IEEE Solid-State Circuits Magazine - Spring 2016 - 78

Vdd

Output

Vdd
Tail
Current

Tail
Current

Input
(a)

Output

Vdd

CM Bias

CM Bias
Input

Vdd

Output

Bias

CM Bias
Input

(b)

Input
(c)

(d)

FIGURE 2: despite their common characteristics, fundamental differences exist among (a) differential, (b) differential power, (c) pseudodifferential, and (d) push-pull amplifiers.

by simply giving this well-defined
architecture a different name. Doing
so is a disservice to the record established in archive journals and the
general literature. When we respect
the archive journals, we must maintain use of their original definitions
and terminology when describing
identical things.
This misnomer also provides an
ambiguity, contributing to confusion
as to what a differential amplifier
actually is. Shown in Figure 2, the differential amplifier dates back to the
days of vacuum tubes (valves) more
than 50 years ago [18]. Besides having a differential signal on its input,
in CMOS any differential amplifier
has both field-effect transistor (FET)
sources tied together and not directly
connected to any supply or reference.
This allows both transistors to share
feedback through their common
sources, which provides important
properties of the differential amplifier. The push-pull amplifier does
have a differential input signal, but
the feedback path between the transistor sources is eliminated by their
common connection to "ground" (circuit 0 V reference).

Why It Matters
Archive journals matter because
they show how our efforts progress
technology improvements across
time. This map of progress is broken

S P R I N G 2 0 16

RF
Input

Output
Load

FIGURE 3: Signal flow in a CS amplifier
does not allow the input signal power to
flow through the pA output load.

if terminology is arbitrarily changed
at any time with no corresponding change in circuit architecture
or operating principles. That said,
when there is a technical advance
that changes circuit architecture or
operating principles, it is to everyone's advantage to clearly identify
the reasons for, and then to adopt, a
new terminology to clearly identify
the new circuit operation as a differentiation from historical practice.
This makes it equally egregious to
reuse an old name with an established meaning to suddenly (also)
apply to a different operating principle and circuit structure. A modern
classic example of this misuse of an
established name is the term "full
duplex," which has nearly a century
of history [19].

Misunderstanding #3:
Power-Added Efficiency of
Common-Source Amplifiers
The definition of power-added efficiency (PAE) is that it measures how

IEEE SOLID-STATE CIRCUITS MAGAZINE

much power an amplifier adds to the
input signal power which combine to
make the output signal power [20].

Clarification
The concept of PAE comes from the
era when negative resistance diodes
were the only devices that could be
used to provide useful power at radio
and microwave frequencies [20]. There
is no corresponding physical process
that justifies this definition of PAE for
common source (CS) amplifiers. Figure 3 shows that there is no path for
the CS amplifier input signal power to
flow through the load. The input signal power must dissipate within the
CS amplifier. All of the current through
the load flows from the transistor
drain, and so all output power is developed by the drain circuit. Unlike in the
old diode amplifier structures, there
is no power adding happening in any
CS amplifier. Yet we can certainly do
the calculation
PAE =

^POUT - PINh

P DC

(1)

and compare results of this calculation (1), but the result of this calculation has no physical meaning.
This means that all RF input
power to a CS amplifier cannot get
to the load for dissipation there.
Without resistive direct or parasitic
impedances in the CMOS FET input

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