|
Power
amplifiers
November 2, 2007
Click
here to go to our main page on amplifiers
Click
here to go to our page on compression point
Click
here to go to instructions for our downloadable (and free!)
Power_Amp_Desingner_101 spreadsheet
Click
here to go to our page on efficiency of active devices
Our apologies, the content on
this page is still pretty weak, but we're moving it up the list
of "things that suck and we need to improve right away".
Where are all the amplifier sponsors out there?
Fellow Microwave Dudes, we sure
could use some pictures to go along with this text! Send us a picture
and you'll receive a cool Microwaves101
pocketknife! Don't make us draw stuff with Microsoft Word, that
gets ugly for everyone.
Power amplifier technologies
Power amplifiers are used to
boost a small signal to a large signal. Power is relative… and frequency
plays a big part in this. Solid state amplifiers and tube-amplifiers
have different connotations when we talk about power amps. The table
below represents the state-of-the-art for different microwave bandwidths.
The info in this table is easily googled from the world wide web
and represents no ITAR-restricted or classified
information, Mr. FBI!
Frequency band
|
Solid state
|
Tube type
|
| L-band through
C-band |
100 watts (LDMOS) |
|
| X-band |
20 watts (GaN
HEMT device) |
3000 watts (TWT) |
| Ka-band |
6 watts (GaAs
PHEMT device) |
1000 watts (klystron) |
| Q-band |
4 watts (GaAs
PHEMT device) |
|
| W-band |
0.5 watts (InP) |
1000 watts (EIKA)
even more! (gyrotron) |
| D-band ? |
|
|
We have a page on semiconductor
tradeoffs, as well as a page on microwave
tubes, check them out!
Peak versus continuous wave
power
Heat that an amplifier dissipates
can be reduced by periodically shutting it off (pulsing it). The
temperature in the active channel follows an exponential decay curve,
at short pulses it looks like a saw tooth, at long pulses it looks
like a square wave. The short pulse situation results in increased
gain and power compared to continuous wave operation. The only truly
accurate way to analyze this is with finite element design tools
such as Fluent.
Let's propose a rule of thumb here (give us some feedback!) A pulse
width of 1 microsecond is considered short and will always result
in improved performance. If your pulse width is 100 microseconds,
the steady-state channel temperature is reached and you won't get
better than CW performance.
Temperature considerations
What happens to an amplifier
over temperature? In the case of a FET amp, the gain drops and the
noise figure increases, all very predictably. Use these temperature
coefficients and a simple Excel spreadsheet and you can model what
happens to your design over temperature:
For gain, use -0.006 dB/stage/degree
Centigrade
For noise figure of an LNA, use
+0.006 dB/degree Centigrade (no need to consider stages in an LNA,
the first stage will dominate the temperature effect).
Note that specific amplifiers
may behave differently from the coefficients we have provided; if
you need to know what happens with extreme accuracy, guess what?
You better get out in the lab and start measuring. One thing to
look out for: if you reduce the temperature of your DUT below the
dew point in your lab, the moisture condensation may cause problems,
especially if you are looking at a hybrid amp with the lid off!
We have a page that ties together
other temperature and thermal effects,
check it out!
Amplifier efficiency
This discussion is now covered
on a separate page.
Amplifier classes
We're working on a spreadsheet that
will help illustrate this, come back in a month or so!
Class A
When a power amplifier is running
in Class A it is biased at close to half of its saturated current.
The output conducts during all 360 degrees of phase of the input
signal sine wave. Class A does not give maximum efficiency, but
provides the best linearity. Drain efficiencies of 50% are possible
in Class A.
Class B
In Class B the power amplifier
is biased at a point where it draws nearly zero DC current; for
a FET, this means that it is biased at pinchoff. During one half
of the input signal sine wave it conducts, but not the other half.
Class B amplifier can be very efficient, with theoretical efficiency
of 80 to 85% depending on FET I-V characteristics. However, you
are also giving up six dB of gain when you move from Class A to
Class B, so if power added efficiency is important to you the optimum
bias point may not be obvious.
Class AB
Most microwave power amplifiers
are operated in a compromise between class A (higher linearity)
and Class B (higher efficiency). In this case, called Class AB,
the output conducts for more than 180 degrees of the input sine
wave, but not over the full 360 degrees.
Class C
Class C occurs when the device
is biased so that the output conducts for even less than 180 degrees
of the input signal. This can be even more efficient than Class
B operation, but the distortion is even worse. And output power
and gain also take a hit. Class C is almost never used in microwave
amplifiers.
Classes D, E and F
Yes these also exist. Mostly
these classes of power amplifier get weirder and weirder, with careful
attention paid to harmonic terminations. If you are designing a
Class F amplifier, you probably don't need any help from Microwaves101!
|