IEEE Solid-State Circuits Magazine - Winter 2014 - 12

to write a text book on analog ICs,
based on my lecture notes. The
same day at lunch I found out that
Paul had signed up too - same editor, same book. Rather than write
competing books, we decided we
should do it together and Gray and
Meyer, Analysis and Design of Analog Integrated Circuits, was born.
My research in those days concerned the noise and distortion performance of BJT circuits, and built
on my Ph.D. research experience. I
was fortunate to attract some outstanding graduate students, including Willy Sansen from Belgium who
attacked the problem of BJT agc
amplifier design for wide dynamic
range [3]. Willy enjoyed great success in his Ph.D. work and in his
later career in research and teaching, and ultimately became head of
the Electrical Engineering Department at K.U.Leuven.

Consulting at H.P.
In the early seventies I began acting as a technical consultant on a
regular weekly basis in tandem with
my academic duties. In addition to
the enormous value of the resulting
technical stimulation and research
leads, such consulting helped augment the rather meager salary of an
Assistant Professor. My first position was as a consultant on amplifier design to a new group formed
to design and produce CATV (Community Antenna Television or Cable
TV) amplifiers at Hewlett-Packard
(HP), then still an equipment and
device manufacturer and not the
present-day computer company. HP
was a wonderful company to work
for, either as a full-time employee or
a consultant, and I greatly enjoyed
my association with them. The engineering camaraderie was contagious
and the measurement equipment at
our group's disposal was the stuff of
engineers' dreams. I learned about
S-parameters and microwave design
and found the challenge of a lifetime
in their need for wideband amplifiers (3-300 MHz) with unbelievably
low levels of distortion combined

w I n t E r 2 0 14

Figure 7: The author delivering the
Commencement Address at the College
of Engineering Graduation, University of
Melbourne, 2012 after receiving the degree
of Doctor of Engineering Honoris Causa.

with low noise. This was a fit to my
background made in heaven. The TV
signals of the day were all analog,
and a CATV amplifier typically had to
amplify simultaneously about 50 TV
carriers separated by about 6 MHz in
frequency division multiplex. Interference between channels generated
by the amplifier nonlinearity (which
manifested itself as annoying lines
and fuzz on the TV screen) was the
primary limitation of these amplifiers in the field, and was specified
by crossmodulation (CM) and intermodulation (IM) limits. The most
widely used distortion theory of the
day was based on a simple power
series representation of circuit nonlinearity and predicted a fixed relationship between CM and IM [4]. HP,
however, had a previous generation
of amplifiers in production that had
unexplained major deviations from
this theory. I immediately smelled a
hot research topic and threw myself
into the quest to explain this phenomenon. Naryanan at Bell Labs
had applied the method of Volterra
Series to frequency-dependent distortion analysis of active circuits [5]
in 1967, and this technique seemed

IEEE SOLID-STATE CIRCUITS MAGAZINE

ideally suited to the problem I was
analyzing. Using Volterra Series, I
was able to derive the rather remarkable and compact result that, in a
negative feedback amplifier, the
values of CM and IM as commonly
measured are not always related as
had been previously assumed, but
in fact, the CM becomes a combination of amplitude CM and phase CM
with the amount of each being determined by the phase angle of the
return difference of the FB amplifier.
Fortuitously, when the phase angle
of the return difference was 90°, all
the CM became phase CM, which was
not detected by the amplitude detectors of AM TV receivers. Thus CM all
but disappeared. Since AM CM had
been the major limiting factor in all
prior CATV amplifiers and systems,
this was a major discovery with
significant economic implications.
Also fortuitously, the phase angle
of the loop gain (and return difference) of a well-designed FB amplifier is inherently close to 90° over
most of its frequency range, so this
AM CM reduction is relatively easy
to achieve. At the lowest frequencies in the amplifier passband, the
angle becomes less than 90° but in
those regions the loop gain magnitude itself is very large and all forms
of distortion become very small
anyway. This research was reported
in [6] and was instrumental in my
design of high-performance CATV
amplifiers operating over the 3-300
MHz band with distortion lower than
any prior designs in the business
[7]. I might add here that the Volterra Series approach can be used to
prove the very general and important
result that all forms of distortion in
a feedback amplifier are reduced in
amplitude by the magnitude of the
return difference at the distortion
product frequency.
My research in wideband amplifiers in those days was greatly
influenced by my experience in
designing amplifiers for the CATV
market. Aside from low noise and
extremely low distortion requirements, these amplifiers had to

Table of Contents for the Digital Edition of IEEE Solid-State Circuits Magazine - Winter 2014