> >Modern RF Fets are pretty robust devices and have an MTBF a couple
> orders of
>
> >magnitude better than thermionic devices, provided the junction/channel
> >temperature is kept to a sensible level, and the breakdown voltage and
> >dissipation limits of the device are observed. Monitoring the drain
> >currents, heatsink temperature and load VSWR is really all that is
required.
> >A $2 microcontroller could be used to automate the monitoring/trip
> >functions......
> >
> >
> I am troubled by this opinion. When I study the hardware to the R&S box
> I see evidence of great attention to details that I sense are associated
> with longevity of the devices.
Can you be specific?
>I am also told that many high power RF
> solid state devices have a definite life cycle.
They do. As I mentioned above, a solid state device has about two orders of
magnitude greater MTBF than a tube
>The issues are as I see
> it survival in the keying ON/OFF mode, some issues of linearity and
> instability below, at, and above the operating frequency. My sense is
> that we need some serious development of protection for open or short
> antenna feedlines or relays, even to the extent of a momentary test
> every time one goes into transmit mode. I suspect that we need some
> rather serous protection efforts as well as a way to understand what the
> protection system did when it protects the devices.
This how folklore develops. There is no more need to attempt to monitor the
performance of a solid-state PA device on a cycle-by-cycle basis than there
is with a tube. The reasons for a solid state device failing are voltage
breakdown, overcurrent, or overdissipation. There are no 'magic' effects.
Much of the reputation within the amateur community of the supposed
unreliability of solid state devices stems from bad design. As I've said
previously, solid state devices, operated conservatively will have an MTBF a
couple of orders of magnitude better than tubes. No electron device has an
infinite life....FWIW it's easy to show that adding 'protection' circuitry
will have an adverse effect on the MTBF of an amplifier!
Using devices conservatively is the first step. In the same way as you
wouldn't run, say, a 4CX250 with 3kV on the anode, and expect it to survive
a sudden severe mismatch, it's not a good idea to run transistors at too
high a voltage. Running '28V' devices at 24V (or choosing devices with a
larger breakdown voltage than the 70V usually suggested for 28V amplifiers)
will result in considerably increased MTBF, and an amplifier which should
laugh in the face of any passive mismatch. When I design amplifiers, even if
the spec. only requires that the amplifier will survive into a 10:1
mismatch, I test for all phases of the worst vswr I can generate (>50:1) and
expect the amplifier to survive undamaged.
A power supply with effective current limiting is a good idea.
Choosing the right device technology is also important. At HF and VHF, DMOS
devices, such as those manufactured by Polyfet and Semelab, have better
breakdown voltage performance than some of the more recent processes, such
as LDMOS.
The super twitchy can always fit an isolator after the PA, and sleep well
knowing that the amplifier will always see a good load, even if the antennas
are completely blown away!
A major aspect is thermal design. Expecting SS devices to work for any
length of time when the silicon is approching liquidus is clearly rather
optimistic. Properly designed heatsinking is essential. Spending time
ensuring that the flange, or better still, the die temperature of the device
is well within the manufacturers limits is an important part of SSPA design.
Too many amateur (and not a few professional) SS amplifier designs pay scant
attention to this. A channel/junction temperature of 120C is often specified
for professional designs. That's a good figure for the amateur designer to
aim at. Although SS amplifiers will continue to work (for a while!) with
channel temperatures of 200C, as a rule of thumb, the MTBF reduces by around
a factor of two for every 10C rise.
With a big amplifier, don't expect to be able to get away without some form
of fan in the cooling system. Relying on simple radiation to cool a heatsink
is relatively inefficient: a small fan can improve efficiency by a factor of
two or more.
> If protection and management was simple we would already have it in
> place. Since we have no available systems I must assume that the task
> is far beyond trivial.
With competent design of the basic hardware, fancy protection and management
systems simply aren't necessary. That's why there is little evidence of
them. In practice, they would reduce the reliability of the complete system,
and to do the job properly they'd require sensors integrated into the active
device. Although I saw this years ago in LF power transistors, to the best
of my knowledge no RF power transistor manufacturer has been down this
route. Hi-rel transmitters for avionics and professional comms for which
I've designed PAs have all used very simple PA monitoring circuitry - or
none at all.
A simple system to monitor load vswr, heatsink temperature, fan activity and
Id of the active devices could be a very nice addition to a solid-state
amplifier, but providing the amplifier itself was well designed, it would be
a luxury.
Vy 73
Chris
GW4DGU
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|