>On 12/14/2013 6:09 AM, Jim Lux wrote:
>> I'm not sure this is true in general. What is the mechanism for the
>> increase in distortion?
>
>I'm quoting my neighbor, K6XX, an engineer intimately involved with the
>design of Elecraft amps. I don't know the mechanism, but my guess would
>be a non-optimum operating point for the output devices.
>
>73, Jim K9YC
That's very close - the cause of the distortion is a non-optimum load
line.
Most of us are familiar with the idea of drawing a 'load line' on the
characteristic Va/Ia curves for a tube amplifier. Exactly the same
concepts can be applied to MOSFET and bipolar devices but we don't see
SSPAs discussed in those terms so often, so I'll mostly describe this in
the more familiar language of tubes.
The zero-signal end of the load line (near the lower right corner of the
chart) is fixed by the DC bias conditions. Va is the voltage of the
power supply, and in a class AB...B amplifier the zero-signal anode
current Ia will be set to a relatively small 'idling' value. When drive
is applied, the operating point moves along the load line towards higher
anode currents and a lower instantaneous anode voltages. The *slope* of
the load line depends on the load impedance that is presented to the
tube, while the *length* of the load line depends on how hard the
instantaneous operating point is driven along that line.
A typical set of characteristic curves will be quite straight and
parallel over most of the chart, and this is where a linear amplifier
should aim to operate. But at the upper left of the chart, the curves
curl sharply upward. This simply means that when the tube is driven
harder, it becomes increasingly difficult to drive the anode voltage
down closer to zero. Exactly the same happens in solid-state devices
where the drain or collector voltage is said to be 'bottoming out'. This
is the area you need to avoid, for whenever the operating point enters
this upper left region, there will be a sharp reduction in the
instantaneous power gain which is the main cause of 'flat-topping',
non-linearity and IMD.
When an amplifier has been correctly designed, and is correctly loaded
(read: sees a 50-ohm resistive load at its output port) and is not being
overdriven, the upper left end of the load line will always be well
clear of that non-linear region.
Now what happens if that same amplifier sees a load with a higher VSWR?
The correct answer is: "You don't know and neither do I!" because any
given value of VSWR can hide an infinite variety of resistive/reactive
load impedances. Always remember: what matters to the tube or transistor
is the actual load IMPEDANCE that is presented to its
anode/drain/collector.
Recall that the load line is 'pinned' at its lower right-hand end by the
DC bias conditions, so varying the load impedance will cause the whole
load line to pivot about that fixed point. If the load line happens to
rotate anti-clockwise, you're in luck because that will swing the
left-hand end farther away from the region of high IMD. This is exactly
what we're doing in the final step of tuning and loading a tube
amplifier, because we've all been told to "Increase loading a little to
improve the linearity." We are deliberately rotating the load line
anti-clockwise, more safely into the linear region.
But depending on the VSWR and the feeder length, your luck could go the
opposite way. If a high VSWR happens to rotate the load line clockwise,
the left-hand end will move upwards into that distortion region. And
for a number of reasons your luck is MUCH more likely to go that way!
One reason is that the characteristic curves bunch together very tightly
in the distortion region, so if you stray towards that area the IMD
levels will increase very quickly. The other reason is that a 'load
line' is actually a special case that only applies when the load
impedance is purely resistive. In reality 'a high VSWR' is MUCH more
likely to represent a reactive load impedance, and then the phase
difference between voltage and current will cause the load line to open
out into a long elliptical loop. That open loop shape makes it much more
likely the instantaneous operating point will pass through the
distortion region at *some* part of the drive cycle.
So an increased load VSWR could either improve the IMD or make it worse;
but the odds are heavily weighted towards the bad. Even for amateurs,
"Are you feeling lucky today?" is no way to do RF engineering!
[ Sorry to stray so far into AMPS territory, but the source of the
problem is almost sure to be either up the tower or along the feedline.
]
73 from Ian GM3SEK
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|