. Unlike vacuum tubes, an SSPA needs no warm-up, lives longer, and runs from a single supply — a good match for LEO ground infrastructure premised on mass deployment and unattended operation.
Takeaways for Designers
The GaN HEMT has moved past the stage of “use it because you need high power” and become a tool for system-level optimization spanning efficiency, size, and reliability. Three points for the designer. First, the essence of GaN is high-voltage, high-power-density operation delivered by its breakdown field. Second, precisely because of that power density, thermal design — GaN-on-SiC, Rth, Tj management — governs both performance and lifetime. Third, linearity is only completed in combination with a linearizer or DPD. The designers who can read not just Psat on the datasheet, but linear output, efficiency, thermal resistance, and the guaranteed temperature range as one coherent picture, are the ones who will carry the ground segment of the LEO era.