[Amps] Ampleon BLF189XR 1900W LDMOS transistor

[Manfred Mornhinweg]

Kevin, Jim, and all,

 > W6PQL.com
 >
 > He originally started using the LDMOS devices on 6m and up and then
 > built the HF SSPA. I believe the schematics are in the article.

Yes, I had a look at his site. His BLF188XR HF amplifier has a 1:9 
imepdance ratio at the output, and a 50V supply, so it can only produce 
about 750-800W without touching saturation, and 1 to 1.1kW with 1dB 
compression.

The same transistor used with a 1:16 transformer would be able to run at 
1500W with less than 1dB compression, but the dissipation would become 
unmanageable. So that would need two of these transistors, with 1:16 
matching, and the big problem is making a bifiliar feed choke that 
actually works as such, up to 30 (or 54) MHz. When the bifiliar choke's 
leakage inductance gets high enough to be a significant reactance at the 
1.6Ω drain impedance, the efficiency and linearity of the amplifier 
suffers. And it's MIGHTY hard to make a feed choke that fullfills this 
requirement. The leakage inductance needs to be less than 3nH or so for 
30MHz, and well under 2nH for 54MHz.

Thus a better aproach is two separate modules, each using one BLF188XR 
and a 1:9 transformer, and then combine the outputs. The resulting 2.8Ω 
drain impedance is still hard enough to handle...

 > He has built a two device amp for someone using two BLF188XR's.

It's the way to go, at this time.

 > His filter is rated at 1500W. I'd feel better if it was 2K or higher.

I wonder what the current rating is, for the capacitors he used. When I 
have been shopping for such capacitors, the current rating is either not 
given, or is way too low for this application, or the capacitors are 
really expensive. Instead the voltage rating is no problem, there are 
plenty of capacitors available with high enough voltage rating.

Some of the capacitors in a 1500W low pass filter need to carry roughly 
10A of RF current during normal operation, and more under some high SWR 
conditions.

 > An LPA design I saw in the handbook used a Cauer architecture for the
 > filter. The capacitor values are critical and they are expensive. You
 > also need to tune the filter to land the nulls right on top of the odd
 > harmonics.

For those reasons I prefer staying with low-ripple Chebyshev filters.

 > The LDMOS devices are designed to tolerate 65:1 SWR, 3:1 probably
 > wouldn't be a problem so where is the need for a tuner?

Caution with that rating! It holds true only as long as the dissipation 
capability of the transistor/heatsink combination isn't exceeded. That's 
why this 65:1 survival is usually specified in pulsed service only, 
typically with 1% duty cycle!

A 65:1 SWR can result in very low to very high power dissipation, 
depending on the scircuit used, and on the phase angle of the load. A 
ham accidentally transmitting into the wrong antenna, and taking a few 
seconds to notice, can easily burn out the transistor, even if the SWR 
never got as high as 65:1. That's why an SWR protection circuit is still 
needed. The good thing is that the transistor will easily survive the 
high SWR for long enough until the SWR protection kicks in, while some 
older transistors might burn out faster than the protection circuit 
could react.

The fact that the dissipation varies widely whith the phase angle, even 
at constant high SWR, allows advertisers to give great-looking demos of 
the device's ruggedness: They will adjust the load to 65:1 SWR at such a 
phase angle that the dissipation is low enough, and then demonstrate to 
the public how the amplifier can take a brick on the key into that load. 
  If any of you has a chance to witness such a demonstration, PLEASE ask 
nicely for permission to vary the phase angle setting of the load, 
slowly, through the full 360 degrees, while the brick remains on the 
key! You won't get permission! :-)

 > If you're trying
 > to load a wet string with a tube amp your going to need an outboard
 > tuner so why not use it with an SSPA? The auto tuner argument is a straw
 > man argument.

I agree.

I think a good approach is building legal limit solid state amps without 
built-in tuners, but include safe, effective SWR protection. The 
protection should be such that it limits the peak current, peak voltage, 
and average power dissipation of the transistors, to safe values. So the 
limit is NOT a fixed SWR of 2:1 or 3:1, but rather the acceptable SWR 
varies depending on phase angle, frequency, and so on.

With such an amp, a ham who has antennas with a decent SWR in the middle 
of each band can use the amp without any tuner over the full bands. At 
the band edges, with SWR like 2:1 or 3:1, the power will decrease by 
maybe 1 dB. So what. And if he wants to use a nonresonant antenna, he 
must use an external tuner, which can be manual or automatic at his choice.

The tuner should be considered a part of that antenna system, NOT a part 
of the amplifier!

