Protection Devices and Circuitry
While MOSFETs are very rugged devices from a current standpoint, like
all semiconductors, they are subject to immediate damage or destruction if
the voltage ratings of the device are exceeded. The most common failure modes
of MOSFETs in class E transmitters are gate puncture due to voltage spikes at
the gates of the MOSFET, and drain-source (high voltage) breakdown.
Gate Protection
Gate voltage spikes are generally a result of gate overdrive. This is usually
remedied by using a stable driver, and by adding protection devices to the
gate circuit.
You can easily protect the gates of MOSFETs by using TVS (transient
voltage supressor) devices (sometimes called TransZorbs) from the gate bus
to the source bus or ground plane. Try to keep the TVS lead impedance
as low as possible to afford maximum protection to the gates.
TVS devices work well on 75 and 160 meters without any device heating, and
may work on 40 meters, although I personally have not tested this as of yet.
The TVS devices are very effective at protecting the gates from any overvoltages or
spikes.
The 1.5KE18CA TVS works very well in protecting the gates. This is a 1500 watt,
18 volt device. The device starts to clamp at around 19.5 volts, and effectively
keeps the gate voltage at or below that point. The TVS is is used to clamp accidental
high gate voltage. It is not designed to be clipping the gate waveform
under normal, steady-state operation, and excessive TVS device heating and device failure
may result if the gates are overdriven for too long.
Parasitic Oscillations and Drain Protection
The primary cause of drain source failure is a parasitic oscillation in the
class E RF amplifier. Another very common cause is related to poor or failing shunt capacitors (capacitors
usually lose value when heated - some will momentarily fail) allowing large
voltages to occur during MOSFET turn off, which would otherwise not be present if the capacitor
were working properly. Shunt capacitor problems can be hard to find, and are
generally intermittent in nature. Use high quality, high current shunt
capacitors. Using two capacitors in parallel is much better than using a single
capacitor.
Parasitic oscillations come about when the MOSFETs are allowed to
operate in their analog (or linear) region for a long enough time for a
parasitic oscillation to get going. This region exists between the
MOSFET gate threshold voltage and saturation voltage. MOSFETs have extremely
high voltage gain, combined with very high capacitances, which makes the
devices very prone to parasitics unless steps are taken to prevent them.
In
audio circuits, and switching power supplies operating below 1mHz, it is usually
sufficient to insert a resistor between the gate of each MOSFET and any other
components, to stablize the MOSFET against parasitics. In class E amplifiers
operating above 2 mHz, this is not always practical. Even a small resistance in series
with the gate will significantly reduce the gate drive, and will slow the
rise and fall times of the gate waveform. Putting a 10 ohm NON-INDUCTIVE resistor
from gate the source can often help.
The best way to prevent parasitics is to use a gate driver circuit
that guarantees the MOSFET gates are actively held in either the ON or OFF
state at all times. The gates must never be allowed to "float" or be subject
to a high impedance driving source. If the gate drive drops under modulation,
you may also be in danger of allowing the MOSFETs to operate in their linear
range, giving rise to possible parasitics. The most fool proof driver for a
MOSFET is a driver IC connected directly to the gate, with short leads and good
grounding and bypassing.
You can protect the drains from the possibility of overvoltage
caused by parasitics and other anomalies by using a TVS across
the drain bus. A couple of 1.5KE540A TVS devices, connected from the drain bus to the
source bus has proven effective in helping to reduce or eliminate voltage spikes
that would otherwise cause device failure.
The modulated DC input to the RF amplifier should also be protected with
130V (1.5KE130A) TVS devices. This will prevent high voltage spikes caused by
inductance in the modulator circuitry from reaching the RF amplifier. These
devices should be located right at the RF amplifier, one for each module.
Overcurrent Protection
By using a high speed current sensor such as a Hall Effect device, connected
between the output of the modulator and the RF amplifier, combined
with circuitry to quickly remove power from the RF amplifier in the event
of an overload, it is possible
to provide very effective protection against even short duration current overloads.
Most of the power supplies shown here incorporate some type of overcurrent
protection. These are relatively
slow shut-down circuits, and will generally protect the power supplies
and modulators from damage which would otherwise result from severe
overload.
Do not leave these important circuits out of your implementations.
Cooling
Although very efficient, class E RF amplifiers, and their modulators
can generate a considerable amount of heat, if operated at more than a
hundred watts or so. An RF amplifier operating at 350 watts will
produce at least 75 watts of heat, when you factor in gate drive power
(converted to heat). If you have included a driver on the same heat
sink, you must conclude that all of the power used in the driver will
be converted to heat, either in the gates of the RF MOSFETs or the
drain of the driver MOSFET. A class H modulator, supplying modulated DC
to a 350 watt transmitter will produce approximately 130 watts of heat
at carrier, and somewhat more than this at full modulation.
Sufficient heat sink area is essential for proper cooling of the
modulator and RF amplifier. A 8 inch by 12 inch heat sink with 20 3/4
inch fins (or equivalent) equipped with a small fan should be more than
sufficient for the RF amplifier. For the modulator, a similar or
somewhat larger heat sink, equipped with a good fan, should work.
