IEEE Solid-State Circuits Magazine - Summer 2016 - 58

Bo
un
Co
da
Bo mp
ry
un res
da sio
ry
n

0.2
0

0.5

Lin

1.5
2
V DS (V)

2.5

Drive Range

0.4

Linear Amplifier

0.6
Knee Pro
file

Drain Current (A)

0.8

Switching Amplifier

Sw
i

tch
ing

ON Resistanc
e

3.5

(a)

V DD
RL
P SAT

V DD
RL
Large V IN

P SAT

Large
V IN

R ON

(b)

0.4
0.2
0
0

file

0.6

Knee Pro

Drain Current (A)

0.8

0
5
0.5

Sw
itc
h in
gA
mp
lifi
er
Ra
ng
e

Lin
ea
rA
mp
lifi
er
Ra
ng
e
Cl
as
s-A
Cl
as
s-A
B

Lin

1
5
2
1.5
V DS (V)

25
2.5

Cu
t-O
ff B
ou
nd
ar
y

ON Resistanc
e

Sw
itc
hin
gB
ou
Co
nd
m
ar
Bo p
y
un res
da sio
ry
n

Figure 1: Driving a transistor hard to access switching operation: (a) the physical actions at
the transistor and (b) the transition to a circuit model which is valid only when the input drive
is sufficiently large.

35
3.5

Figure 2: The transistor characteristic curves and load line is the same as for a linear amplifier, but operation along the load line is now only at its two endpoints instead of linearly
along the load line.

su m m E r 2 0 16

IEEE SOLID-STATE CIRCUITS MAGAZINE

a switching mode. What that means
to the design of a PA is presented in
this article.
Any switch is a two-state circuit
element, where the states are called
OFF and ON. We require this behavior
also in any switching PA: within each
RF cycle, the PA transistor must spend
the majority of its time in both of the
steady-state conditions of OFF (transistor cut-off) and ON (transistor a resistor). At no time should the transistor
operate as a current source. Looking at
the transistor characteristic curves in
Figure 1, the ON condition is to the left
of the knee voltage profile, at the farleft edge of its characteristic curves.
The drive signal must be increased
from the maximum available for linear amplifier operation so that the
transistor is forced into acting as a
resistor. Drain (load) current increases
slightly from that at the linear compression boundary, showing that the
output power increases slightly. The
input drive magnitude increases much
more, which directly means that the
SMPA gain must be lower than that of
its linear value.
A more complete view of how
switching operation of any amplifier
relates to linear operation of that same
transistor is seen in Figure 2. The output signal range from a switching amplifier is greater than that of a linear
amplifier using the same transistor.
This is a consequence of our not caring about, or needing, any amplifier
circuit linearity, and therefore there
is no need to keep the output voltage
above the transistor knee voltage. Indeed, there cannot be any circuit linearity in an SMPA. Small input drive
magnitudes are not possible, as emphasized in Figure 1. When the transistor spends time only at the load
line endpoints, these intersect very
low value power dissipation curves,
something emphasized in Figure 2.
This is the necessary condition for
high energy efficiency as pointed out
in Part 1 of this series.
When the transistor operates as a
switch, it is impossible to have a sinusoidal output waveform. The output
waveform must be a square wave,

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