[Heaven431]

I have designed an patched array antenna over an electromagnetic band-gap (EBG) surface operating at 3.3–3.8 GHz. Now, I need to design a similar system for the 140 GHz D-band. However, I have never worked at such a high frequency before, so I do not know how to start. I would really appreciate your help.

The array must be fed by a CMOS beamforming transmitter, must support beam-steering requirement of ±45° in both planes. and I need to choose either an on-chip implementation, an antenna-in-package (AiP) implementation, or a hybrid approach. I am using CST Studio Suite for the design and simulation.

I would like to understand the following:

• When I scale my design from 3.3 GHz to 140 GHz, which design principles will remain the same, and what things will I need to change?
• Which radiating element (patch, slot, dipole, dielectric resonator, etc.) should I use for this application, and why would it be suitable?
• If I choose an antenna-in-package (AiP) implementation, which substrate technology should I use, and why would it be suitable for 140 GHz?
• How can I check whether the selected approach will work at 140 GHz? What element spacing should I use for this array, and what port-to-port isolation should I target?
• How should I model conductor loss and the multilayer dielectric stack at 140 GHz? What mesh strategy should I use, and how can I keep the full 16-element simulation computationally manageable in CST?
• How should I characterize and measure the 140 GHz array? Which de-embedding structures should I include, and how can I obtain the gain and radiation pattern? What would be the dominant sources of measurement error?

 

[rainyliveshere]

The jump to 140 GHz changes several practical constraints. Basic array principles still remain largely unchanged. However, losses and fabrication tolerances become critical. You can start with a half-wavelength element spacing. This helps reduce grating lobes during beam steering. For plus minus 45 degree, verify spacing using the scan range. Patch and slot elements are both worth comparing. The package stackup will strongly influence this choice. For measurements, carefully characterize every RF transition. At 140 GHz, connectors can affect measurement accuracy. The coaxial cable connectors guide may help. De-embedding should move the reference plane appropriately. You'd better validate one element before simulating sixteen. Then optimize the complete array using adaptive meshing.