IEEE Solid-State Circuits Magazine - Summer 2016 - 59

Any switch is a two-state circuit element, where
the states are called OFF and ON.
magnitude is shown by the phasor and its circle. When the bias is
adjusted to class-AB, the drive magnitude needs to be increased to attain
the same drive into compression for
the SMPA ON state. This is shown by
the phasor anchored at the AB bias
point and its associated larger circle.
Also in Figure 4 is a class-C bias situation, where the input drive magnitude

has to increase still further. It is interesting to note that, for one of the most
widespread historical uses of SMPA
operation, plate-modulated amplitude
modulation broadcast transmitters,
the bias was always set to class-C
[4]. As Figure 4 shows, this decision
forced the use of larger magnitude
drive waveforms than the minimum
available for SMPA operation.

4
V OUT

Power (Watts); Signal (Votts)

3.5
3

P LOAD

2.5
2
1.5

V IN

P FET

1
0.5
0
0

0.2

0.4

0.6

1.2
0.8
1
Time (Cycles)
(a)

1.4

1.6

1.8

(Square Wave)
Amplifier Characteristic (dB)

similar to that of VOUT shown in Figure 3(a). This particular waveform is
not completely square to allow the
consequence of extended transition
times to be observed. Since the control input to the transistor remains
at the gate, the amplifier must still
transition along the load line between
its ON and OFF states. This path intersects power dissipation contours of
higher values, as seen in Figure 2, and
travel along this path results in the
"ears" on the transistor power dissipation curve ^PFETh seen in Figure 3(a).
The time integral of this power dissipation is lost energy, which is something we want to avoid. It is for this
reason that we desire very fast transition times in an SMPA. The result is
C-mode operation, described in Part 3
of this series [2].
The power transfer function in Figure 3(b) shows that the output power
(at the fundamental frequency) from
an amplifier continues to increase as
the output waveform is increasingly
distorted by amplifier compression.
There are two reasons for this. First
is the rms power in a square wave is
higher than that of a sine wave with
identical peak-to-peak amplitude,
and, in particular, the Fourier component at the fundamental frequency
being 4/r (approximately 2 dB)
larger than the square-wave amplitude. Adding to this is the greater
available SMPA square-wave peak-topeak amplitude than the maximum
available linear sine wave peak to
peak amplitude seen in Figure 2. It
is very important to understand that
this increase in output power is a
result of waveform distortion due to
compression. Circuit operation here is
not linear at all, and any multicarrier
signal that operates in this region is
subject to intermodulation distortion
as required by the Fourier transform.
Using the approach introduced in
Part 2 of this series [3], the relation
between bias settings for the SMPA
and their associated drive magnitude
requirements is illustrated in Figure 4.
Starting from class-A bias (midpoint
between the cut-off and compression boundaries) the necessary drive

PSAT

0
-5

Ratiometric Gain
Fund. Output Power

-10
-15
-20
-25
0.01

Linear
(Sine Wave)

Clipping

1
0.1
Input Voltage (Norm. to Clipping Onset)
(b)

Figure 3: Driving the output waveform into a square shape is necessary for switching PA
operation. (a) The transition between the on and oFF states still happens along the load line,
resulting in "ears" on the field-effect transmitter power dissipation waveform (PFeT). (b) As the
output waveform is increasingly distorted into a square wave, output power at the fundamental frequency continues to increase.

IEEE SOLID-STATE CIRCUITS MAGAZINE

su m m E r 2 0 16

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