Access Official Technical PDF

Share Summary:LinkedInEmail

Extending low-noise amplification across a wide millimeter-wave band is a demanding receiver-design problem. A proposed InP HEMT LNA addresses the 125–211 GHz range with a four-stage monolithic microwave integrated circuit (MMIC) designed for radio astronomy. Simulations indicate low noise across the band, but the design remains to be validated through fabrication and measurement. Its results therefore offer a promising engineering direction—not yet a demonstrated receiver replacement.

Why consider an LNA for 125–211 GHz?

Superconductor-insulator-superconductor (SIS) mixers have been widely used in millimeter- and submillimeter-wave astronomy, including across the frequency ranges served by ALMA. They provide low-noise reception, but require operation at cryogenic temperatures around 4 K. High electron mobility transistor (HEMT) amplifiers offer another approach: advances in transistor materials and device geometry have improved their high-frequency gain and noise performance.

Indium phosphide (InP) HEMTs are of particular interest because of their high electron mobility. Earlier work on cryogenically cooled InP devices has demonstrated noise performance close to several times the quantum limit. The design considered here applies a 35 nm gate-length InP HEMT process to a continuous 125–211 GHz operating range, which spans the frequencies of two adjacent ALMA receiver bands.

InP HEMT LNA design and stage allocation

The MMIC uses four common-source transistor stages. Each transistor is a two-finger device with a 10 μm gate width, selected for its noise performance across the intended band. Microstrip transmission lines, resistors, and capacitors form the matching networks and bias circuits. Series blocking capacitors at the input and output prevent DC bias signals from passing through the circuit interfaces, while a resistor in the output matching network contributes to stabilization.

The stages have different priorities because noise contributions in a cascade depend on the gain of the preceding stages, as described by the Friis equation. The first stage is optimized for low noise. The second is also designed with noise in mind: at these frequencies, the gain from one stage may not be enough to make later-stage noise negligible. The third and fourth stages focus more on flattening gain and reducing output reflections. Independent gate and drain bias lines allow each stage to be tuned.

Simulations were conducted at a physical temperature of 20 K, with electromagnetic analysis of the passive matching networks. This distinction matters: the reported figures are simulated circuit performance, not measurements from a fabricated amplifier or a complete receiver.

Simulated noise, gain, and stability

Across 125–211 GHz, the simulated noise temperature stays below 58 K, with a reported average of 38.8 K. At 170 GHz, that average corresponds to less than five times the quantum limit. The minimum simulated noise temperature is 34.6 K. Results also meet the stated noise targets for the two sub-bands: below 40 K across 80% of 125–163 GHz, and below 45 K across 80% of 163–211 GHz.

Average forward transmission (S21) is 20.5 dB, with variation of ±0.85 dB. That is below the 35–40 dB gain target in the proposed specification. The paper suggests that two LNA modules connected through a microwave isolator could provide the required gain; the isolator would also help prevent reflections between modules from producing excessive gain ripple. This is a proposed system configuration, not a measured result for a completed two-module receiver.

Other simulated results indicate input and output reflection coefficients better than −6 dB and −12 dB, respectively, across the band. Reverse transmission (S12) reaches −47 dB at 211 GHz. The Rollett stability factor remains above 7 from 0 to 400 GHz, indicating unconditional stability in the simulation.

What the results mean for receiver integration

The noise performance suggests that an LNA-based receiver could be competitive with SIS mixer systems over this frequency range. A single broadband receiver could potentially cover the span now divided between ALMA Bands 4 and 5. The proposed system-level rationale is that consolidating two 4 K receiver coolers into one operating at 15 K could reduce cooling-related operational costs and free a slot in the front-end cryogenic system.

Those benefits depend on performance at the receiver level, not just the MMIC. A fair comparison must account for noise contributions from components such as the feedhorn, orthomode transducer (OMT), and coupling optics. The amplifier would need to be assembled in a suitable cryostat and evaluated as part of the receiver chain before system-level performance or operational benefits could be established.

Fabrication and validation are the next steps

The proposed MMIC was intended for a future wafer run using the 35 nm InP process. Planned validation includes on-wafer and in-module measurements of S-parameters, noise figure, and linearity, including 1 dB compression point and third-order intercept, as well as power dissipation. In-module testing also requires attention to ancillary components and the waveguide-to-microstrip transition, which must transmit the input signal effectively.

For engineers and system integrators, the central takeaway is the difference between promising device simulations and a qualified receiver. The design shows a path to low-noise, broadband amplification from 125 to 211 GHz, with modeled unconditional stability. Fabrication, cryogenic measurement, and full receiver integration will determine whether that potential translates into practical performance. If validated, the approach could provide an alternative architecture for broadband millimeter-wave astronomy receivers.

Access Official Technical PDF

About The Author