[Amps] Price per Watt Conversation

[John Lyles]

There has been excellent discussion here about the new LDMOS devices and 
their potential use for HF amplifiers. Manfred (XQ6FOD), Karl-Arne 
(VK5KLT) and Leigh (VE7RF) have each made strong points about the 
ratings of the new parts, aspects of cooling these tiny power devices, 
and the concerns about linearity. Others have also given good advice 
here. This is such a quality group of technical-minded hams loving 
amplifiers! Without repeating it all here, I recommend re-reading the 
comments of the past few days above this, if you really want to 
understand the challenges of using the new high power transistors.

I looked at the new NXP MRFX1K80H preproduction datasheet. Its quite 
obvious from the gain compression graphs that these devices would not be 
operated linearly about about 800 watts. 65 volts is moving into good 
direction for matching. Remember that the output Z being proportional to 
the Vdd and Power, this sort of higher voltage will lead to wider 
bandwidth circuits with less transformation. I remember an article in 
the 1980s in the old RF Design magazine that Gary Breed published. It 
was by a fellow at Siliconix at the time (bought out by Vishay 
subsequently), where Vertical diffused MOS (VMOS) was the new 
technology. He published a HF PA circuit that used no output matching at 
all, running a device at 125 VDC. I'm sure it was rich with harmonics 
but it did make the point! Lateral DMOS (LDMOS) is limited in how high 
the voltage can get, being silicon technology. Wider bandgap Gallium 
Nitride can tolerate higher voltage and temperatures, and has an edge on 
this, although it hasn't been pushed like LDMOS is right now (more 
expensive fabrication). It has taken a lead in some higher frequency 
designs however.

I have been faced with similar design decisions at work, where solid 
state is making inroads into the territory of thermionics, but there are 
so many caveats and pitfalls to watch for. I am concerned with VHF 
power, not HF, but so many of the devices work all the way up to 450 MHz 
now that it's not so different except for the size of the pallets.

About 9 years ago, we were looking at how to get 5 - 20 kW of RF at 200 
MHz in a more efficient way than we did in the 1990s, piling up MRF151G 
DMOS devices. One amplifier we ran for 20 years used 32 push-pull 
devices to get 5.5 kW, much like the TV transmitters were using in that 
time. Before LDMOS was above 1 KW, I was looking at some new devices 
from Microsemi. They had bought Advanced Power Technology, which had a 
development going in Silicon Carbide transistors. In 2008 they were 
promoting Silicon Carbide static induction FETs. One part, 0150SC-1250M 
was a single ended transistor to produce 1250 watts in class AB, at 155 
MHz. Another part was for 2200 watts at 406-450 MHz, the 0405SC-2200M, 
in 2010. These parts had 250 Volt max Vds ratings. We bought some 
0201SC-800L parts that were for 200 MHz at 800 watts each transistor 
(not push pull!). We proceeded to kill one with improper bias 
sequencing. They were just like tubes, bias first, then HV. Bias was 
negative voltage, and if it went to zero, the drain current blew up.
Eventually, we got it right. Meanwhile Microsemi proposed even higher 
power, with a 1600 watt target in one transistor. One limitation of 
these devices was that they were fabricated so that they would be common 
gate, not common source connection. The power gain was < 9 dB which was 
not competitive with LDMOS at the time. These were prototypes so we paid 
for fixtures (test pallets) as well as a few transistors. We were 
getting close to needing an answer. As it turned out, LDMOS came to the 
rescue, and we got commercial amplifiers for 5.5 kW using 8 x 50V 
push-pull transistors. We are pulsed but linearity is also important. 
The next requirement is for the 25 kW stage, and again, LDMOS may be the 
solution.

In Europe right now, a transistor power amplifier is being tested for 
use in a ~1.5 MW CW linear application at 200 MHz. So far the 200 kW 
level is working and integration into the larger amplifier is planned. 
It will fill a floor of a large building when done. Hopefully more 
information will be available later this year. It is an impressive 
undertaking for scientific RF generators.

73
John
K5PRO