This may not be germane to the building of the amplifier and it is by
necessity an over simplification, but it is the temperature problem
explanation for those not familiar to transistor failure and I hope it
is of interest to some.
Transistors do age with use. Typically, unlike tubes the shelf life of
modern transistor is almost indefinite except for contaminates introduce
during manufacture which is rare now days.
However with use, the dopant (intentional impurities induced during
manufacture to create N and P type material.) migrates with voltage and
current flow, but under normal conditions this migration is miniscule,
but complicated. For simplicities sake we just say the primary carriers
(and minority) carriers move toward junctions in bipolar transistors.
A similar effect takes place near the gate in FETs.
Of course if the temperature gets high enough we end up with a
catastrophic failure, rather than aging.
It might help to explain that during manufacture the device is placed in
a furnace, heated and exposed to a gas to produce a layer of oxide.
Later the structure the device is etched in the oxide to expose the
Silicon. The device goes back in the furnace and exposed to a gas
containing the dopant that creates the N or P type material at an
elevated temperature by diffusing into the Silicon. This operation is
usually repeated many times to make a device.
So when a transistor is exposed to elevated temperatures it is pretty
much in the same conditions used to manufacture it. Normally we are no
where near this temperature, but the warmer the device the faster the
dopant migrates.
The modern materials are extremely pure. Far more so than the industry
was capable of 30 or 40 years ago, let alone 50 years ago when the
industry was in its infancy, BUT all material does contain some
contaminants. Typically for N type material, it is a P type and for P
type it is N type and these are called minority carriers.
With age and heat the migration "slowly" creates more minority carriers
which I guess a very loose analogy would be a tube becoming gassy or
contaminated. I say very loose because the entire mechanism is
different. Only the results are similar in causing a degradation of the
device. I did say at the beginning this had to be an over simplification.
Referring to the purity level I believe we are now in the parts per
trillion whereas we were in parts per billion just a decade or two back.
I was given the analogy for parts per billion, think of as cube of
white bricks one city block on a side. some where in that cube is "one
black brick". We are now talking about purities on the order of a
thousand times greater. Poly crystal Silicon coming out of the reactors
is now more pure than we could get from float zone refining in the early
days and that stuff was expensive! (As much as $165 USD a gram).
As a side note, in the early days reactors produced poly rods of 3/4
to 1 inch in diameter and 16 inches long and operated at around 10" of
water pressure. Today's rod size and the operating pressures are
proprietary, but I can tell you that it takes a strong, chain hoist to
remove the rods from the reactors and pressures are in PSI rather than
inches. Growth rates are also much faster. But the industry has always
been "Feast or Famine". I believe poly has sold for as little as $6 USD
a kilo (long time ago) and as I believe as high as over $400 USD per
kilo. As a SWAG I'd guess it's presently around $60 a Kilo. None of the
manufacturers want their competition to know what they are getting but
this price has a direct impact on device development, availability, and
price. IIRC there are 3 main manufacturers that are capable of
weathering the Famine and right now it is a major famine with I believe
around 30 some manufacturers.
One of the main producers just finished an expansion project of over a
Billion dollars as well as developing a new facility. They were in
another Billion dollar expansion when the recession set in.
Back to devices:
Keeping the temperature of the device low is a science by itself. The
first and main problem is getting the heat from the Silicon die (the
device itself) to the case as this device is surprisingly small. The
temperature difference between the device and case ( called delta T) is
the single most important limitation of the device. Far more so than
cooling the case and its fixed within the design so we can do nothing
about it other than to cool the case as much as possible and this is a
real challenge. This is also why we find high power devices listed in
pulsed service.
Without going into a long explanation of the cooling I'll just say it is
essential to keep the device as cool as is practical as there is little
difference between a temperature that will allow normal operation, one
that will produce accelerated aging and one that will produce
catastrophic failure. It's much easier to cool multiple devices with the
power divided between them. That's not to say cooling at solid state
QRO is simple as it's a long ways from that with exotic heat sink compounds.
73
Roger (K8RI)
.
Rather than a single big device, I like and prefer the idea of multiple
parallel/push-pull RF power devices to share the load and more evenly
distributed the massive heat into the copper spreader and heatsink system.
Devices tend to prematurely fail when their thermal aspects are not properly
attended to, rather than a failure attributed to load / VSWR mismatch.
Leigh
VK5KLT
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|