[Amps] Ampleon BLF189XR 1900W LDMOS transistor

Manfred Mornhinweg manfred at ludens.cl
Thu Oct 20 10:55:03 EDT 2016


Kevin, Jim, and all,

> W6PQL tested the NXP BLF188XR, the device he uses in his 1KW amps,
> and came up with 1100W before hitting 1dB compression and it's a
> 1400W "peak" device.

Is there a web page or something where we can see what exact circuit he
used? I ask this because the 1dB compression point isn't a fixed
characteristic of a given transistor, but rather one involving the
transistor, the rest of the circuit, the frequency, and the supply
voltage. It makes no sense at all to say "the BLF188XR has 1dB
compression at 1100W". Instead it makes a lot of sense to say "this
particular amplifier circuit, using a BLF188XR, and a 50V supply, has
1dB compression at 1100W and 30MHz", or something like that.

To answer the question of whether to use one or two devices for a given
set of conditions, it goes roughly like this: First you find out what
supply voltage, load impedance, and amount of feedback you need to
achieve the desired power at 1dB compression, at the highest frequency
of interest, using a single device. Then you calculate how much power it
would dissipate, under those conditions. Then you see whether you can
reasonably build a heatsinking system that would make the device run at
a safe junction temperature (providing the desired minimum life span).
If you can, you may go with a single device. If you cannot, you use two
of them, which makes the thermal handling VERY MUCH easier.

Even if in some situation you might just be able to use a single device,
it can be better economy to use two of them. That happens when two
devices on a simple, smallish heatsink come cheaper than a single device
on a super duper peak performance watercooled copper block, radiator,
pump, etc.

> If I was building one with the BLF189XR I'd still use two devices.

Me too, probably, for thermal reasons, but first I need to see the
datasheet. And if using two, I would use each of them in a complete
amplifier block, and then combine the outputs. The more obvious solution
of making a single push-pull amplifier with one complete device per side
is very hard to implement properly, because of the extremely low
impedance. Also making two amplifier blocks allows physically separating
them by a significant distance, allowing much easier cooling. For
example, you can use one of those heatsinks that form an air duct,
having the fins inside, with a single fan blowing through them, and
mount one amplifier block on each side of that heatsink. That's much
better than mounting them side-by-side on a flat heatsink.

> The big question is the low pass filter. The single most expensive
> part of an SSPA.

Yes.

> ### This is my problem with these SS amps, the LP filter.  And you
> require a myriad of them, not just one of em.   On a NINE band ham
> amp, you would require a bare minimum of SEVEN 1.5 kw CCS rated LP
> filters... AND you have to be able to switch them in / out.

The coils are easy enough. Switching can be handled by inexpensive
relays. The problem are the capacitors. Factory-made capacitors with
adequate ratings are rare and expensive. Like a 180pF capacitor rated
for 15A of RF current, along with being small enough to have an
equivalent series inductance of only a few nanohenries. Very few
capacitors meet such specs, and they tend to be expensive.

I have been considering making a whole legal limit low pass filter board
  based on homebrew capacitors. Metal-clad multilayer capacitors are very
easy to make from copper foil and some dielectric. The dielectric
can be either mica sheet or teflon film. Each has its own plusses and
minuses. Teflon is soft, so one has to be careful with burrs in the
metal punching through it. Mica comes in uneven thickness, so that it
takes some cut-and-try to achieve the correct capacitances. Mica has a
higher dielectric constant, so the capacitors end up being smaller. But
even with the larger teflon caps, the series inductance is very low,
because of the wide and short conductors that result from the 
construction of metal-clad capacitors.

If anybody has a good idea what other dielectric material could be used, 
that's easy to get, please let us know! It needs to be heat-resistant 
and have a very low dielectric loss, of course. A somewhat highish 
dielectric constant would be a plus.

> You dont  require ANY LP filters on any tube amp.
> 
> ##  You also forgot the SS amps also require the fully automatic,
> 1.5 kw CCS rated antenna tuner.  Again, a tube amp doesnt require a
> tuner.

Well, I would like to remind everybody that there is no law requiring
that solid state amps must be broadband or autotuned! In the ham world
it's just a custom that most tube amps are manually tuned, with a few
being autotuned, while solid state amps are broadband, with some having
an add-on or built-in autotuner. But there is no technical reason
whatsover preventing us from building a solid state amp that has a
high-Q tank like a tube amplifier! That tank would replace both the
switched low pass filters and the autotuner. And the operating
inconvenience would be exactly the same as that of a conventional tube 
amp: You need to tune the amp when changing bands, or when making a 
large frequency change.

