for learning

1. Modern approaches to envelope restoration usually dispense with the limiter as
a means of generating the necessary constant amplitude,
1
phase modulated signal. In
a modern system, it is now the current trend to assume that the necessary phase and
amplitude modulation drive signals can be generated directly in the system digital
processor, giving a basic system block diagram as shown in Figure 10.21.2
Confus-
ingly, such modern derivatives of the Kahn EER technique have become known as
“Polar” PA systems, having some generic similarity to the “Polar Loop” feedback
linearization system (see Chapter 14), but with the notable distinction that in some
cases they remain open-loop configurations.
ER systems present a number of challenges, in addition to the obvious one of
realizing a high efficiency power converter. The assumption that the phase modula-
tion on the input to the nonlinear RFPA will be preserved is quickly invalidated
when the supply is varied, and there is a serious issue of dynamic range when
attempting to implement ER for a zero-crossing signal. The latter problem has
potential solutions, such as the use of tracking bias controls on the driver stages.
Linearity problems, as always, can in principle be handled using DSP techniques,
however this introduces some serious overheads to the cost and complexity of the
system. In particular, the need to calibrate each individual RFPA device characteris-
tics is a detraction from the ER and Polar techniques.
10.6 Envelope Tracking
In the EER configurations considered in the previous section, the amplitude modu-
lation was created by modulating the supply to a well-saturated RFPA stage. This
requires great accuracy in the generation of a suitable supply voltage and may have
dynamic range limitations. There are some interesting potential benefits in applying
10.6 Envelope Tracking 311
Figure 10.21 “Polar” modulation system.
1. Indeed, it is not clear as to how Kahn realized this function for zero-crossing modulation systems; his much
quoted publications gave no details on this critical item.
2. It is not clear why such systems are still given the generic label of “EER,” since there is neither elimination
nor restoration of the envelope, it is simply “constructed” by suitable amplitude modulation of the supply to
the RFPA. Replacing “EER” with “C” does not appear to have much support at present, and “ER” would
seem to be a good compromise in terminology.

2. a similar envelope-derived modulation to the supply voltage of a conventional linear
RF amplifier. Such a technique is generally termed “Envelope Tracking,” or “ET,”
and is shown schematically in Figure 10.22.
Recalling the analysis of a conventional Class B amplifier (Figure 10.1), we have
already seen that if the RF load resistor is decreased in inverse proportion to increas-
ing drive voltage amplitude, the efficiency remains constant at its maximum value
while the power increases as the square root of the drive power; this is an important
mode of dynamic behavior in the Doherty amplifier (Section 10.2). An alternative,
and considerably simpler, scenario is that the load resistor remains fixed and the
supply voltage is increased in proportion to the increasing drive voltage. In this case,
maximum efficiency is maintained (due to full rail-to-rail voltage swing) and the
output power increases linearly with input drive power. This process can be contin-
ued down to a selected point on the lower side, and up to a point where the RF swing
reaches breakdown level. In this manner, maximum efficiency can be maintained
over a wide linear power range.
One of the appealing aspects of this technique is that the modulation control
voltage does not have to replicate the signal envelope with great accuracy, as is the
case in the ER process. For example, the supply voltage could be tracked for just the
upper few dB of the signal envelope range; this would then show an overall PBO effi-
ciency characteristic somewhat comparable to a Doherty PA. The still-present chal-
lenge of the tracking power converter can also be further reduced by the use of a
power supply having two or more discrete switched output voltage levels. Problems
of dynamic range, particularly in zero-crossing signal environments, are essentially
eliminated by allowing for normal linear operation at a reduced supply voltage in
the small signal regime. An additional attraction of ET is that the efficiency enhance-
ment process is completely decoupled
3
from the RF matching. This is in contrast to
the Doherty and Chireix configurations, which depend heavily for their operation
on resonant RF circuit elements.
It is instructive to consider a simple case, where two supply voltages are avail-
able, as shown in Figure 10.23. The supply switch can, in principle, be realized using
low cost and readily available semiconductors, which for envelope speeds in the
MHz range do not have to be exceptionally fast, by modern standards. The drive
signal to the switch would conventionally be derived by using an envelope detector
312 Efficiency Enhancement Techniques
RF IN
DC
RF OUT
Pwr. cond
Figure 10.22 Envelope Tracking (“ET”) RF PA system.
3. Both literally and metaphorically; the physical integration of power converter and RFPA circuitry is one of
the less publicized challenges in implementing ER systems.