 > These LDMOS devices are NOT like your fathers Bipolar and JFET, or even
 > more modern MOSFET devices. They won't pop when facing a 2:1 SWR and
 > have amazing amounts of gain at HF.

Yes. I would condense the differences like this:

- Modern LDMOSFETs have lower capacitances than the old VMOS or 
TMOSFETs. This gives them much better high frequency response.

- They have very sensitive gates. So they need far lower drive power 
than the older devices. BUT their gates are also much more fragile! Old 
MOSFETs could take ±20 or ±30V on the gates, some even ±40V, while 
modern LDMOSFETs have much tighter limits, like +11/-6V. Unfortunately 
this compromises their usability in certain high efficiency operation 
modes. Even simple class C can be a problem for many modern LDMOSFETs.

- Modern LDMOSFETs are more cost-effective.

- The current "extra rugged" crop of LDMOSFETs have better avalanching 
capabilities and larger peak current handling than power-equivalent 
older MOSFETs. This gives them the ability to survive high SWR, but only 
within their dissipation limits!

 > They do require copious amounts of clean DC, 50V, 60-65A for 1500W.

65A at 50V for 1500W output means an amplifier that is only 46% 
efficient at its peak output. That's a very poor amplifier indeed! That 
would be almost the efficiency of class-A! Instead a good class AB 
linear HF amplifier with LDMOSFETs should achive over 60%. I have done 
65 to 68% using cheap switchmode MOSFETs, at the 1dB compression point! 
Some builders claim 75% efficiency at 1dB compression, with LDMOSFETs. 
Considering that the theoretical limit is 78% without any compression, 
it seems plausible.

Given the many 50V switching power supplies cheaply available on the 
surplus market, it's not really a problem to get 65A at that voltage. 
But managing the heat in a 46% efficient legal limit amplifier is indeed 
a problem.

I want an amplifier, not a room heater with auxiliary RF output! :-)

> From several Youtube videos it appears that the LDMOS FETs.are quite
> capable of withstanding opens and shorts indefinitely. 

Take them with a grain of salt. A few days ago I watched one advertising 
the ruggedness of a certain LDMOSFET. The test was done with a _pulsed_ 
signal source at a low duty cycle, and while the narrator _said_ that 
the amplifier could survive 65:1 at any phase angle, the video did _not_ 
show this! The part where supposedly the engineer changed the load's 
phase angle through the full range was cut out of the video!

What was shown, quite impressively, was the guy drawing sparks from the 
amplifier's output by shorting it with a screwdriver. The power supply 
voltage and current were shown too. Thanks to the pulsed signal source, 
and to water cooling, the total power input always remained well within 
the dissipation capability of the LDMOSFET. But that test just showed 
two conditions: Open and shorted. And hams often use continuous 
transmission rather than pulsed...

> The next question would be what about other components in the finals and
> lowpass filters and how would they react to continuous 3:1? Coils,
> capacitors, etc.

Yes, that's an important point. A low pass filter designed to just 
barely survive 1500W at 1:1 SWR will fail when run for any significant 
time at a high SWR. And of course the filtering characteristics of the 
filter will suffer when the load isn't a clean 1:1. So, I would say that 
we should still try to keep our antennas well tuned and matched. But 
losing our sleep over a 1.5:1 SWR at the band edges is pointless.

> How would the power supply to the FET output stage be affected? Would the
> excess voltage or current at the amp LPF output cause the signal to degrade
> due to starvation of either?

That's why I think that a good solid state amplifier should use a 
protection circuit that senses drain peak voltage, and drain current, in 
addition to SWR and power. The combined result from these four signals 
should be fed into the ALC circuit. So that the amplifier runs at 1500W 
PEP while everything is fine, but reverts to SWR-limited, 
current-limited or voltage-limited operation when the antenna isn't good.

With such a protection circuit, the power supply will never be called to 
deliver more than the normal current, and the transistors will never be 
stressed above their continuous safe ratings, except very briefly during 
the time the control circuit takes to kick in - and that amount of time 
is easily survivable by modern ruggedized LDMOSFETs.

If the voltage and current sensing circuits also detect the relative 
phase, then it's possible to derive MOSFET dissipation just from them. 
In this case the SWR sensor is no longer required. Just the current and 
voltage sensors. This is convenient, because with SWR sensors there is 
always the problem of whether to put them before or after the low pass 
filter! Put them before the filter, and they will always show some 
elevated SWR, because of the amplifier's harmonics bouncing off the low 
pass filter. But put them after the low pass filter, and they won't 
protect against the amplifier trying to transmit above the selected 
filter's cutoff frequency!

The ancient Kenwood TL-120 uses two SWR sensors for this reason. One 
before and one after the low pass filter. With the first being less 
sensitive than the second.

Manfred

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