WD2XES Transmitter
Linear or switching? Most LF amateur/experimental transmitting is done with high-efficiency Class "D" or "E"
final amplifiers. Such amplifiers are straightforward to construct, and offer excellent power-line to RF output efficiency,
but they are not linear. For on/off keyed CW or FSK, this is not an issue, except for key-clicks. With the low data rates
typically in use, even clicks aren't much of a problem. But for BPSK, HELL, and other modes generated by sound cards, linear
transmitters are significantly cleaner, and may be necessary to fit the signal in the regulatory bandwidth, as well as offer
some relief to other users of the band. Realizing that I was interested in operating with such modes, I considered building a
linear power amplifier. Then I started thinking about the circuitry in MOSFET-output audio amplifiers. Several of the U.K. guys
used such amplifiers on 73 kHz when they had that band, and the LF Experimenter's Handbook had some tips for keeping the
equipment happy.
An AUDIO amplifier? I don't think that Hafler intended such an application, but when I learned that they
rated the power bandwidth of some of their amplifiers at 300 kHz, I checked out a schematic for the P3000. After some head-scratching,
it was obvious that this was a good linear amplifier that would easily cover the 137 kHz range. I snagged one on eBay, figuring
that I would have to make some minor mods before it would be completely suitable. Not so! The Hafler P3000 was my main transmitter
as my LF from March, 2004 to September, 2006 without modification. Not convinced? A (750k) PDF file of the manual and schematic is
available here. In September, 2006, I acquired a used Hafler 9500 on eBay. Jay
Rusgrove, W1VD, had begun using a similar amplifier earlier that year, with excellent results. I have been running (unmodified) it at the 500
watt level since that time, and the P3000 is sitting on a shelf.
Considerations: Tips in the LF Experimenter's Handbook suggested that you could not achieve full power
output from typical MOSFET-output power amplifiers, even at 73 Khz. Since Hafler had rated the P3000 to -3 dB (half power) at
300 kHz, I figured this wouldn't be a major limitation. The suggestion was made that the amplifier be terminated in a load
impedance greater than that used at audio frequencies. By using a 1:2 transformer, I could provide a 50/4 = 12.5 Ohm load,
which seemed to be reasonable for a nominal "8 Ohm" output. Later tests showed that this was correct. It also seemed a good
idea to reduce the output coupling at audio frequencies by using a series capacitor between the amplifier and
the transformer. An earlier choice of a capacitor to resonate inductance seen when looking through the transformer turned out to be
unnecessary after winding the transformer as described below. Comments about output filtering indicated that an inductor-input low pass filter might provide more stability on an amplifier with negative
feedback than a filter with a shunt capacitor at its input. Finally, a 50 Ohm termination on the input would be a good idea,
along with an RF transformer to again limit the low frequency response. Since switching to the Hafler 9500 amplifier, I have kept
the same input and output coupling circuitry.
Results: The schematic above represents a number of changes since the rig first went on the air. The
output filter components and output transformer have been "beefed up" to reduce heating at high power. 500 watt operation
with the Hafler 9500 is entirely practical, assuming that cooling air is supplied to the heat sinks as described in the manual.
I am presently using two 4 3/4 inch AC-motor fans a few inches under the amplifier, each aimed at a heatsink. I also have
two 3 3/4 inch DC-motor fans on the top of the amplifier, centered on the vent holes, and blowing air up. With this arrangement,
the case runs cool at 500 watts continuous key down. I have not attempted to measure temperatures inside the case.
You may want to derate the power if the amp runs very hot with your cooling setup. 300 watts is probably a
"happier" power level in that case. But with any reasonable duty cycle, such as CW, full power is no problem. Note that the present
output transformer does not use windings on opposite sides of the core. Jay Rusgrove did some measurements on different winding styles,
and there is less residual reactance if the windings are either interleaved, or simply wound over each other. You can see the results of
his experiments here.
The input circuit was changed to a 1:2 step-up to raise the input level after I put a bandpass
filter between my DDS exciter and the Hafler amplifier. That filter reduces some spurs from the exciter, and should not
be necessary in most applications.
Testing: Jay Rusgrove, W1VD/WD2XNS, has done considerable testing on this setup, showing that the amplifier
is clean and linear up to 300 watts. Check out his report.
Back to Home