Compact Hierarchical Bipolar Transistor Modeling With HiCUM

Determining the base resistance of InP HBTs: An evaluation of methods and structures

“…Using two tetrode structures with different emitter lengths allows the elimination of any impact of the foreside and edges by 1 The latter dependence is valid for b E0 ( l E0 and was implemented in the compact model for the purposes of this comparison, but will differ if b E0 approaches l E0 [26,[33][34][35]. 2 In fact, this has been the case in many SiGe BiCMOS foundries for many years.…”

“…where g i is a geometry function ranging from 1/12 for a transistor with two long double-base stripes to 1/28.6 for a square-emitter transistor with surrounding base contact according to [26,[33][34][35]. w is the DC emitter current crowding function; its value equals 1 for sufficiently low forward bias and is very close to 1 even at higher bias in transistors used for high-frequency applications.…”

“…This still leaves the question open how accurate either of the methods actually is. Therefore, in order to be able to evaluate the absolute (achievable) accuracy, in this study the extraction methods under investigation will be applied to simulation data generated by the physics-based compact model HICUM [26], which has been shown to accurately model InP DHBTs [27]. The advantages of this approach are:…”

Correlated noise in bipolar transistors: Model implementation issues

“…where a B qf and a B it are NQS model parameters and s B f (denoted as s Bf in [7]) is the transit time through the quasi-neutral base. Taylor expansions of the formulations given in (17) up to the first nonzero terms yield for the elements of (14)…”

“…However, proper guidelines are laid down in [5] for correlated noise implementation in Verilog-A. The diagonalization method of [4] was applied in [6] to implement noise correlation in the HICUM model [7]. Simulation results, however, led to excessively large noise levels in some cases [8] due to an erroneous x 4 term.…”