Table 1 Verification structures.

From: Low-cost quasi-global optimization of expensive electromagnetic simulation models by inverse surrogates and response features

 

Case study

Structure I

Structure II

Structure III*

Substrate

εr = 3.5

εr = 3.5

εr = 3.5

h = 0.76 mm

h = 0.76 mm

h = 0.76 mm

Design parameters

x = [l1 l2 l3 w1 w2 w3]T

x = [l1 l2 l3 d w w1]T

x = [l1.1 l1.2 w1.1 w1.2 w1.0 l2.1 l2.2 w2.1 w2.2 w2.0]T

Other parameters

l0 = 30, w0 = 3, s0 = 0.15, o = 5

d1 = d +|w − w1|, d = 1.0, w0 = 1.7, and l0 = 15

win = 1.7, wout = 0.4

Design specifications

Minimize reflection coefficient |S11| at two operating frequencies, 2.5 GHz and 4.8 GHz

Ensure 3 dB power split, |S31| −|S21|= 3 dB, at the operating frequency f0 = 1 GHz, and minimize max{|S11|,|S41|} at f0

Minimize the maximum in-band reflection coefficient |S11| within the range 1.8 GHz to 4.0 GHz

LPW

25

15

30

Simulation time#

60 s

160 s

120 s

Parameter space

l = [15 5 5 0.2 1.5 0.5]T

l = [1.0 5.0 10.0 0.2 0.5 0.2]T

l = [2.0 0.1 0.5 0.1 0.2 2.0 0.1 0.1 0.1 0.2]T

u = [50 15 30 0.6 5.0 5.0]T

u = [6.0 15.0 25.0 1.2 1.5 1.2]T

u = [5.5 0.5 1.1 0.8 1.1 4.5 1.0 0.6 0.6 1.3]T

  1. *Input and output line widths, win and wout, set to ensure 50-Ω input and 100-Ω output impedance.
  2. #EM simulations have been performed on Intel Xeon 2.1 GHz machine with 64 GB RAM.