[Amps] Fwd: Re: Fwd: Linear Amplifier Tuning---PROPERLY!

[Gary Schafer]

Hi Jeff,

I haven't seen these responses come thru AMPS yet? Wonder if you have?

I think that I understood what you were saying when you talked about
harmonics mixing to cause IM3. As I understand the thesis 2xF1 - F1 will
cause modification / distortion of the fundamental frequency by the IM
product falling at the same frequency as the fundamental. If the IM product
is in phase it will add to the amplitude of the fundamental or subtract
depending on how far out of phase it is. The phase of that IM3 product can
be dependent on what the second harmonic frequency sees as a load impedance
on the device.
But that seems to only be a problem at wide modulation frequencies of
several megahertz? With our narrow band  operation phase shift should not be
a problem for IM3 of the fundamental.

What I was referring to was that typical IM3 is expressed as 2xF1 +- F2
where F1 and F2 are two tone signals causing the IM3 product (2xF1-F2) to
fall next to the F1 and F2 frequencies (in band).  (note that 2xF1 - F1
falls right on the fundamental of F1)

I see that the thesis went on to say that IM5 products mix with harmonics to
modify the level of IM3 products at higher power and that is the reason for
seeing IM3 levels actually dropping somewhat at certain power levels as
power is increased while IM5 levels increase.

73
Gary  K4FMX

> -----Original Message-----
> From: jeff millar [mailto:wa1hco at wa1hco.net]
> Sent: Tuesday, December 27, 2011 3:53 PM
> To: garyschafer at comcast.net
> Cc: 'Amps'
> Subject: Re: [Amps] Fwd: Re: Fwd: Linear Amplifier Tuning---PROPERLY!
> 
> Thanks, Gary...  I'm trying to be helpful.
> 
> On the subject of harmonics vs 2xF1.  When I mentioned harmonics before,
> it
> wasn't as
> clear as could be.  I'm not referring to harmonics at the amplifier
> output, I'm
> referring to
> harmonics at the _device_, before the output network.  The plate,
> collector or
> drain
> waveform is _not_ a sine wave for any amplifier other than class A.  The
> clipped
> plate/collector/drain waveform generates a lot of harmonics at the
> device that the
> output matching network filters out to make the clean RF sine wave sent
> to the
> antenna.
> 
> The device is switching on and off (if not class A) and generally acting
> non-linear, so in
> addition to generating harmonics, it also mixes all the harmonic
> products and
> generates
> all those inter-modulation frequencies.
> 
> The output matching network puts the correct impedance on the device for
> best
> performance, and whether that makes a conjugate match, load line match,
> or whatever
>   is not at issue here.  What matters is, What does the output matching
> network
> present
> as an impedance at the 2nd harmonic, 3rd harmonic, etc?  Tube Pi-
> Networks have a
> capacitor across the plate terminals and present a low impedance for the
> higher
> harmonics.  This has some effect on the phase of the 2nd harmonic
> currents running
> around the tube's output network.  At the high harmonics, the parasitic
> inductor, stray
> wiring may presents a higher impedance to the plate which again changes
> the phase
> of the harmonics currents.
> 
> The output 3rd order IMD appears at 2xF1 +/- 1 x F1.  But that IMD also
> includes
> components from 3 x F1 +/- 2 x F1, 4 x F1 +/- 3 x F1, etc and all those
> components
> land on the exact same frequency as the 3rd order components.  When two
> RF signals
> land on the same frequency, they can add _in phase_ and make a larger
> signal or
> add _out of phase_ and make a small signal.  Since the output network
> doesn't
> control
>   the phases of the high harmonics reflected back to the device, it's
> impossible
> to predict
> how all those components will add up.
> 
> VHF output matching networks tend to use transmission lines and those
> have
> resonances
> at or near the higher harmonics...which tends to present a wide range of
> impedances to
> the devices at those higher frequencies...resulting in even less
> predictability
> and control.
> 
> All the IMD mixing takes place at the non-linear device...namely the
> tube/transistor/FET.
> The mixing depends on the phases and amplitudes of the all the currents
> running
> around
> at the various harmonic frequencies of all the components that make up
> the modulated
> signal.  The resulting mixing products that land near the fundamental
> are properly
> matched by the output network and make it to the antenna.  But mixing
> products that
> fall far from the fundamental, get filtered out.
> 
> As an aside, Class F amplifiers try to raise and control the impedance
> at higher
> harmonics
> to make the output transistor switch more completely and improve
> efficiency.
> But in
> general the impedance and phases of harmonic voltages and currents
> running around
> the output network is uncontrolled.
> 
> jeff, wa1hco
> 
> 
> On 11-12-27 01:16 PM, Gary Schafer wrote:
> > Nice article Jeff! I got through part of it but it takes some
> digesting!
> >
> > I think that the kind of IM products that we are most concerned about
> in our
> > amps is based on 2xF1 +- 1xF2 for 3rd order IM rather than the
> harmonics of
> > the fundamental. This assumes two signals being amplified by the
> amplifier
> > as in a two tone test. Of course things become a little more
> complicated as
> > the number of tones increase.
> > Of course we are also concerned with the higher orders of IM also such
> as
> > 5th, 7th etc. as they all can mix with one another once the amp is
> operating
> > in the non-linear mode caused by any order of IM products.
> >
> > I don't think that the phase of the output of the amp has much to do
> with IM
> > with narrow band signals such as we use for SSB. When the bandwidth is
> much
> > wider such as that used by cellular transmitter amplifiers maintaining
> a
> > near constant phase match over the bandwidth seems to be more
> important.
> >
> > Our output matching network is not providing an impedance match to our
> > amplifiers as the actual device (tube or transistor) output impedance
> is
> > typically much lower than what our network is set for to obtain
> maximum
> > output. We really do not provide an impedance match we just provide an
> > optimum load for the tube to work into.
> >
> > <...>
> >
> > 73
> > Gary  K4FMX
> >
> >