[Amps] SS Amps are expensive...why?

[Manfred Mornhinweg]

Hi Art, and all,

> Thank you Manfred for having the nuts to tell it like it is!!!

Isolation in the distant wilderness makes me nutty! ;-)

> Especially the last couple? of paragraphs and the statements that no 
> solid state mosfet amp is clean!

Facts are facts. But to avoid giving a biased impression, I would like
to add that no tube type amp is clean either!

> I'd like to point out that even the best bipolar transistors are far 
> from linear, look at the curve tracer results and see how severely
> the hfe differs as the base current increases.

That's right, but hfe is irrelevant at RF. Instead the current gain at
RF is determined almost exclusively by the transistor's Ft and the
operating frequency.

At lower frequencies the hfe becomes important. Bipolar transistors show
a certain range of collector current, reasonably wide, in which the hfe
  is quite constant, and what's more important, this range covers the
useful current range used, for example, in a class-A driver stage.
Instead MOSFETs reach a reasonable linearity only at pretty high drain
currents, which is usually above the range used in said class-A driver
stages. For that reason MOSFET class-A drivers tend to be far less
linear than drivers made with bipolar transistors.

For that reason the predriver stage in my SDR is using an ancient
2N5109, rather than a modern MOSFET! I tried to build that stage around
an RD06HHF1, and the standing current required to get acceptable
linearity was far too high.

> The fact is evenn the best conventional bipolar rf amps aren't
> linear! And there's no way to make a mosfet linear-case closed.

No amplifier ever will be absolutely linear. Regardless of whether it 
uses bipolar transistors, MOSFETs, triodes, tetrodes or pentodes. The 
point is making an amplifier LINEAR ENOUGH for a given application. And 
that can be done with either of the mentioned devices. It's a matter of 
deciding what degree of linearity is required, and then design for it. 
The more stringent the requirements are, the more complex the amplifier 
will become. Typically by requiring linearization loops, which means 
adding stages.

Take the HiFi audio scene as an example: They use bipolar transistors, 
MOSFETs, and there is a group of dieharts who use tubes. Interestingly 
those who swear that tubes sound better seem to like some of the 
specific distortion tubes cause. Instead the cleanest amplifiers are 
invariably solid state ones. Cheap audio amplifiers, even those 
implemented in ICs, easily achieve an IMD level of -60dB, and a 
full-blooded HiFi amplifier will deliver 100 to several hundred watts 
with an IMD level better than -80dB. Often much better. And that's done 
with common, "crummy" transistors, in a frequency range where the 
non-linear hfe dominates gain! It's done by using enough stages to have 
a lot of excess gain, and then using this excess gain in negative 
feedback to linearize the whole thing.

I see no reason why we can't do the same at RF. While audio guys have a 
choice between applying direct feedback over the whole amplifer, or 
separately over each stage, we RF guys need to apply it per stage, due 
to delay and phasing problems. But nothing forbids us for applying 
ENVELOPE feedback over the whole transmitter! After all, unlike audio 
guys, we don't need to accurately reproduce a waveform. All our signals 
are essentially pure sine waves, and we can reconstruct a good sine wave 
in a lowpass filter, regardless how bad the waveform is before the 
filter. We only need to get the envelope and the phase correct. By 
feeding back the envelope and phase into a correction system, we can 
transmit signals with the IMD down at -80dB. And modern SDR 
implementations that do this are widely available. The software for them 
is free. A few different hardware options are available.

Now of course this deviates a lot from the main topic of this forum, 
which is add-on amplifiers that mostly amplify a 100W signal into a 
1500W one. But this also illustrates that add-on amplifiers should be 
considered obsolete. There are simply too strong advantages in 
integrating the 1500W stage into a transmitter, rather than adding it 
externally.

Given the sizes and costs, this approach will produce something like a 
legal limit amplifier with built-in transceiver. The one I'm developing 
should end up being a black box with just a power switch and a speaker 
on the front panel, and an AC power inlet, antenna connector and 
ethernet connector on the back panel. Most of the signal processing, and 
the entire user interface including microphone and PTT, are centered on 
a computer. The black box can easily be remote-located, for example at 
the antenna tower, with just a power line and an ethernet cable (or 
optical fiber) running there.

That's bad news for hams who love playing with many knobs, buttons, 
needle meters and so on, but it would produce better performance at 
lower price.

Manfred

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