At 09:33 AM 2000-02-14 -0800, Steve Thompson wrote:
>In my experience, the formula holds at 0.5-2dB gain compression, but only if
>the amplifier if truely linear up to when it starts to compress. If the
>formula is mathematically derived, it has to assume linearity up to V/2,
>then look at compression at full output. It can't differentiaite between
>hard and soft compression either (although the effect in practice wasn't
>that great in my tests).
It is not commonly realized that "soft" compression is due to harmonic
energy. Power meters cannot differentiate between the fundamental and
harmonic power so the harmonic energy contributes to the total power
measured. In addition, grounded-grid amplifiers feed through driver energy
which is also measured.
The only way to measure fundamental power without harmonic contribution is
with a tuned receiver or spectrum analyzer.
>Many power mosfets have gain which drops
>continuously with power (Po/Pin is banana shaped)
If this is being observed with an untuned power meter, it is likely because
MOSFETs and FETs readily generate copius amounts of odd-order harmonic
energy, most notably 3rd and 5th harmonics. Helge Granberg wrote about this
characteristic extensively in the earlier Motorola RF Power Transistor data
book application notes.
>.. latest LDMOS has an 'S' shaped Po/Pi characteristic at most bias levels.
Likely the same problem: untuned power meter and harmonics being measured.
It's also not commonly realized that spectrum analyzer measurements are
notoriously inaccurate at amplitude measurement; even the latest Hewlett
Packard instruments with digital readout of the cursor's location are
specified to no better than about +/- 1 dB accuracy. This inaccuracy can be
reduced considerably by comparing the power measured by the analyzer to a
known, clean signal source level established by a power meter.
>Most bipolar amps are
>dreadful because the particular transistor isn't suited to linear operation
>and/or the biassing arrangements are rubbish.
Bipolar RF power transistors are built with and without emitter resistance
which introduces negative feedback. The emitter resistance helps raise the
real part of the base resistance for easier matching. Motorola called these
devices "Controlled Q". A very-low-Q, low-pass network is usually used to
connect the base junction to the base lead of the package and raises the
base impedance further (in V/UHF transistors, this network is often the
reason why the high frequency response drops drastically beyond the
intended operational frequency range).
>Tubes aren't exempt - poor bias and/or anode voltage regulation, or wrong
>loading can mess things up a treat.
Tetrodes driven into drawing grid current will cause the grid bias to shift
further negative if the bias supply does not have sufficiently-low source
impedance, both positive AND negative. Take the common 4CX250B, for
example. The typical grid bias regulator is a zener with a shunt 10K
potentiometer, the bias taken from the pot wiper. These tubes typically
require a grid bias of between -50 and -80 volts, and the zener voltage
most-commonly used is 100 or 105 volts. Thus, there is a differential
voltage of 20 to 50 volts between the control grid and the zener diode
itself. A 10K ohm potentiometer used to adjust the bias voltage results in
a resistance of between 2000 and 5000 ohms in series with the grid to the
zener-regulated source. If the grid current is just 1 milliampere, Ohm's
Law shows that the voltage developed between the grid and zener will be
between 2 and 5 volts, a barely-acceptable variation. If the grid current
is 5 milliamperes, the voltage difference between the grid and zener will
rise to between 10 and 50 volts, clearly a very undesirable state of
affairs. A shunt regulator is necessary to prevent the bias from being
more-negative.
>Where amplifiers show low IP3 compared with 1dB compression, there's often
>something to be seen in the gain/power graph, or there's more than one
>factor contributing to the compression (e.g. multiple gain stages
>compressing at the same time, or device saturation and bias failure both
>occuring together).
... as noted above.
73, Steve K0XP
--
FAQ on WWW: http://www.contesting.com/ampsfaq.html
Submissions: amps@contesting.com
Administrative requests: amps-REQUEST@contesting.com
Problems: owner-amps@contesting.com
Search: http://www.contesting.com/km9p/search.htm
|