In practice you would build the typical broadband push pull amplifier
block, or two of them plus a combiner, and pass its output through a
tank circuit looking much like a tube amplifier's one, with a big,
simple bandswitch, a large coil, and all. Only that instead of a low 
capacitance, high voltage tuning capacitor, both variable capacitors 
would be of the same kind, having the same ratings as the loading 
capacitor of a tube amp.

It would be a good idea to analyze the various possible tank circuit
configurations, as maybe the Pi circuit isn't the best for this
application. In any case it would boil down to two 
front-panel-adjustable elements, plus a single band-switched element, 
and would look much like a tube amp's output circuit.

> A  SS amp will blow its brains out with swr.

Not if a correctly designed protection circuit is used in combination
with these "extra rugged" LDMOSFETs. Their ruggedness allows them to
survive the few milliseconds the protection circuit takes to kick in,
even at infinite SWR.

> None of em will operate full power into a 2:1  swr.

Not without tuning. But a tube wouldn't either. It's just that
essentially all tube amps have tunable matching circuits included. As
soon as you include them in a SS amp, it will handle SWR as well as a
tube amp, and even better: Because SS amps always have protection
circuitry, and tube amps often do not! A serious operator mistake can
blow out a tube (and many of them cost a lot of money!), while a
properly protected SS amp should be impossible to blow up.

Then again, of course it's possible to include total SWR protection 
circuitry in a tube amp, but most tube amps don't have it.

> ##  Your typ  SS amp operates at 50% eff at best.

It should be possible to get to 60% or so in linear class AB, with a
well-done design. It's just that most SS amp designs are poor! And these
UHF-capable LDMOSFETs open the doors to high-efficiency linear
amplifiers that simply can't be built with tubes. Such amps have
efficiencies ranging from 85 to 95%. This side of things NEEDS to be
explored by hams, NOW! Almost all of us (myself certainly included) seem 
to be too old and too lazy to get serious, sit down and do it!

> So for a 1.5 kw CCS  rtty / FM signal, the heat sink would have to be
> able to dissipate 1500 watts  CCS.

Anybody serious about RTTY and FM operation should either build an amp
dedicated to these modes, or one that can be switched into these modes.
Because RTTY and FM don't require a _linear_ amplifier. They can
perfectly well be done using these LDMOSFETs in saturated class AB.
That's the mode most commonly used in the "typical application" circuits
in the datasheets, and an efficiency above 80% can be obtained without
any strange tricks or complex circuitry.

A practical legal limit ham amp might simply switch the supply voltage
according to mode. With the higher supply voltage (50V) it would run
1500W PEP with 1dB compression, at about 55% efficiency, for SSB, and
with a lower voltage, roughly 40V, it would run the same 1500W but with
  6dB compression or such, requiring higher drive power (but still far 
less than 100W), and achieving over 80% efficiency, for RTTY and FM. No 
other change than the supply voltage switching is needed, and that's 
trivially easy to do when using a switchmode power supply.

The low pass filters, or tank circuit, need to be able to provide enough
attenuation of harmonics even when the amp operates in the nonlinear 
mode. That's a point to watch.

An amplifier operating in this way would dissipate around 800W average
during 1.5kW PEP SSB transmission in linear class AB, and only around
400W during 1500W RTTY or FM transmission in the saturated class AB 
mode. It might be possible to handle this with a single device, the 
critical condition being in the SSB mode, not in continuous-duty RTTY!

> ## Will  2 x BLF189XRs even do 1.5 kw pep with real low IMD ??

The real question is: Can the BLF189XR, mounted on the best heatsink 
system you can make, handle the heat resulting from such operation? To 
calculate this, I first need the data sheet of the BLF189XR.

Third IMD levels of -40dB are pretty easy to achieve with LDMOSFETs, as
long as you use a good amount of negative feedback, and stay out of
saturation. Many ham designs use too little negative feedback, because
they copy datasheet circuits intended for nonlinear, saturated class AB.
And of course, many hams try to milk the amps for the last watt, which
means entering severe saturation and creating horrible levels of IMD.

If you want third IMD levels much lower than -40dB, combined with good
efficiency, techniques like predistortion or t least active feedback 
must be used. At that point things get more difficult. But it's 
pointless to go below -40dB in a ham amp, since the radios used for 
driving them won't be as good. And no one on the air will notice, due to 
QRM and QRN being much stronger than the residual IMD. With an SDR 
transceiver using predistortion to drive a matched LDMOSFET amp, you can 
get better than -50dB at full output, but the practical value of this is 
debatable. It's probably useful only in rather rare cases, when several 
hams live and operate very close together in distance and frequency.

I think that -35dB is a level good enough to use as a target value for
3rd IMD in general purpose linear amps. Most HF transceivers are
significantly worse than that at 100W, specially on 10 meters.

Manfred

========================
Visit my hobby homepage!
http://ludens.cl
========================


More information about the Amps mailing list