■ PHOTO 2. The output
directly from the voltage
doubler (without filtering)
shows a DC signal with
switching glitches. The
simple filter shown in
Figure 3 reduces these
glitches to about 200 mV.
current of one amp to the
gate of the MOSFET and
typical switching times of
15 ns into 1,000 pF. They
can support MOSFETs
operating at up to 90 volts
and can supply a bootstrap
voltage of up to 108 volts. The Schmitt trigger inputs
accept TTL and CMOS level signals.
Unfortunately, these drivers do not come in standard
DIP packaging. I had to get SMT parts and fabricate an
adapter as shown in Photo 1. It requires a steady hand
with the soldering iron but is not too difficult. The trick is to
put the component carrier in a protoboard (as shown) to
support it and act as a heatsink for the pins. Put a blob of
solder on a carrier pin and use #30 solid wire. Solder the
carrier first because it acts as a heatsink. If you solder a wire
to the SMT driver pin first, it may become unsoldered when
you solder the wire to the carrier pin. Be careful with the
heat because the carrier is made of thermoplastic and melts.
Last part of the circuit is the power MOSFETs
themselves. I chose IRF540 devices. They cost about
$0.75 each, operate up to 100 volts, and can handle 28
amps continuously with an on resistance of 0.77 ohms.
In theory, this design can handle 2,500 watts of power.
We won't be anywhere near that value.
The input voltage is limited to 30 volts because the
voltage regulator that supplies power to the timer and
other circuits is only rated at 35 volts. If you use a
separate low-voltage power supply, the input can be
theoretically increased to 90 volts where the driver chips
fail. More realistically, the maximum safe input to the
H-bridge is about 75 volts as shown.
The maximum continuous DC current — derated for
heating — is about 14 amps for the MOSFETs. Since each
MOSFET is on only 50% of the time and the switching
speed is high, a full 28 amps through the bridge is a
reasonable figure. Note that each MOSFET will have
to dissipate 50 to 60 watts in this scenario, so good
heatsinking will be a requirement.
■ PHOTO 3. The switching glitch is about
165 ns. This can be reduced or eliminated with
proper fine tuning of the switching signals.
simple filter as shown. The capacitor ratings in this voltage
doubling circuit are critical. The use of improper
capacitors can cause them to explode. (See the sidebar
on Switching Capacitors for details.) Because of this failure
possibility, it is strongly recommended that the project be
housed in a sturdy box. Since the box only has input and
output jacks, there seems little reason to show it.
If you want to do something other than voltage
doubling, you can use Figure 4's circuit. It's more
conventional and easier to understand. Just remember
that the switching frequency is 32 kHz, so ordinary 60 Hz
transformers are marginal performers. (Of course, you
can always change the circuit's frequency.) A toroidal
transformer is usually incorporated here. Generally, they
provide better efficiency and high frequency performance.
I tried an ordinary 60 Hz power transformer and got
Output Circuits and Variations
We will be using a simple capacitor voltage doubler,
shown in Figure 3. It connects directly to the output of
Figure 2. (In addition to the H-bridge outputs, connections
to H-bridge power and ground are required.) Note that the
doubled output voltage is theoretically a perfect DC signal
right out of the bridge rectifier. In practice, switching
glitches are present and can usually be removed with a
Bridge Power Ratings
In theory, a bridge circuit can provide up to four times
more power to the load than a standard power control
circuit. This seemingly impossible task is accomplished
fairly directly and logically. However, it is a bit subtle.
Suppose you have an ordinary audio amplifier running at
eight volts and driving an eight ohm speaker. The voltage
swing at the speaker will be a maximum of ± 4 volts. Four
volts into eight ohms is 0.5 amps. The peak output power
is then two watts (Power = Current x Resistance).
However, a bridge circuit changes the polarity to the
load. This results in eight volts going in one direction and
8 volts going in the other direction for a ± 8 volt swing.
Doubling the voltage causes twice the current to flow,
or one amp. Doubling the current results in four times as
much power (P=I x R). So, a bridge amplifier can provide
up to four times as much power (eight watts) to the load
as a non-bridge type.
The next question is why bridged power amplifiers
are only rated at twice the non-bridged power, or less.
This is because of the power handling of the amplifier
rather than the theoretical power delivered to the load.
If one amplifier can only dissipate enough heat to support
X watts of power output, then two amplifiers can still only
dissipate 2X watts, regardless of the voltage swing.
February 2009 55