IEEE Solid-States Circuits Magazine - Spring 2020 - 18

Bult's category 2 ideas tend to be obviously
correct and useful the moment one reads
about them but largely unknown until he
publishes them.
Conference (ISSCC) in 1992. The idea
is useful in many applications, such
as current-mode digital-to-analog
converters (DACs), and it works astonishingly well. Yet, even though Bult
explained the concept clearly with a
simplified version of the circuit that
contains only two transistors, the
ISSCC audience had trouble believing it at first. As Chris Mangelsdorf
from Analog Devices recalled, it was
like "cold fusion, in that it had a sensational effect on the conference and
was hotly debated in the halls (not
cold fusion in the sense that it turned
out to be bogus)." Even back in 1992,
it seemed hard to imagine that such
a simple-to-implement and practical
idea could be unknown. Also, people who consider themselves expert
circuit designers were puzzled and
frustrated that they could not immediately understand such a fundamental principle embodied in such
a simple circuit. And the fact that it

caused so many spirited arguments
at ISSCC makes it clear that, unlike
some ideas presented at the conference, it really had not been previously known.
One could be forgiven for thinking, "Well, sure, that was 1992, when
the field was less mature, but the
field is so mature now that there is
no way to come up with fundamental new analog circuit principles."
While this seems like a reasonable
position, Bult has proven it wrong on
multiple occasions. The most recent
example, which also falls into category 1, is the capacitively degenerated dynamic residue amplifier for
pipelined analog-to-digital converters
(ADCs), des-
cribed in the April 2018
issue of IEEE Journal of Solid-State Circuits. The principle identified by Bult
is that capacitive degeneration causes
a dynamic amplifier to first exhibit
an "expanding nonlinearity" and then
a "compressing nonlinearity," and,

The attendees of a Mead Education short course dinner hosted by Vlado Valence in beautiful
Montreux, Switzerland (from left): Behzad Razavi, Richard Schreier, Shanthi Pavan, Ian
Galton, Marcel Pelgrom, Elizabeth Pelgrom, Klaas Bult, Vlado Valence, and Ali Hajimiri.

18	

S P R I N G 2 0 2 0	

IEEE SOLID-STATE CIRCUITS MAGAZINE	

amazingly, if the amplifier output is
sampled at the point in time when
the expanding nonlinearity crosses
over to the compressing nonlinearity, then the sampled value is nearly
devoid of nonlinearity. He experimentally demonstrated that the idea
can yield 100-dB linear amplification.
Bult explained this idea to me in person before he published it, and I was
highly skeptical; it just seemed to me
that achieving 100-dB linearity without feedback or calibration would be
impractical. I could not find a flaw in
his reasoning, but I figured there had
to be one. He subsequently proved me
wrong, and I am still amazed that the
principle works so well.
In contrast, as mentioned previously, Bult's category 2 ideas tend to
be obviously correct and useful the
moment one reads about them but
largely unknown until he publishes
them. A representative example is
the technique his group presented
at ISSCC 2009, which improves highfrequency DAC linearity via trickle
current sources that prevent the cascode transistors in current-steering
DACs from ever switching off. While
the idea seemed obvious to me as
soon as I heard the ISSCC presentation, I had built several current-steering DACs previously without the idea
that would have worked better had I
used it, so I had to grudgingly admit
to myself that the idea really wasn't
obvious-at least to me!
One reason Bult's ideas often seem
like they should have been known
long before he first presented them
is that he explains them so clearly.
Most researchers tend to jump to their
inventions by intuitive leaps and then
retroactively explain their reasoning,
which often causes the papers that
first introduce the inventions to be
confusing and overly complicated.
In such cases, it is often not until
the inventions become widely used
that they get explained more clearly
in subsequent papers, courses, and
textbooks. Hence, if an idea is clearly
explained from the start, it sometimes feels like the concept must have
been known already.
IEEE Solid-States Circuits Magazine - Spring 2020

Table of Contents for the Digital Edition of IEEE Solid-States Circuits Magazine - Spring 2020
