Roger, and all,
> IF and I have to emphasize the IF these transistors were rugged enough
> for prime time and I'm quite willing to take your word on the power you
> are getting out, the manufacturers would be jumping on them like flys
> on...er... honey because they could run them PP/parallel for the legal
> limit out plus comfortable orverhead and couldn't build them fast enough
> to meet the demand. There has to be a reason they are not doing so.
In fact I have been wondering about this question. The reasons I can see
for commercial manufacturers not doing it this way might be both
technical and commercial:
- Hams are obviously willing to pay 5000 bucks for an amp, so why
destroy the market by making amps that sell for 700 bucks? The
intermediate option, making amps that are cheap to make, and then sell
them at 4000 bucks, probably won't live long with hams. They want to see
where the cost is.
- There are several 30 to 40 year old designs with 50V FETs that
actually work, sort of. So let's just use these, instead of investing
money in developing a new circuit around new FETs. There you have all
those MRF150 amps. And the old designs are copied COMPLETE, even with
the obvious bugs many of them have!
- New designs = new problems, so... STAY AWAY! The present amps are
still selling well enough, so don't touch them!
A real technical difficulty with the type of amplifier I'm working on is
that it really only gets good with water cooling. Trying to use air
cooling, more individual devices are required, and the additional
capacitances quickly stop the show, specially when combined with the
additional wiring.
And another technical problem is that my amplifier probably won't end up
particularly rugged. If you set your mind on it, you can probably blow
it up just by slowly tuning an antenna tuner through its whole range,
until you find a setting with sky-high SWR that just happens to create
high enough drain voltage peaks to fry the FETs. Protection circuits can
keep FETs alive, as long as the high SWR comes up slowly (milliseconds),
but if you have a loose connection in the antenna system, that provides
1:1 SWR most of the time, but suddenly goes up to a condition that
causes those high voltage peaks, this amplifier can be blown up more
easily than one that uses just a few, rugged FETs. The reason is that
I'm using many small FETs, directly in parallel. That's simple to do,
and inexpensive, but any overvoltage will make a single FET enter
breakdown, and that single FET cannot survive the big discharge. When
using just a few large FETs, which is much more expensive, the big FET
that breaks down has a much larger chance to survive.
For this reason, an amp like I'm developing is more suitable for the
homebrewer than for the appliance operator. FETs will blow just under
totally abnormal conditions, such as arcing over a loose antenna
connection. With some care, such a situation can be largely avoided. And
IF it happens, the FETs are inexpensive, and easy to replace. But an
appliance operator is more likely to have such poor connections in his
antenna system, and if he blows a FET, he has to send back the amp to
the manufacturer, have a professional spend time on it, and this sort of
masks the cost difference between a 1.5 dollar FET and a 150 dollar FET.
So far I have killed no FET at all during all of my experimention with
this amp project! But I'm far from done. I probably will kill some,
while testing to the limit at high SWR. Anyway, they seem incredibly
forgiving of mistakes. When semiconductors started, over half a century
ago, oldtimers successfully made many people believe that those
newfangled tiny things were terribly fragile, and this reputation has
stayed. In truth, they are far more robust than most people think. Most
FETs can take a high voltage spike that makes them break down (note to
newcomers: "breaking down" in this context means "abnormally and
unintentionally entering conduction", and not "failing"), and suffer no
damage at all, as long as a certain energy per event isn't exceeded, and
of course as long as the device isn't overheated through repeated
breakdown or other causes. And regarding static charges killing MOSFETs,
in all my 30 years working in electronics I have killed only one single
component that way! And that was a "joint venture" with a colleague: I
had just successfully repaired an ultra sensitive front end card that we
urgently needed for the place's scientific work, and I triumphantly
waved it at my workmates. One of them came over, walking across the
carpet, without any antistatic gear (we didn't use any, back then), and
took the card from me. ZAP! We both saw the spark from his hand to the
card's edge, and we both felt the jolt. And then I had to repair the
card again.
But that event was highly exceptional. Most semiconductor parts,
including all power FETs I have ever used, have built-in protection in
the form of zener diodes, that keep the devices alive despite the usual
small sparks that happen in daily life. It takes an exceptionally large
discharge to kill a part. Some specific parts, such as certain ultra low
noise microwave FETs, and many laser diodes, are unprotected against
static. They need to be handled carefully, but it's nothing out of this
world.
On the other hand, I remember a terrible event in my other hobby, that
of restoring antique radios. I needed a good-enough 2A7 tube for the
front end of a beautiful 1933 cathedral radio. So I took my big box of
old American tubes out of the cabinet, and went through it looking for a
2A7. I found two. Then I turned to my tube tester to see which was
better, and something, somehow, I don't know how, got tangled, and the
whole box of tubes fell to the floor! More than half of the tubes were
broken, around 30. Talk about fragility! Semiconductors at least don't
have this particular sort of fragility.
