Passive detection of millimeter wave (PMMW) radiation for imaging holds great promise as a means to see concealed weapons and to aid aviation in low visibility conditions [1]. This requires compact cameras sensitive to the pW radiation typical of most scenes. Our passive millimeter wave (PMMW) cameras are based on 2D planar arrays of total power radiometers serving as the imaging focal plane array (FPA). The overall architecture includes a set of imaging lenses that focuses the incoming MMW radiation onto the FPA. Each receiver includes an LNA followed by a rectifying diode to convert the received signal into a video rate voltage whose amplitude is proportional to the power received by the element (Fig. 32.3.1). The final image is generated by assigning a gray scale to the voltage and plotting the 2-D array of data. Image quality depends strongly on receiver sensitivity, of which noise figure, bandwidth, and video rate integration time are critical parameters. Figure 32.3.2 shows two 94GHz images. The imaging of the camera is diffraction limited (the angular resolution achievable by the camera is inversely proportional to the product of the aperture diameter and the frequency of the radiation being imaged). Thus, to maintain the angular resolution while using a smaller lens, the operating frequency must be increased.Reported here are amplifiers, using 70nm and 35nm InP HEMT transistors, for PMMW imaging made possible by these technologies. Using state-of-the-art III-V HEMT MMIC technologies with increasing speed [2,4,7], a series of LNAs have been developed at frequencies ranging from 90GHz to 200GHz for the PMMW camera [3,5,6,8], as detailed in Fig. 32.3.3.More recently, a cascode LNA has been designed using our 70nm InP HEMT MMIC process (Fig. 32.3.4). The cascode topology demonstrates higher gain compared to a common source, which is advantageous for high frequency applications where available gain is limited [9]. Additionally, conductor losses increase with frequency, while device gain is harder to achieve. Generally, cascode stages do not demonstrate as low noise figure as commonsource stages, so an optimum amplifier topology is to cascade the first stages as low noise common source stages followed by high gain/stage cascode stages to achieve the best overall performance for these high frequency LNAs.The cascode design uses two-finger devices with a periphery of 30µm. A low-impedance transmission line at the gate of the common-source device provides a modest amount of matching. This single-stage cascode design is used for capability evaluation and model extraction. A shunt stub with an RC network at the drain provides stabilization at the output. Measured performance of the circuit is shown in Fig. 32.3.5. The drops in gain at 170GHz and 200GHz are attributed to "suck-outs" occurring from coupling to adjacent structures in the on-wafer measurement. However, the measured circuit performance illustrates the benefit (a higher gain per stage) of the cascode design at high frequencies. For example, 8dB gain from ...
