IEEE Solid-State Circuits Magazine - Winter 2016 - 19

Q-factors of the inductors was poor
(Q of 3 at 1 GHz for 9 nH). I decided
almost immediately to switch my
focus to transformers on silicon
because their potential was promising. My decision was inspired by
Gordon Rabjohn, a colleague from
Bell-Northern who had developed
symmetric transformer baluns as
part of his M.Sc. research with Prof.
Barry Syrett at Carleton. The transformer on silicon proved useful in
RF front-end circuits, and the idea
appealed intuitively to Prof. Copeland when I pitched it to him as a
topic for Ph.D. study. Once agreed, I
developed models for transformers
and inductors on a silicon chip and
benchmarking circuits to validate
their application in the front end of
a radio. A program to extract partial
electrical equivalent circuit (PEEC)
models called GEMCAP had been
developed by Rabjohn and Syrett.
Gordon graciously agreed to let me
hack his original code, and with
Miles's support I developed a scalable, SPICE-compatible PEEC model
that captured silicon substrate losses.
The first transformers (1:1 in noninverting and inverting versions and
a 5:4 symmetric balun) were integrated on silicon BiCMOS [Nortel's
Z31 code in 1992, see Figure 3(a)]. The
model with verification results was
published in early 1995 [2]. Similar
PEEC modeling programs developed
later by others confirmed our results
[4], [5]. A balanced mixer driven by
a 5:4 symmetric balun and low-noise
preamplifier (4:1 step-down feedback
transformer) were also designed in
Nortel's 10-GHz silicon technology
(Z88 code). These 1.9-GHz front-end
benchmarks verified working transformers on silicon in RF circuits
and were reported at the IEEE International Solid-State Circuits Conference in 1995 [3]. The GEMCAP2
model also predicted that the differential drive of on-chip inductors
and transformers would give a large
improvement in Q-factor, which
inspired symmetric inductor prototypes [actually autotransformers,
see Figure 3(b)] integrated at Nortel

in 1994 using IBM's SiGe technology.
Nortel's Hadaway made this possible
by securing early access to IBM-SiGe

the fall of 1995 that he was retiring the following year. His plan was
to literally sail into retirement (on

Prof. Copeland nurtured links between
the university and local industry, which
in the Ottawa region focused primarily
on communications and microelectronics.

for the company. The results of the
inductor modeling and verification
study for Nortel were presented at
the IEEE Custom Integrated Circuits
Confonference in 1996 [6], [7].
My work was similar to that of
Lakshmikumar, in that it straddled
the boundary between semiconductor technology and circuit design,
and our results helped open up the
field of RF circuit design to silicon
implementations.

Post-Carleton
During the TRIO years, Prof. Copeland was forced to devote many
hours to administrative tasks rather
than the technical work he enjoyed.
He reserved one day/week at Nortel for technical study and research,
which was where I normally saw
him as we shared a small office.
Although I could sense that he was
itching for more freedom to explore
new ideas, I was still surprised when
he told me before I left Carleton in

a boat he sailed as a hobby). Miles
didn't stop working entirely, however, and Sorin Voinigescu tells the
story of Miles' post-Carleton work
in his article in this issue, "Of Miles
and Oscillators: Reflections on Miles
A. Copeland."
Looking back, I see that many students mentored by Prof. Copeland
made their own contributions to technology and industry. For example,
Prof. Copeland's original Ph.D. dissertation work focused on electric motors
(radial flux hysteresis machine) and
magnetics (permanent magnets). His
first Ph.D. student at Carleton, M.A.
Rahman, also worked in this area. At
Memorial University, Newfoundland,
Prof. Rahman made pioneering contributions in developing modern permanent magnet motors, including higher
efficiency. The world's first mass-produced hybrid car, the Toyota Prius,
uses Dr. Rahman's permanent magnet
motor drive in the millions of vehicles
produced to date.

(a)

(b)

Figure 3: The first transformer implementations on silicon (three-level metal): (a) a 1:1 transformer
from Nortel in 1992 [2], and (b) a symmetric inductor (autotransformer) integrated at Nortel in 1994
using in IBM's first commercial SiGe technology [6].

IEEE SOLID-STATE CIRCUITS MAGAZINE

W I N T E R 2 0 16

http://www.M.Sc

Table of Contents for the Digital Edition of IEEE Solid-State Circuits Magazine - Winter 2016