> Steve, what must be gaurded against is overheating of the transistor.
> Overheating is related to load mismatch and thermal time constant of the
> transistor.
Mismatches can also reduce dissipation. If the load impedance
presented to the transistor increases, as a general rule dissipation
decreases. The dissipation curve is almost always very lopsided,
with dissipation reaching a minimum with a load mismatch since
most solid state fixed tuned PA's are set to deliver maximum
energy transfer into a load higher than the recommended load Zo.
This makes the device more linear on peaks, like over-tuning a tube
PA to slightly beyond maximum output at full drive. You open up
the loading control a tad beyond optimum efficiency. Of course if
load impedance increases and the network is fixed, you move the
PA back to optimum point.
As the load impedance is raised, the efficiency generally increases
while dissipation drops.
Dissipation pretty much always tracks PA current when the output
network is fixed. A load change that increases current increases
dissipation.
However, transistors can fail very rapidly from excessive voltage.
You have zero-time to prevent that failure, since it is a punch-
through failure. Temperature really isn't the major concern. Excess
voltage is, and exciters with keying transients (common today) can
really hammer the transistors with collector or drain voltage spikes.
That's another reason why the optimum load impedance is
generally set to be higher than the working impedance.
> a transistor you will see that the die has a typical response of 10ms or
> so. So if you can react quicker than this (say in about 1ms) then you can
> adequately protect the transistor. They do not instantaneously vaporize
> under adverse load conditions.
They most certainly do. Thermal concerns from excessive
dissipation are almost never the failure mode. Voltage breakdown
is immediately catastrophic.
Therefore simultaneous detection of forward
> and reflected power is unnecessary, and a single A/D converter is
> typically used.
You MUST use a directional coupler, and you should always detect
reflected power as well as forward power. Forward power sets the
drive limits, and reflected power sets the SWR limit.
> time of the amplifier is this quick if that's the method used for shutdown
> under high SWR conditions. Severe load mismatches can cause the
> transistors to see higher voltages, but typically not as high as the
> breakdown rating of the device.
Not so. They can easily produce such voltages unless the voltage
breakdown of the device is several times the operating voltage.
While commercial equipment often has enough headroom, amateur
gear doesn't.
One of the cures for the common PA failures in Drake TR5
transceivers was to use a higher voltage output device.
Have a look at Motorola's data, and you will see transistors are
typically recommended to be used with supply voltages just less
than half of the working voltage. In normal operation, that device
would be coming close to peak voltage rating on every cycle at
maximum output. If the collector impedance increases, the voltage
can easily exceed twice the supply rail.
I've measured voltages of 100 volts on the drains of FET's operated
at 12 volts, under some conditions of load mismatch. If the device
have a breakdown voltage of less than that, it will instantly fail. Just
one RF cycle over the peak voltage breakdown will do it.
73, Tom W8JI
w8ji@contesting.com
--
FAQ on WWW: http://www.contesting.com/FAQ/amps
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
|