> As to home brew, there are very few of us who actually do that be it SS
> or tubes.
I can't help it, I worry that this is spelling the end of ham radio. To
me, ham radio is about being interested in radio technology. I can
perfectly well accept the fact that not all hams are professionals in
electronics, but I would expect that all hams, without exception, should
at least be interested in electronics, and try to learn, at their own
rate. I just can't understand those people who say they love ham radio,
but whose love for radio ends precisely at the front panels of their
rigs! To me, they DON'T really love radio. Instead, they are after the
utilitarian side of ham radio, such as keeping in touch with friends and
family, or having emergency communications. In my surroundings, at least
90% of hams, and probably much more than that, belong to this group that
doesn't have any interest whatsoever in electronics. Hear them talk on
the bands: The weather, the brand and model of their rig and antenna,
and that's about it. I call them "empty QSOs".
> I'd be interested in the device, and circuit design.
What I'm doing is loosely based on DL9AH's designs. Google for his call,
and you will find lots of information, most of it in German, but at
least you should understand the schematics and photos... But I'm using
adequate cooling, which he didn't, and also I'm using a much more
decent, even if still very simple and inexpensive power supply. The
intention is to package all that with a set of auto-bandswitched low
pass filters, and sufficient protection to keep the FETs alive in all
normal situations. I have tried several newer FETs than the ones Arno
used, obtaining better performance, but I don't want to give away all
details of what I'm doing, until I'm through with it and get it published!
Anyway, if these messages spike some of you to start working in the same
line, I would consider that to be very good, and would love to exchange
experiences.
Take DL9AH's published design, put in a larger number of newer and
better FETs, add water cooling, replace his half-wave power supply by a
switching supply with improved power factor, replace his somewhat
problematic RF transformers by ones that work well, add autoswitched low
pass filters, and there you have what I'm doing!
I'm not nearly ready, in any case. I'm rather in the early stages!
> At any rate all commercial amps and projects I've seen had very
> elaborate protective circuitry which I said before was pretty simple
> when looking at each section, but kinda intimidating when looked at
> overall. Course most people make the mistake of looking at the whole
> thing instead of one step at a time.
People who have never built any equipment of significant complexity can
of course get scared by such a circuit, that might have 70 parts or so.
But it falls into proportion when you realize that these 70 parts cost 2
20 dollars total, and fit in three square inches of printed circuit space.
Tell these people to look into their HF transceiver, and count the
parts, and try to understand it all. That's quite a bit more complex.
But even the most advanced HF transceiver consists of nothing but very
plain, simple individual circuits.
Of course, anything like that is too complex for the "typical ham", the
one at whom the ARRL is addressing QST. But even the simplest tube type
amp is neither for the typical ham, because that typical ham would never
dare to build anything having 3000 volt in it. Or rather, he would never
attempt to build anything at all, period!
> I still maintain SS is not ready for prime time, but for those with the
> knowledge and desire to experiment it's a great field.
Well, I think differently: I think solid state is the way to go,
definitely, at the power levels hams are allowed to use. Only at much
higher power levels do tubes make real sense. Simply because tubes are
available for several tens of kilowatts in a single device, while
transistors are not, and would need to be stacked up in too large
quantities. But at 1.5kW, and even 10kW and some more, solid state is
more practical than dealing with tubes and their inevitable
narrow-banded tuned circuits, their lifetime of a few thousand hours,
and sockets costing a thousand dollars!
One could argue that a lifespan of a few thousand hours for a tube will
last a ham's lifetime of normal operation. But practical experience
shows that tubes are less reliable: Among the local hams I know and who
own power amps, the spread is about 2/3 tube amps, and 1/3 solid state,
changing fast to increase the proportion of SS. Most of these hams turn
to me when something fails. Over the last several years I had to replace
several 3-500Z, and a few other tubes, but not a single high power
transistor. The only time I got a solid state amp for repair, a Quadra
that had seen extensive contest and DXpedition use, the problem was a
broken solder joint that took five minutes to find, and one second to
repair.
In several tube amps I had to replace electrolytic caps, and I had to
rewind two power transformers. In SS amps, so far I have seen no power
supply trouble. Of course, the number of amps that could possibly fall
in my hands when broken is small, but I believe that it's enough to do
at least a broad statistic about their reliability.
Years ago I looked into the possibility of making broadband tube
amplifiers, because I just hate having to retune an amp after changing
bands or even changing frequency a lot within a band. But I quickly
found that the tubes' combination of high internal capacitance and
required high operating voltage makes this totally impossible. Tubes
force the use of tuned circuits, while transistors give the designer the
choice between tuned and broadband impedance matching.
> It's also, often at pulse with enough off time to allow the heat to
> migrate to at least the spreader
Indeed the datasheets of many RF transistors rate the RF output power
under pulsed condition, and I have seen hopeful hams who thought they
could run that power continually...
