600W HF Amplifier
This describes the planning and construction of a 600W HF amplifier for
Ham radio. This project is stalled. I completed the filter/diplexer board and did some low power
testing of the main amplifier board. It is sitting on my work bench, but it may be another year before I finish it.
I wanted more power (especially for voice on 75 meters) and I figured
it would be a good learning experience.
I have an electronics/engineering background, but am fairly new to RF.
The "RF deck" was bought from Communications Concepts. You can buy the
entire kit, or just the PCB or parts.
This design is from Helge Granberg (SK) while he was at Motorola. It of
course uses Motorola (MACOM) transistors
(MRF150 MOSFETs). The kit will get you started, but you still need to
add several sections to make it ready
to get on the air. When I am all done, I should have an amplifier
VERY similar to the Ameritron ALS-600,
but with automatic band switching….. and the wealth of knowledge that I
have hopefully learned in the process.
Most people use a large copper spreader (a thick sheet of copper) to get
heat out of the four MOSFETs.
Cooling has to be one of the most important parts of this project. If
you don't get the heat away from the
transistors, they are going to have a short life span. I agree that
using a copper spreader is required.
Copper conducts heat much better than aluminum and a thick sheet will
get the heat spread out to the sides well.
My first job out of school was at a custom power supply company. My
mentor was an older Englishman who had
been working in power well before I was born. One of his favorite
things to say was "lets put a cooper
spreadah on it" (my best spelling to try and give him an English
accent). We use to harass him about that,
but he was right. If there was a high concentration of heat, the best
way to get that heat out to the
large aluminum heat sink was through a copper spreader. The idea
of the spreader is to pull the heat
out of the almost point source heat producing transistors and get it
distributed to the larger and cheaper
aluminum heat sink. You can see the weight of the copper spreader is
almost 6 pounds in the picture to the right.
I am so-so when it comes to mechanical things. Luckily my build partner
Ed (KI6JMK) is a mechanical engineer (we met at school) so I let him
take care of the mechanical challenges. When we (he) gets farther along
with the mounting and chassis, I will post more details.
Protection will consist of both analog and digital control. It is hard
to beat analog for fast acting
(you don't have to take samples and do math before you change
something). Also, analog is what I am most
familiar with. I have played with microcontrollers both for hobby and
work, but I am a beginner in the
uC area. The uC will shut down the amplifier if the reflected power
becomes too large (greater than 70W-100W)
or if the VSWR becomes too great (4:1) for any power. My hope is to
have good amplifier protection with
the use of both analog and digital control.
The uC also drives the LCD display. I decided that it would be nice to
have amplifier information displayed
on a LCD display. The display doesn't add much cost or complexity to
the project. I really like the pretty vacuum
florescent displays, but I decided on the much less costly LCD
The Wavenode sensor gives a proportional voltage out (both forward and
reverse) to the V on the coax.
I used (Vforward + Vreflected) / (Vforward - Vreflected) to calculate
SWR from the Wavenode data.
The uC uses a 10bit A-D converter. I have not decided if I am going to
add an analog SWR/FWD/REV meter to
the amplifier. You can't beat an analog meter for displaying certain
types of information.
I plan to use the PCBs made by K6IF for the output filters. Although
the PCBs were a little on the expensive
side (compared to hand wiring), I thought it was worth it to have a
circuit board with a nice ground plane
and all ready to accept the inductors/capacitors. The PCB file provided
on his (K6IF) website is in an ExpressPCB format.
You can only have the boards made at ExpressPCB. The output filters
are necessary to knock down the
harmonics of the amplifier. The broadband output transformer of the
amplifier lets harmonics come out the output.
The low pass filter attenuates the harmonics. The push-pull nature
of the amplifier means the even order
harmonics are fairly low to start with (also true for push-pull audio
amplifiers) I read a few articles
on using diplexers instead of the more standard straight low pass
filters. I found the diplexer to be
just a fancy name for a crossover. Like the crossovers used in audio,
the diplexer's job is to send
the highs one way and the lows the other (as the crossover of a 2-way
speaker sends highs to the tweeter
and lows to the woofer). In this case, the crossover sends lows out
to the antenna and highs to a 50
ohm power resistor. The highs are unwanted harmonics and the lows
are the frequencies we want to get
to our antenna. The reason for using a diplexer over a simple low
pass filter is the high frequency
information (the harmonics) can't get out of the amp (we don't want
them on our antenna) so the
energy can get reflected back to the output of the amplifier.....
perhaps being mean to our MOSFETs
(however the Ameritron ALS-600 that is almost identical to the EB104
uses just a low pass filter).
