Study on Broadband Millimeter-Wave 2-Stage Power Amplifier Design in 40-nm GaN pHEMTs

We present two design examples with measurement data to highlight the benefits of using a stacked topology over a common source (CS) topology for broadband millimeter-wave (mm-Wave) power amplifier (PA) design. These medium power PAs are designed using a low-voltage state-of-the-art 40-nm gallium nitride (GaN) pseudo-high-electron-mobility transistor (pHEMT) process. Two PAs are designed and fabricated, one using a CS driver with another CS second stage (i.e., CS/CS PA), and the other using a CS driver and a 2-stack second stage (i.e., CS/2-stack PA). Their measurement data is compared to post-layout extracted (PEX) electromagnetic (EM) simulations, and also with other state-of-the-art designs found in literature. As expected, the CS/2-stack PA can achieve considerably higher power gain than a similar 2-stage CS/CS PA with similar power-added-efficiency (PAE). Good agreement on measured vs. PEX EM simulated results have been observed. However, performance degradation was discovered in measurement for high drain bias voltage not shown in simulations. Measurements taken on the CS/CS PA at 37/41/44 GHz yielded saturated output power (P
OUT,SAT
) of 19.7/18.9/19.1 dBm with gain > 13 dB, with 3-dB bandwidth (BW) of ~ 21% from 37 – 45.6 GHz On the other hand, the CS/2-stack PA attained measured P
OUT,SAT
of 23.2/24.6/23.2/23.3 dBm with over 20 dB gain at 21/24/28/31 GHz, and with a much wider 3-dB BW (i.e., ~ 55% from 19.9 – 35.1 GHz). As the DC bias voltage, V
DD
, on the 2-stack PA increased from 20 V to 24 V, performance degradation was observed in measurement repeatedly, even though the P
OUT,SAT
is still only less than half a Watt. Therefore, we need to be more conscious of the DC breakdown limits when utilizing these low-voltage GaN transistors for mm-Wave PA design.