Well maybe if the controller lost its voltage feedback or its internal
reference the output of the switching supply could go wild. Most of the
time in a linear supply the output goes wild because the power
transistors weren't proper connected to their heat sink, then they
overheated, and melted into short circuits. In order to try to balance
the currents amongst multiple paralleled pass transistors, they have to
use emitter resistors to drop more than half a volt each. Which puts the
minimum unregulated voltage generally 4 to 6 volts above the output
voltage and that's what shows on the power supply output when the pass
transistors short. Hence the desire for crowbar protection. The typical
crowbar circuit consists of a zener diode, a resistor, and a power SCR.
The resistor is connected from SCR gate to cathode, the zener diode from
SCR gate to output voltage, the SCR cathode to the negative output, the
SCR anode to the positive output, or sometimes the output of the filter
capacitor. Sometimes there's a fuse for the SCR to blow to shut down the
output. Its a juggling act to select the zener voltage. Get it too close
to the output voltage and the crowbar is too touchy. Get it too high and
the radios fry before the crowbar does its thing. Sometimes theres a
capacitor across the gate resistor because one of the failings of the
simple zener and SCR crowbar is that it can be triggered by stray RF.
That makes it unfriendly. There ought to be some resistance in series
with the zener diode so a rapidly rising output doesn't melt the zener
diode or the SCR gate. Often the SCR in the crowbar is presumed to
effectively short and reduce its power dissipation rapidly so that its
not mounted on a heat sink, but it probably needs at least 50 amp rating
for a 25 amp supply because its going to have to discharge the filter
capacitors rapidly and then blow some fuse that only blows fast with
current twice its continuous rating.
There is a pass transistor in the switching supply, its turned on and
off rapidly by the controller. If the pass/switching transistor shorts,
most of the time there's no output at all because DC doesn't pass
through its high frequency transformer. Hence that most common failure
mode of the linear supply is a safe condition in the switching supply.
The failure that can cause radio grief is a loss of voltage feedback.
Voltage feedback typically uses a programmable zener like a TL-431C (it
would be a super trigger zener for the SCR in a crowbar too) and an
optical coupler for isolation. The parts in the voltage feedback circuit
aren't stressed with heat and so their life is long.
I get around the need for a crowbar in my shack supply by making the
unloaded unregulated voltage only 16 volts and use power MOSFETs for my
series regulator which drastically reduces the power dissipation and the
heat stress of those regulator transistors. Runs 24/7 supporting my
packet node, my packet station and the rest of my ham shack. And its
efficiency is very high while its noise is not detectable.
73, Jerry, K0CQ
Entire content copyright Dr. Gerald N. Johnson. Reproduction by