On Fri, 27 Aug 2004 19:28:39 +0100, Paul Burridge
wrote:

On Fri, 27 Aug 2004 08:07:02 -0700, John Larkin
wrote:

I think the RF guys (I'm not one!) call an amplifier "linear" if the
RF output amplitude follows the input drive amplitude. You can do this
with a transistor that has very low quiescent bias. So "linear" does
not mean "class A" to them. The key here is that an RF amp has a tuned
output, whereas an audio amp doesn't. So the lopsided bias would
normally produce intolerable distortion in something like audio, but
the tuned output circuit changes the pulsey-looking collector/drain
current back into a nice sine wave. So you don't need a lot of idle
current, and the transistor really amplifies half of the incoming sine
cycle.

Most mosfets are pretty nicely linear (ie, straight-line Ic/Vd curve)
beyond the initial knee. You could get gobs of watts at zero standing
current, but then you'd have some zero-clipping (no output) for the
smallest drive levels, so a little idle current helps.

The only sensible way to do it AFAICS is to operate the MOSFET in
class C as a high speed switch and reconstruct the pulsed output into
a sine wave carrier by means of a suitable tuned circuit. I wouldn't
consider driving a MOSFET for RF use in any other way. The efficiency
should be pretty darned good, too.

That means, to get a linear amp, the input signal has to be converted
to PWM gate drive. That's hard to do at high frequencies. At 300 MHz,
a power mosfet doesn't much look like a high-speed switch any more.

John

On Fri, 27 Aug 2004 13:00:56 -0700, John Larkin
wrote:

That means, to get a linear amp, the input signal has to be converted
to PWM gate drive. That's hard to do at high frequencies. At 300 MHz,
a power mosfet doesn't much look like a high-speed switch any more.

Certainly not at that kind of frequency! But for the lower HF bands,
it's *perfectly* feasible.
--

"What is now proved was once only imagin'd." - William Blake, 1793.

On Sat, 28 Aug 2004 00:28:21 +0100, Paul Burridge
wrote:

On Fri, 27 Aug 2004 13:00:56 -0700, John Larkin
wrote:

That means, to get a linear amp, the input signal has to be converted
to PWM gate drive. That's hard to do at high frequencies. At 300 MHz,
a power mosfet doesn't much look like a high-speed switch any more.

Certainly not at that kind of frequency! But for the lower HF bands,
it's *perfectly* feasible.

Have you actually built a class C linear RF power amp? Tell us how it
works.

John

On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/
--

"What is now proved was once only imagin'd." - William Blake, 1793.

In message , Paul Burridge
writes
On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/

Don't forget that you cannot really modulate a 'linear' amplifier by
varying the supply rail (which is what 'plate & screen' mod does). The
modulated stage has to be non-linear (eg Class C) where the power output
varies as the square of the supply volts. If the PA was biassed in Class
A, there wouldn't be any modulation.
Ian.
--

On Sat, 28 Aug 2004 10:56:03 +0100, Ian Jackson
wrote:

Don't forget that you cannot really modulate a 'linear' amplifier by
varying the supply rail (which is what 'plate & screen' mod does). The
modulated stage has to be non-linear (eg Class C) where the power output
varies as the square of the supply volts. If the PA was biassed in Class
A, there wouldn't be any modulation.

Certainly there would be no *amplitude* modulation, but that doesn't
preclude FM and various other schemes.
--

"What is now proved was once only imagin'd." - William Blake, 1793.

On Sat, 28 Aug 2004 10:56:03 +0100, Ian Jackson
wrote:

In message , Paul Burridge
writes
On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/

Don't forget that you cannot really modulate a 'linear' amplifier by
varying the supply rail (which is what 'plate & screen' mod does). The
modulated stage has to be non-linear (eg Class C) where the power output
varies as the square of the supply volts.

---
Nonlinear? Yes. Class "C"? No.

From http://sound.westhost.com/class-a.htm :

"
Class-A Output device(s) conduct through 360 degrees of input cycle
(never switch off) - A single output device is possible. The device
conducts for the entire waveform in Figure 1

Class-B Output devices conduct for 180 degrees (1/2 of input cycle) -
for audio, two output devices in "push-pull" must be used (see
Class-AB)

Class-AB Halfway (or partway) between the above two examples (181 to
200 degrees typical) - also requires push-pull operation for audio.
The conduction for each output device is shown in Figure 1.

