Abstract: This article discusses the design of a simple Class-E switching-type LowFER transmitter.
[Reproduced from Western Update #59, September 1988, by permission from the author]
My LowFER beacon, MPM, has been on the air for some time, utilizing a dual MOSFET push-pull transmitter design. Despite it's complexity I've been quite proud of the circuit, which I considered to be hard to beat for DC to RF efficiency. Because I have received several requests for a description of the circuit, and since efficiency has recently become a "hot" topic, I had planned on writing a short article on the transmitter.
However, no sooner had I began than I ran across Frank Cathell's article (WU #58) on Mike Mideke's "simple beacon". I'd previously glanced over Mike's circuit without being particularly impressed: "Maybe 50% efficiency on a good day" I thought to myself. But Frank's article started me thinking that maybe the efficiency could be pretty good if the active device could be operated as a switch. I decided to do a rigorous analysis of the circuit, optimizing component values, and then check the results with a circuit analysis computer program. The results were encouraging, so I went ahead and built a prototype and WOW! - was I ever impressed!!! Not only is it considerably simpler than my push-pull design, the efficiency is higher! As a result, I no longer recommend my push-pull design! However, my analysis uncovered a few surprising characteristics (as well as some minor errors in Frank's article) so, since I still have this ego-driven desire to write something, I thought I'd share my findings with everyone:
As Frank's article pointed out, operating the device as a switch can theoretically achieve 100% efficiency. What originally frightened me about this circuit was that the switch was directly in parallel with the tank circuit. Any voltage on the capacitor when the switch is turned on would be discharged by the switch, dissipating the capacitor energy and wasting power. Since the MOSFET drain voltage oscillates about the supply voltage (rather than ground) when the switch opens, it would not normally return to zero volts after one-half cycle. To remedy this situation, the tank circuit should be DETUNED so that it oscillates for more than one-half cycle while the switch is opened. It turns out that if the tank is made resonant at 1.2915 times the operating frequency (assuming a 50% switch duty cycle) the voltage on the drain will return to very nearly zero volts when the MOSFET turns on. (Contrary to Frank's article, a 50% duty cycle is not necessary - any duty cycle may be used, the only difference being that the detuning factor will change. However, since 50% is easily produced, it is used throughout this analysis.)
It is also desirable for high efficiency that the current in the
be zero immediately following turn-on: This tends to
MOSFET current for a given power output. This will occur if
is completely discharged during the MOSFET "off" period, i.e., all
stored in the inductor is transferred to the load.
For a given supply voltage, frequency, and power output, it turns
that there is an ideal inductance value and load resistance which
the above condition. When these values are combined with the
for tank resonance above, we obtain component values and and load
for our "optimized" circuit:
|L = tank inductance (Henries)||Z = load resistance (Ohms)|
|C = tank capacitance (Farads)||P = output (or input) power (Watts)|
|V = supply voltage||F = operating frequency (Hertz)|
|Vmax = peak voltage on MOSFET = 3.6311 * V|
|Irms = rms MOSFET current = 1.1638 P / V|
The above equations (Figure 1) are the result of several hours of slaving over a table of Laplace Transforms, and numerical equation solving using a programmable calculator. Since I easily could have made an error in the calculations I decided to check my results using a circuit analysis computer program I have access to at work. I first calculated component values using the above equations, for P = 1 watt, V = 12 volts, and F = 179,000 Hz. These values worked out to:
|For reference, the values for L, C, and Z referred to
(i.e. "Simple Beacon" schematic) are as follows:
L = 75 uH
C = 0.015 uF and
Z = 72 ohms
Again, note that these values are NOT derived from the equation in figure 1.
L = 167.7 uH
C = 2826 pF
Z = 182 ohms
The analysis was run and the output voltage and switch current were plotted (see Figure 3.) Plots were made for both no load and full load. The output voltage closely approximates a half cycle sinewave at no load, but departs somewhat from sinusoidal at full load. The full load MOSFET current is seen to be zero when the switch first closes, as desired. Note also that the no- load MOSFET current is initially negative and ramps positive, with the result that the average current is essentially zero, which would be expected for no load.
Encouraged by these results, I decided to build a real working
stage. I picked up an IRF-511 MOSFET at Radio Shack, dug
my junk box and found a T-150-2 powdered iron toroid core, and
several mica capacitors from work which totaled about 2800
added turns to the toroid until it resonated with the caps at 231
times 179 kHz) - this required 105 turns of #30 wire. I
gate of the MOSFET with a 5 volt 50% duty cycle square wave from a
generator. For a dummy load, I used a 150 ohm resistor in
with a 3 mH inductor and 300 pF variable capacitor - the
this combination was very nearly 182 ohms at resonance.
I first powered up the circuit with no load, and observed a
identical with the no load output voltage plot. The supply
was about 1.2 mA. I then connected the load, and by tuning
cap obtained a waveform identical to the full load plot, with a
current of 83 mA. Using the 'scope I then measured the load
and calculated the output power and efficiency, which was an
The highest efficiency I ever measured with my push-pull design
The circuit was well behaved with changes in tuning, with maximum
and maximum current drain occurring very nearly together.
A tune-up procedure which works well is as follows: With no
the tank is adjusted for minimum supply current (this adjustment
too critical and the circuit will work well over a +/- 10 kHz
fixed values). The antenna is then connected and the loading
tuned for maximum RF. Since most LowFER antennas have a
of less than 182 ohms some means of impedance transformation is
Probably the simplest way of doing this is to connect the antenna
coil to a tap on the tank inductor. The position of
should be adjusted so the transmitter draws 1 watt DC (83 ma for a
supply) when the loading coil is tuned for max. RF output.
much current is drawn, tap the inductor farther from the MOSFET
These numbers may look scary, especially the 2nd, but shouldn't
cause for alarm. A LowFER antenna with a Q of 100 (a rather
value) will provide 37.5 dB attenuation of the second harmonic, so
radiated second harmonic will be over 43 dB below the fundamental.
For proper circuit operation, the load impedance at all
should be high compared to the impedance at the fundamental.
accomplished by placing a series tuned circuit in series with the
A LowFER antenna / loading coil combination [will] provide this
but if the transmitter is going to be driving a transmission line,
filter, pi-network, dummy load, or a tap on a grounded loading
series tuned circuit designed for an operating Q of 5 or greater
be connected in series with the load, and tuned for maximum
a low distortion sinewave is desired (e.g., for efficiency
the Q can be selected based on the
above harmonic levels to provide the desired harmonic distortion.
Finally, as Frank's article mentioned, it is important to ensure
the MOSFET gate be driven with a fast risetime squarewave to
operation. Since MOSFETs have considerable input
impedance driving source is needed. I have had excellent
driving MOSFETs directly from HC series CMOS. The HC parts
driving source resistance of 30 ohms or less, as compared to
ohms for the 4000 series. I also recommend AC coupling the
to protect the MOSFET (what if the oscillator stops and the
is stuck high??) (A sample design may be seen in Figure
In summary, I'm just about convinced that the 'simple beacon"
is the Ultimate Design! With the changes outlined here, it
unmatched DC to RF efficiency, as well as being just about the
circuit one could hope for. What could be better?
|Fundamental: 0 dB||6th -30.24 dB|
|2nd -5.54 dB||7th -33.19 dB|
|3rd -16.92 dB||9th -35.20 dB|
|4th -22.66 dB||9th -37.62 dB|
|5th -27.06 dB||10th -38.93 dB|
Additional comments (not necessarily from the
Go to the "CT" MedFer PSK Beacon page
This page converted to HTML by Clint, KA7OEI
Updated on 20110729