3. and a threshold circuit which sets the level at which the switch was activated. A
modern implementation would probably use a drive signal generated by the system
DSP, which could conveniently be time-synchronized with appropriate pre-
distortion of the input RF signal. This will include necessary compensation for the
small changes in gain and phase in the RFPA as the supply is switched. The “break-
point,” or envelope level at which the supply is switched, is clearly an important
design choice. This choice will have an optimum value, dependent on the statistics
of the signal environment. As with the selection of the breakpoint in a Doherty PA,
the intuitive choice would be to set the lower voltage to switch in at the mean power
level, the higher supply being used up to the peak power. The efficiency characteris-
tic shown in Figure 10.24 assumes that the PA follows the ideal Class B PBO/effi-
ciency characteristic for each supply voltage setting, reaching a maximum value of
70% at the onset of envelope clipping in each case.
In order to evaluate the impact of such a characteristic in a specific signal envi-
ronment, it is necessary to generate a representative signal burst and evaluate the
average efficiency. This process can be repeated for various breakpoint and PEP
backoff settings. For example, Figure 10.25 shows the results from performing this
exercise for a two-level supply voltage on an EDGE signal. The average efficiency
10.6 Envelope Tracking 313
RF IN
Switch drive
DC
RF OUT
DC-DC conv.
Figure 10.23 ET System using a two-level switched supply.
100
50
0
-10 dB-20 dB Pmax
Efficiency
(%)
Output power
backoff (dB)
Vss
Vss/2Vss/3
Figure 10.24 PBO-Efficiency characteristic for two-level switched ET system.

4. for the two-level supply switch is plotted for a range of breakpoint values. It is clear
that a simple two-level switch, in this signal environment, can give a useful increase
in efficiency (53.5% at optimum break point) in comparison to a standard Class B
PA (45.3%), but falls significantly short of the Doherty PA result for the same signal
environment (60.3%). In the case of an EDGE signal, it will in practice be necessary
to back the PEP level down from the clipping point in order to obtain acceptable
ACP performance, this will reduce all of the efficiency results by a similar factor.
The efficiency gains in Figure 10.25 seem to deliver somewhat less than the PBO effi-
ciency curves might suggest. The returns are, however, quite critically dependent on
the signal environment. Some further details of the method used to perform the effi-
ciency computations contained in Figure 10.25, and some more simulation results,
will be presented in Section 10.10.
As is customary when promoting efficiency enhancement techniques, little has
been said about the possible effects of the technique on the PA linearity. As far as
envelope tracking is concerned, this has to be raised as a potentially serious issue.
The next chapter in this book, Chapter 11, is largely devoted to minimizing the
harmful effects of small amounts of unintended modulation on the supply voltage to
an RFPA. Clearly, when it is proposed that this same supply voltage be intentionally
yanked around over most of its allowable range in order to improve the efficiency,
questions about linearity have to be addressed. The standard answer, as always in
the modern era, is that DSP will save the day. The gain and phase variations which
the tracking will create can in principle be characterized and compensated by suit-
able adjustment (DPD) to the input signal. But there can be little doubt that the large
supply variations will make the DPD task more challenging.
10.7 Power Converters for EER and ET
The implementation of a high efficiency, broadband power converter design for
EER and ET implementation in RFPAs has been the focus of much research activity
314 Efficiency Enhancement Techniques
Break point (normalized)
Efficiency
(%)
100
50
0
0 0.5 1.0
Doherty (6 dB breakpoint)
Figure 10.25 Average efficiency for EDGE signal, Class B PA with switched voltage supply;
(switch break point plotted as x-axis). Doherty PA efficiency shown for standard 6 dB break point.