EMRFD Message Archive 6545

Message Date From Subject
6545 2011-08-16 09:50:09 Dan Mills Norton feedback power amplifier?
Hi all,

While seeking a quiet PA design, I have been reading the old papers on
norton feedback RF amplifier stages and the thought occurred that the
same thing may be applicable to RF power stages. The Clifton labs
'preamp' circuit manages almost a watt at 1db power compression after
all from a push pull variant.

We have some very high gain HF power fets available today, but the
linearity sometimes leaves a little to be desired, could a common gate,
transformer feedback amplifier be built around such a part (The source
would have to be isolated from the heatsink)?

With some of them, setting the transformer to provide say 11dB of gain
(10dB in practise) would leave almost 20dB of feedback to linearise the
thing.

Anyone here played with norton transformer feedback PAs?

Regards, Dan.
6546 2011-08-16 11:08:03 Chris Trask Re: Norton feedback power amplifier?
>
> While seeking a quiet PA design, I have been reading the old papers on
> norton feedback RF amplifier stages and the thought occurred that the
> same thing may be applicable to RF power stages. The Clifton labs
> 'preamp' circuit manages almost a watt at 1db power compression after
> all from a push pull variant.
>
> We have some very high gain HF power fets available today, but the
> linearity sometimes leaves a little to be desired, could a common gate,
> transformer feedback amplifier be built around such a part (The source
> would have to be isolated from the heatsink)?
>
> With some of them, setting the transformer to provide say 11dB of gain
> (10dB in practise) would leave almost 20dB of feedback to linearise the
> thing.
>
> Anyone here played with norton transformer feedback PAs?
>

It would seem to be a logical progression, but power FETs as well as
higher power bipolars do not function at all in common-gate or common-base
configurations due to the finite inductance in series with the gate or base
terminal. The result is usually an expensive power oscillator.

Some manufacturers many years ago made power bipolars where the base
contact was the bottom of the die so as to eliminate the inductance problem,
but western designers did not care for common-base power amplifiers the way
that Soviet designers did. Motorola actually made a dual common-base power
device, but I cannot recall the part number.

In addition to that problem, the time delay through power devices is
usually large to the point where any negative feedback configuration will
result in oscillation. I've been able to apply augmentation to 1W and 2W
bipolar amplifiers, but anything beyond that is not practical due again to
the base inductance.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
6547 2011-08-16 11:56:40 Dan Mills Re: Norton feedback power amplifier?
On Tue, 2011-08-16 at 11:07 -0700, Chris Trask wrote:

> The result is usually an expensive power oscillator.

I think we have probably all built some variation on that theme at some
point!

> In addition to that problem, the time delay through power devices is
> usually large to the point where any negative feedback configuration will
> result in oscillation.

Got to hate the miller effect!

Possibly a really silly idea, but run an emitter (or source) follower
with (possibly, might not be needed) transformer feedback from the
emitter to the base and a impedance step up in the emitter circuit to
give the power gain?

Ccb still hurts as we are now swinging the base real voltage, but Cbe
mostly goes away as does the miller cap. multiplier and emitter lead
inductance is usually by design small, possibly a better trick with FET
then a bipolar.

Of course the transformer primary will be a bit on the heavy side, but
with folks making 100W class PAs with only a 12V supply that is not
exactly unheard of.

I need some junker power devices and to do some experimenting.

Regards, Dan.
6549 2011-08-16 13:21:38 William Carver Re: Norton feedback power amplifier?
On Tue, 2011-08-16 at 19:56 +0100, Dan Mills wrote:
>
> On Tue, 2011-08-16 at 11:07 -0700, Chris Trask wrote:
>
> > The result is usually an expensive power oscillator.
>
> I think we have probably all built some variation on that theme at
> some point!
>
> > In addition to that problem, the time delay through power devices is
> > usually large to the point where any negative feedback
> >configuration will result in oscillation.
> Got to hate the miller effect!

Pushpull PA? Neutralize the Cdg with cross coupled series R-C? Doesn't
help delay though.


The source-gate transformer feedback amplifier is discussed in "Radio
Frequency Design" (W7ZOI), chapter 6. Pages 214-218 might be a useful
place to start.

W7AAZ
6550 2011-08-16 14:51:21 kb1gmx Re: Norton feedback power amplifier?
6551 2011-08-16 15:07:27 kb1gmx Re: Norton feedback power amplifier?
6552 2011-08-16 16:38:34 Dan Mills Re: Norton feedback power amplifier?
On Tue, 2011-08-16 at 22:07 +0000, kb1gmx wrote:

>
> To get a quiet PA [system] the driver stages have to be quiet.

Understood, and that actually is what I am working on, as I already have
an old BLW50/BLW96 block that will serve as the PA proper, but it needs
~ 5W PEP drive to get to full output.

> Most of the power output devices are fairly quiet but many of the
> driver chains use excess gain or take no care to minimize noise in the
> low level and milliwatt states and by time you get to the final
> the noise is already there.