So we take the energy and dump it into a resistor. The amplifier
still sees a 50ohm load
(assuming your antenna/antenna tuner is doing its job). Helge
Granberg also noted that a diplexer can make
the amplifier more stable by presenting it with a 50ohm load over a
wider frequency range.
The K6IF filter PCB was laid out to include relays that can be
controlled to switch the filters in/out.
That means that if your transceiver has a way of communicating it's
band to the amplifier you can use that
information to automatically select the filter. Many Icoms use a band
status output whose DC output changes
with the band. Since I already have a uC in the amp, I plan on using a
single A-D input to monitor this band
info. Several uC outputs will switch the relays. An analog method
could be used for switching, but the digital
method makes more sense. There is not a who is better (analog or
digital), just a who is better for the job
Power supplies are 85% of what I do for a living, so at first I thought
about using a nice active PFC
(power factor corrected) switching power supply that could take
90VAC-264VAC (Universal input).
I also looked at some switchers made for the telecom industry (48VDC) as
that is about the perfect voltage.
Making my own switcher even using a previous design was just too much
work… a project in its own.
Ebay is certainly a good alternative as several thousand watt 48V
supplies can be found for $75 - $200.
However those supplies need 220VAC. That isn't too hard for me since
my electrical box is on one side of
the ham shack, but my partner in crime didn't have easy access to
220VAC and I really wanted to make all
three amplifiers the same. So I decided to go with a linear. That
means big, but easy and quick to make.
Also, linear power supplies get the reputation of being inefficient.
That is true in many conditions….
mostly when you are trying to use a linear regulator and drop a lot
of voltage. Since this supply is not
regulated (there are no active devices trying to keep the output
voltage constant), the efficiency won't
be too bad. The amount the DC output voltage varies is dependant on
the input voltage staying steady
(we have great power in the U.S.), the amount of output capacitance,
and the load (how much power you
put out and for how long). Regulating is a fine thing, but it takes a
lot more work and heat for a
small return (how small of a return is certainly up for discussion).
I looked at some transformers from Toroid of Maryland and Plitron, but
decided to go with the transformer used in
the Ameritron ALS-600. The part number is 406-1246 and I was charged
$59.64 and $41 shipping. That is about
$100 to the door. The toroid option was going to be at least $50 more.
Although a toroid is pretty and has low
magnetic leakage, price ultimately made me go with the Ameritron
transformer. It has a small winding that can
be put in or out of the primary to bump the output voltage up or down.
voltage is one of the big advantages of a toroid since you can simply
add a turn (or several) in the "right" or
"wrong" directions to bump up/down the output voltage. That can be very
hard to do with a standard transformer.
These are the output MOSFETs used in the amplifier. Of course MOSFETs
really shine when they used in
switching power supplies since they are being used as
switches (full on or off). The MOSFETs in our amplifier are working in
their linear region (we hope).
Helge Granberg identifies a few reasons to use MOSFETs in linear
At first I had a hard time locating the MOSFETs at a "reasonable" price,
but eventually I found them at Future
Electronics for $35.22 each. Communication Concepts wants $256 for
four matched MOSFETs ($64 each).
They are matched (meaning their RDS-on should be fairly close between
devices for a given gate to source voltage),
but since each MOSFET is individually biased using a trimpot matching
nearly as important.
I plan to use voltage clamping devices across the transistors (drain to
source) to help protect against voltage
surges caused by high SWR. It is hard to know if the devices can act
fast enough to help MOSFET survivability.
The Wavenode outputs two voltages that represent the forward and
reflected power. The forward power information can
be used to operate an ALC. The ALC control voltage of the amplifier
will connect to the ALC input of the transceiver and tell the
lower the transceivers power when the output approaches maximum output
(probably make ALC kick in at about 500W).
A negative voltage is sent to the transceiver. From what I have read,
there is not an ALC standard so
the output level will
be set with each radio. If the reverse power
becomes too high, the amplifier can be shut down automatically.