[Amps] SPE 1.3K-FA and FTDX-5KMP / IMD at rated PEP

[Manfred Mornhinweg]

I agree on the fact that there is currently too much hype about signal purity on 
the side of some hams, while other hams coulnd't care less about signal purity, 
and just want to get out as loud as possible. So we reach the absurd situation 
that ham A is hesitant to put out a signal with the 3rd IMD suppressed "only" by 
30dB, because he could interfere his neighbor, ham B; While at the same time ham 
B uses a radio with the ALC defeated to get "full unlimited" power output, with 
the 3rd IMD 10dB down from PEP, and fills the band with crap for ham A.

I think we should take a certain signal purity level as a minimum base, and 
somewhere in the back of my mind I have a memory of some law requiring -28dB 
minimum suppression for the 3rd IMD products. And from there, of course, we 
should strive for better quality as technology and budget allow, but without 
losing too much sleep over it.

What is very clear is this: When your exciter puts out a signal with -28dB IMD, 
there is no appreciable difference in the final signal quality whether you use 
an amp with -32dB IMD, or one with -80dB IMD. And that -28dB IMD is typical for 
any usual ham transceiver run at 100W power, or close to it.

> Given the age of those designs, it is amazing that
> the new designs are so sloppy.

I see that time and again "designers" of modern amps copy from 40 year old 
application notes, instead of designing a new (and hopefully improved) circuit.

> A pair of BLF-188A based amplifiers with splitter/combiner each running
> 800W PEP for 1600W PEP total should be able to make the -40 dB(PEP) as 
> long as the designer takes care to keep the bias stable and avoid any
> saturation of the output transformers.

In this regard it's very important to understand a fundamental practical 
difference between tubes and MOSFETs: Tubes in ham amps are usually run close to 
their voltage, current and power dissipation limits. MOSFETs instead are usually 
run close to their voltage and power dissipation limits, but relatively far 
below their current limits. So, with tubes we actually milk them for the most 
power we can get with enough linearity. With MOSFETs instead, we milk them for 
the most power we can get without exceeding their dissipation limit, and that 
places the designer in a position to trade efficiency for signal purity.

This in turn means that if we want to get very low IMD from a MOSFET amplifier, 
by the simple method of using devices far below their maximum output power 
capability, then we have to design the impedance matching in such a way that the 
devices work at a relatively low RF voltage, relative to the supply voltage, 
while still running high current. The result is lousy efficiency. So we need 
more MOSFETs, a big power supply, a big heatsink and fans, etc, and the final 
product is big, heavy, clumsy, and very expensive.

This attracts designers to go the other way, and load the MOSFETs to a higher RF 
voltage, its peak value close to the supply voltage. That allows much larger 
output power from a given MOSFET, at the same dissipation, thus better 
efficiency, lower cost, smaller size, lower weight, etc. But the IMD gets 
higher. The game played by designers of solid state equipment - transceivers and 
amplifiers alike - is how far they can push this before the IMD becomes 
unacceptable to the customer or to the law.


Another fundamental difference between tubes and MOSFETs is the internal 
capacitances. With tubes they are fixed, while in MOSFETs they vary with the 
instantaneous drain-source voltage, which means that they vary over the RF 
cycle, and their average value varies with the signal envelope.

These modulated capacitances create phase distortion. The lower the drain 
voltage gets, the faster the capacitances increase, so there is a "soft 
saturation" effect, in that driving the amp closer to voltage clipping increases 
the amount of phase distortion in a progressive way. But even if an amp is 
driven way lower than to saturation level, there is still some capacitance 
modulation remaining, and thus some phase distortion. That's why uncorrected 
MOSFET amplifiers are almost inevitably less clean than tube amplifiers using 
low distortion tubes, such as certain tetrodes.

The good news are three: One is that well applied predistortion can completely 
cancel this unwanted phase distortion. But the technique doesn't lend itself to 
add-on amplifiers. It's better suited to integrated radios.

