IEEE Solid-State Circuits Magazine - Winter 2016 - 23

needed absolute value accuracy for
the resistor-capacitor product 50
times better, on the order of 1%.
We continued to move the yardsticks along on this problem, but it
was Miles Copeland that changed
the game for us, and he did it by
going back to first principles. Miles
deduced that the R-C product rather
than the absolute value of the components was the key. Since Nortel's
silicon process could maintain a
ratio better than 0.2% between like
component values, why not make a
resistor look like a capacitor, or vice
versa? He further postulated that
by charging and then discharging
a capacitor (C sw) at a fixed rate (T ),
that this "switched capacitor" would
synthesize a resistor (R eq) that varied inversely in proportion to the
(C sw) value according to
R eq = T/C sw,
making the R eq-C product
R eq-C product = (C/C sw) # T.
Therefore, semiconductor process
variability was limited to just the
small change in the ratio between
capacitor C and the switched capacitor C sw used to synthesize R eq in
the critical R-C product.
Prof. Copeland developed two
different structures to move charge
(i.e., a current) from point A to
point B. Of course, neither of these
approaches were without further difficulties to implement on a chip, but
he had certainly "cracked the shell"
of the problem.
A test chip (Figures 2 and 3) was
quickly fabricated using Nortel's

double-poly process and analyzed
by Dr. Copeland and Carleton University graduate student Choudury
Rahim. Traditional RC filters in the
audio frequency range were replaced
with fully integrated MOS large-scale
ICs using the new prototype. It was
found that filter characteristics
such as gain, cut-off frequency, or
resonant frequency could be varied
by changing the switched-capacitor
sampling clock rate (1/T ), capacitor values, or both. Moreover, it was
found that conventional theory and
design methods normally used in
RC filters could be directly applied
to switch capacitor filter designs. We
only needed knowledge of the parallels between sampled-analog filtering (analyzed using Z-transform
methods) and continuous-time analog filtering.
The results from this test chip,
titled "Sampled Analog Filtering
Using Switched Capacitors as Resistor Equivalents," was published in
the December 1977 issue of IEEE
Journal of Solid-State Circuits (JSSC)
in a joint publication by Carleton
researchers and Nortel staff [1].
This paper is still cited to today as
a seminal paper in switched-capacitor circuits. Independently, another
first switched-capacitor filter paper
was also published in the same JSSC
issue, "MOS Sampled Data Recursive
Filters Using Switched Capacitor
Integrators" [2].
Prof. Copeland and his graduate
students were also instrumental in
solving many other problems with
switched-capacitor circuits and their
implementation, such as the associated Nyquist frequency problem,
developing a capacitor shape that

maintained its ratio to other capacitors despite etching and processing
variations, assisting in the development of low-noise amplifiers, and
maintaining a small footprint for the
circuits on a silicon chip.
Figure 4 shows Nortel's secondgeneration filter codec chip (code
E99) that uses the switch-capacitor

Figure 2: A schematic of the first
switched-capacitor test chip.

Figure 3: The first switched-capacitor test
structure.

Figure 4: Nortel's E99 filter codec.

IEEE SOLID-STATE CIRCUITS MAGAZINE

W I N T E R 2 0 16

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