In pulsed service it's quite complicate to correctly calculate the
thermal limits, because the distribution of thermal mass and thermal
resistance within the device has to be considered in a detailed way. So
it's good that manufacturers specify the ratings for pulsed use. But it
would be really nice, and in the best interest of honesty and
transparency, if they would also rate the actual power output under
continuous two-tone conditions in class AB, while using real-world heat
sinks! But then, many of these transistors are simply not intended for
us hams, nor for any other communications use. They are intended for
nuclear magnetic resonance imagers and other such big money devices,
which indeed work in pulsed mode.
The power dissipation ratings in the data sheets are instead in steady
state conditions. At least I haven't seen anything other. And this is
what brings about data sheets of transistors that tell that you can get
1.5 kilowatt RF output, while the device has an absolute maximum
dissipation rating of 600 watt. The output rating is in pulsed service,
while the 600 watt dissipation is continuous, but assuming a case
temperature of 25 degrees Celsius, which can hardly ever be achieved in
practice (a beacon transmitter at the south pole might be the
exception). To a ham, these ratings mean that you can run perhaps
300-400 watt output in SSB, class AB, or about 600 watt output in CW or
FM, running class C. But certainly not 1.5kW, unless you go to class E
or class F - which is actually a very good way to make good use of these
transistors!
Tube dissipation ratings instead are true, real, directly usable, as
long as you assure enough airflow. This difference comes from the simple
fact that tubes usually have their heatsinks built-in, while transistors
do not, so transistor ratings have to be given in a way that allows
equipment designers to calculate the true allowable dissipation, after
fitting a certain heatsink.
Note that some solid state devices, such as many diodes and also some
switching MOSFETs, nowadays are rated for dissipation and/or current at
a more real-world temperature, such as 75 or even 100 degrees Celsius.
> Depending on frequency we set the internal limit at 100C and max
> internal operating temp at somewhere between 70 and 90C.
> dopant migration was starting to become noticeable at 100C so when I was
> in the industry.
That's a bit lower than I had known. Anyway one has to keep in mind that
these phenomena are time-related. A ham amplifier can probably get away
with pushing the silicon to 150 degrees when delivering full output,
because this will only be short-term. If the transistors live for 10,000
hours at that temperature, that should be enough. A broadcaster instead,
running 24/7 at full power, will want to get much more than 10,000 hours
of life from his transistors, and so they need to run cooler.
That said, and having done calculations on the thermal aspects of
widespread ham equipment, I can't avoid the suspicion that many ham
transceivers run their finals well in excess of 150 degrees internally!
The typical max ratings are often 175 degrees or even 200 degrees for
transistors that are NOT encased in plastic. Particularly brave
manufacturers even specify 225 degrees. I don't know how much of that is
due to processing techniques that reduce dopant migration. The rest is
probably based on decisions to accept shorter lifespans as "typical". In
any case, silicon by itself survives much higher temperatures. It's the
migration problem that sets a limit, and the attachment of the chip to
the case (usually by some solder) sets another limit.
> You can also use *cold* or chilled water. Just draw the water out of an
> open PVC tank and return to the same tank. 20 gallons will run a KW amp
> for quite a while. Throw in a few chunks of ice and have at it. We use
> the water in plastic bags for our picnic coolers , so you just grab a
> couple and drop in the tank
But one needs to be always careful about avoiding condensation inside
the amp. At my former job, we had tons of actively cooled electronic
equipment. There were big coolers to chill the water/glycol mixture just
enough to keep the outside of the electronic racks precisely at ambient
temperature, to prevent thermal turbulence. The controllers monitored
ambient moisture, to keep the water always above the dew point.
Sometimes something went wrong, and then we got dripping wet
electronics, with all the obvious consequences.
The coolers were mostly of the plain common compressor heat pump type,
but in some vibration-sensitive locations we decided to try Peltier
coolers. They turned out to be a disaster. The place was at 2600 meter
altitude, the Peltiers overheated due to the low air density, and died
faster than we could throw them in the trash.
> You can take it down as far as you like as long as you don't get
> condensation on the circuit board,
Exactly. And with a typical comfortable ambient humidity of 60 to 70%,
condensation happens at just about 6 degrees below room temperature!
It's not worth doing, for such little temperature difference.
> or ice up the pump.
At any normal room temperature and humidity, condensation will occur far
sooner than any ice can form. The dewpoint at comfortable room
conditions is always above 10 degrees Celsius. You would need an
extremely dry or very cold environment, to get the dewpoint below freezing!
> BTW there is a fungicide used in swimming pools that will do a good job
> of keeping *stuff* from growing in your cooling system. I'm not
> referring to Chlorine.
I will have a look in the swimming pool section of the home improvement
store! Thanks for the hint!
Manfred
========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|