Current thinking is to come out of the BPF at somewhere around 0dBm,
then a low noise norton amp to get it up to +10 followed by a push pull
variant to get to the hundred mW level, that bit is all easy, the
question then is what next?
The RD06 or RD16 hold some promise, if only I didn't want 160 - 6M, it
is loosing it more then a little at 6M.
It may be that the power stages of the driver end up being 24V devices
(at least the drain load impedance becomes easy to match!).

> Now Linearity is a different thing. Many of the good RFpower fets
> like the BLF177, MRF151/BLF278 are very linear with good IMD capability
> assuming the prior stages were clean. The problem is
> they are 28 and 50 volt devices. Very few of the 12Volt devices
> are as clean at power of 100W.

28/50V is not a problem, but actually I have that bit from an old
commercial/military rig, the issue is in the predriver/driver sort of
region (Which really would be more convenient on 12 or 24V).

> As to feed back Silconix used the VMP4 (stud mount RF Vmos FET)
> in a wideband amp that was if memory serves was 4db noise and
> 1db compression of several watts (memory says 4W). that would
> make a fine RF amp if the down stream stuff can stand that much.
> It would make a good driver too.

VMP4? Blast from the past!
I remember a circuit using those in an OLD VHF/UHF Handbook.

> The biggest problem with power fets is the Miller capacitance
> and that is a negative feedback problem already.

Same old problems, been fighting the Miller effect for years.

Regards, Dan.
6556 2011-08-17 06:50:58 cbayona Re: Norton feedback power amplifier?
Maybe that explains when I got the bright idea of using a Norton
amplifier configuration as a power amplifier because of it's
predictable power gain, but I could not get it to stop oscillating.

At 01:07 PM 8/16/2011, you wrote:
> >
> > While seeking a quiet PA design, I have been reading the old papers on
> > norton feedback RF amplifier stages and the thought occurred that the
> > same thing may be applicable to RF power stages. The Clifton labs
> > 'preamp' circuit manages almost a watt at 1db power compression after
> > all from a push pull variant.
> >
> > We have some very high gain HF power fets available today, but the
> > linearity sometimes leaves a little to be desired, could a common gate,
> > transformer feedback amplifier be built around such a part (The source
> > would have to be isolated from the heatsink)?
> >
> > With some of them, setting the transformer to provide say 11dB of gain
> > (10dB in practise) would leave almost 20dB of feedback to linearise the
> > thing.
> >
> > Anyone here played with norton transformer feedback PAs?
> >
>
> It would seem to be a logical progression, but power FETs as well as
>higher power bipolars do not function at all in common-gate or common-base
>configurations due to the finite inductance in series with the gate or base
>terminal. The result is usually an expensive power oscillator.
>
> Some manufacturers many years ago made power bipolars where the base
>contact was the bottom of the die so as to eliminate the inductance problem,
>but western designers did not care for common-base power amplifiers the way
>that Soviet designers did. Motorola actually made a dual common-base power
>device, but I cannot recall the part number.
>
> In addition to that problem, the time delay through power devices is
>usually large to the point where any negative feedback configuration will
>result in oscillation. I've been able to apply augmentation to 1W and 2W
>bipolar amplifiers, but anything beyond that is not practical due again to
>the base inductance.
>
>
>Chris Trask
>N7ZWY / WDX3HLB
>Senior Member IEEE
>http://www.home.earthlink.net/~christrask/
>
>
>
>------------------------------------
>
>Yahoo! Groups Links
>
>
>
Cecil
k5nwa


[Non-text portions of this message have been removed]
6557 2011-08-20 11:04:25 kb1gmx Re: Norton feedback power amplifier?
6558 2011-08-20 13:48:30 Dan Mills Re: Norton feedback power amplifier?
On Sat, 2011-08-20 at 18:04 +0000, kb1gmx wrote:

> Likely the 1 turn feedback winding was reversed. I did that
> once and because I got start and finish mixed up.

I suspect that EVERYONE who has played with norton feedback amps has
made this mistake (Sometimes more then once! Who Me? Never).....

> The norton amp has one big issue and is unsuitable for a final
> for that same reason. The output impedance is reflected in
> the input impedance. IF your load (aka antenna) is always a
> controlled impedance and all that may not be a problem.

I was planning a modest pad behind the pre driver to stabilise the
impedance seen further down the chain, but actually it sounds like these
would be best employed in the 0dBm - +20dBm or so region, with a
conventional PP stage used for pre driver and driver. Oh well, it was
just a thought.

> As to 160 to six 6 that's nearly five octaves and miller capacitance is only one issue.
> Layout, impedance control and ferites used are going to interact.
> just getting to 10m is not so easy.

160-10M I have done and know how to skin, that last 3/4 octave is a pain
however, and really does need a change in magnetic material.

The usual simplistic 'swamp the gate capacitance with resistance' style
of 'broadband matching' is also somewhat too lossy up there, I probably
need to tune the capacitance out on 6 and go resistor swamp on the lower
bands or something.

Fun for a wet evening.

Regards, Dan.
6559 2011-08-20 19:29:30 kb1gmx Re: Norton feedback power amplifier?