plored by a filter
designer named
Randy Rhea and
his company Circuit
Busters Inc. Their
product Super-Star,
a linear simulator
with two-port and
nodal capability,
and random, gradient and pattern
search optimization, sold for under
$600 (see Figure
3). In comparison,
a seat of Touchstone v1.5 for the
VAX mainframe
was listed at a price
of $13,500 in 1987.
The Super-Star
product would also earn a reputation
for its filter synthesis capability, a feature that would define the attributes
(electrical or physical) of a filter based
on user-specified criteria. The Circuit
Busters company would later be renamed Eagleware.
Meanwhile, under the direction of
Ulrich Rohde, Compact Software was
busy keeping its promise to make vast
improvements to Super Compact. In
1985, according to Rohde, “both the
main frame and PC versions were unstable and distributed models such as
T-junctions, crosses and others were
fairly inaccurate at higher frequencies.” By the end of 1986, the code
had been stabilized for both platforms
and capabilities were being added.
In that year, the company offered
synthesis capability for filters, PLL
and complex matching along with microwave design, RF and communication design kits. Bi-directional control
of network analyzers was available
for the PC version, AUTOART circuit to layout conversion (a feature
introduced during the Comsat days)
was available on both platforms and
the company introduced a number of
new and unique analysis capabilities,
including unrestricted N-dimensional
nodal noise analysis for linear (and
eventually nonlinear) circuits, yield
optimization and user-defined model
capability. Rohde was especially focused on developing the state of the
art in passive model accuracy, touting
this as a significant advantage over his
competitors.
s Fig. 3 1987 ad for SuperStar from Circuit Busters Inc. (
Eagleware).
MODELING PASSIVE ARBITRARY
GEOMETRIES
RF/mW design software from
EEsof, Compact, HP and Eagleware
relied on netlist and schematic entry
(in the case of MDS) to define the circuit topology. Parameterized distributed elements represented the transmission lines and discontinuities that
defined the geometry of the physical
circuit and were known (or assumed)
to impact electrical performance. In
general, these distributed elements
were parameterized by their physical
attributes which, in turn, would impact the equivalent electrical model.
Artwork conversion programs were
used to translate the netlist (or netlists
generated from schematic) into layout
for circuit visualization and mask generation.
Typically, the microwave circuit
designer would progress from ideal
design (wired interconnections and
lumped elements) to replacing the
ideal interconnects and elements
with transmission lines and real element data (measured or other form);
adjust the design to compensate for
the change to the overall performance
(via tuning or optimization); add discontinuity models to sections where
transmission lines intersected and adjust the design once again. Essentially,
the microwave circuit designer, like
previous generations of microwave
engineers, was a plumber. This time,
the tool box was a library of components and the hammer was a set of parameters such as W or L.
The designer’s goal was to approxi-