Class-C Output device(s) conduct for less than 180 degrees (100 to 150
degrees typical) - Radio Frequencies only - cannot be used for audio!
** This is the sound heard when one of the output devices goes open
circuit in an audio amp! See Figure 1, showing the time the output
device conducts (single-ended operation is assumed, and yes this does
work for RF)

Class-D Quasi-digital amplification. Uses pulse-width-modulation of a
high frequency (square wave) carrier to reproduce the audio signal -
because of frequency limitations (and the fact that they nearly all
seem to sound disgusting), many are only suitable for industrial
control of motors and loud but crappy sub-woofers (this may change if
transistors with an infinite bandwidth become available soon - yeah,
right!) All Class-D amps have a major limitation in the output filter,
whose response is highly dependent on the load impedance.
"

If the PA was biassed in Class A, there wouldn't be any modulation.

---
There could be; all that would be required would be for the gain of
the stage to vary with the modulating input. I don't believe there's
a constraint on class A biasing which inherently precludes a class A
stage from being modulated.




--
John Fields

On Sat, 28 Aug 2004 10:36:21 +0100, Paul Burridge
wrote:

On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/

---
Let's say that you have an audio amp with an input resistance of 1000
ohms and that, with a 1V input, it puts 10 volts across an 8 ohm load.

That's a voltage gain of

Vout 10V
Av = 20 log ------ dB = 20log ---- dB = 20dB
V in 1V

and a power gain of

Pout 12.5W
Aw = 10 log ------- dB = 10log -------- dB ~ 41dB
Pin 0.001W



Now let's say that you up the input voltage to 2V and that the output
voltage goes to 20V. That's still a voltage gain of 20dB and a power
gain 41dB.

Finally, let's say that no matter what voltage you apply to the
input,(up to some reasonable limit) the output voltage is always 10
times higher. That's a linear amplifier.


It's no different with RF.

Let's say, for example, that we have an RF amp with a 50 ohm input and
output impedance and that with a 1 watt input it puts out 10 watts
That is, it has 10 dB of power gain. If it's a linear amplifier and
we exercise its input according to the following table, the
relationships given in the table will be true. If it isn't, they
won't be.

Pin Pout Aw Ein Eout Av
W-50R W-50R dB V-50R V-50R dB
------------------------------------------
1 10 10 7.07 22.4 10
2 20 10 10.0 31.6 10
3 30 10 12.3 31.6 10
4 40 10 14.1 44.7 10
5 50 10 15.8 50.0 10
6 60 10 17.3 54.8 10
7 70 10 18.7 59.2 10
8 80 10 20.0 63.2 10
9 90 10 21.2 67.1 10
10 100 10 22.4 70.7 10

So, that should take the mystery out of why it's called a
"linear amplifier" :-)


--
John Fields

On Sat, 28 Aug 2004 10:36:21 +0100, Paul Burridge
wrote:

On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/

Define "linear"? You must be joking.

I'll take that as a "no" to my question. Not surprised.

John

On Sat, 28 Aug 2004 08:43:21 -0700, John Larkin
wrote:

On Sat, 28 Aug 2004 10:36:21 +0100, Paul Burridge
wrote:

On Fri, 27 Aug 2004 17:43:34 -0700, John Larkin
wrote:

Have you actually built a class C linear RF power amp? Tell us how it
works.

It depends on how you define "linear" basically. But the term is a
total misnomer in RF amp terminology and very misleading. I can't
understand how it got there. :-/

Define "linear"? You must be joking.

I'll take that as a "no" to my question. Not surprised.

Actually I've built *several* class C RF amps, John. However, I
wouldn't call any of them linear. You will be aware than linearity
starts to go out of the window when Class A slides into Class AB and
beyond. Let's not have an argument over definitions. It's an open
invitation to John Woodgate. ;-)
--

"What is now proved was once only imagin'd." - William Blake, 1793.