The second good news is that plain simple negative feedback reduces this 
distortion, and that modern MOSFETs have much higher gain than the typical ham 
add-on amplifier needs. So it's a relatively simple matter to design an amplifer 
to require 80W or so of drive, which means needing only 13dB of gain, and then 
use the excess 15 to 25dB of gain available from modern VHF/UHF MOSFETs on HF to 
implement heavy negative feedback. Some amplifiers actually do this, but too 
many are designed to be excited with flea power, so they have to use lower 
negative feedback, resulting in higher IMD. This is a case where better design 
is easy, but requires the designer to realize the problem.

And the third good news is that all the time new LDMOSFETs are hitting the 
market, that have lower and even lower capacitances. These are UHF-capable 
devices, which have capacitances so small that at HF they are starting to become 
negligible - and that means that the phase distortion caused by them also is 
starting to become negligible!

Last saturday I had a QSO with CE7MCK, who was using his homebrew kilowatt 
amplifier. It uses ten IRFP450 MOSFETs in a 2x5 push-pull/parallel arrangement. 
These are cheap, ubiquituous, old technology switching MOSFETs. Pressing them 
into RF service is rather daring, as their capacitances are huge. That amplifier 
works well on 160 and 80 meters, though. On 40 it's at its limit, creating a 
rather lousy signal, with audible distortion in addition to easily measurable 
splatter. CE7MCK is very aware of this, and is working on a version 2 using 
better suited MOSFETs, which hopefully should be usable through 10 meters.

We did lots of tests that day. We had good conditions, with his station being 
just one easy ionospheric hop away from mine on 40 meters. So I measured the IMD 
while he drove his amp at different levels, and with different idling current 
settings. The results were that of course as soon as he drove it close to 
saturation the distortion increased (duh!), but also that even well below 
saturation there was significant distortion. On 80m that kind of distortion was 
far less, a good indication that it was coming from the modulated capacitances 
of the MOSFETs.

As I see it, we can build a dirt cheap amplifier using many small, high voltage 
low current switching MOSFETs, that provides acceptable although not stellar 
performance, through 10 meters. Or we can use some of the modern LDMOSFETs, at 
higher cost but not outrageously so, to make a pretty clean amplifier. In fact, 
UHF-rated LDMOSFETs should be free enough from phase modulation effects that we 
can treat them pretty much the same way as tubes, in the regard that the IMD 
will be dominated by amplitude distortion arising from plain simple saturation. 
And if we reach that point, it's relatively simple to increase efficiency by 
means of modulating the bias to make the amp run linearized class C over most of 
the envelope.

In this latter regard, I have been thinking about running tubes in linearized 
class C. But I run into trouble with the screen grid specs. Tubes with a 1500W 
plate rating, that will burn out with just a few watts of screen dissipation, 
are not really suitable for this! In that regard MOSFETs are much better.

> ##  Now comes along SPE..with its  2.0K...running a whopping 2  kw pep output  from
> just SIX   MRF-151Gs.    Now if SPE can achieve 2000w pep from 6 x devices, then
> yaesu should be able to get  666 watts pep  from  2 x devices.   Can you imagine the
> FTDX-5K  running 666 w pep out ?      Well it would fare a little better than the SPE junk.

Jim, you are comparing apples to oranges. The MRF-151G is a gemini device rated 
at 300W output and 500W dissipation. Six of them will gladly provide 1500W, and 
with some goodwill and courage 2000W PEP isn't impossible.

The FTDX-5000 instead uses two VRF-150 MOSFETs, according to Yaesu's website. 
Those are single (not gemini) devices rated at 150W output, 300W dissipation. 
They are very comfortable at 200W total output, but certainly not 666W.

So I think you didn't notice the "G" in the part number of the devices used by 
that SPE amp!

I'm very curious about those SPE amps, and would love to get an opportunity to 
test one. I will keep asking around among local hams, but I don't think anyone 
will buy one of those amps, given the price. Maybe someday I find a ham on the 
air who is running one, and has a 2-tone generator at hand, so we can do some 
testing.

Manfred


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