Alternative topologies, and using higher voltage devices make things
much easier. Common-emitter can be troublesome. We ended up going
common-drain (simple source followers), running 28V, and using a
diff-amp as the voltage gain stage, then resistive feedback to the
input stage. So, the feedback was DC to 100MHz or so. Ended up with
about 150W capability and IMD3 measured -58dBc rel one carrier. 5ths
and higher order were down substantially more.
We had tried simple xfmr feedback, problem was that the 2nd/4th/6th
order products, down near DC, were re-mixing with fundamentals to
create 3/5/7, etc. Weren't able to make xfmrs that operated close
enough to DC to get around that prob.
We worked with the semi mfr as well, trying to get internal resistors.
Turns out the technology used for emitter ballasting is fundamentally
non-linear. Good for balancing DC idle currents, but not very good
for negative feedback. In the MRF150 type devices, no such ballasting
is used, rather they depend on having 32 identical devices adjacent to
each other on a wafer, and bonding in direct parallel. Semiconductor
physics allows you to do things like that, expect adjacent devices to
be identical enough to directly parallel. In the MRF154, which was 4
X MRF150 in a single package, the drains and sources were in direct
parallel, and the only isolation resistors were in the gate paths,
cutting down on some form of cross-coupled VHF oscillation between
devices.
At one point, we feared we'd have an expensive hybrid design facing
us, but coming up with the diff-amp input, two stage buffer and final
output FETs, and getting the customer to buy off on 28V operation, did
the job. Complex circuit when we were done, and way more gain than
you want usually from a single stage, but we got the performance
without a custom hybrid design.
Hard to believe I was once in that industry...this was shore-based
marine SSB, we were supposed to meet a final system IMD of 36dBc rel
one carrier, including all stages...so we made them all super-clean,
and beat the spec by 10dB. Adjacent channel power measured far down
as well, but I don't recall the figure.
Actual torture test was to put band-limited white noise through it,
and see how much energy was in the next adjacent channel. A variation
on that test is still in use commercially, known as the "noise power
ratio" test.
Z
--- In ham_amplifiers@..., FRANCIS CARCIA <carcia@...> wrote:
I suspect you would be better off with a 300 watt amplifier running
closer to class A with transformer feed back, Most RF transistors have
internal emitter resistors to balance the parallel cells. Making
feedback work over a wide frequency range takes real talent and good
pc board layouts. gfz
zerobeat40 <zerobeat40@...> wrote: Hey, I tried to put the
NFB into a solid state amp once, being used as
a driver for commercial HF SSB. It was not so easy.
Assuming a pair of NPN transistors delivering 100 watts at 30MHz,
running from 13.8VDC, producting -30dBc IMD that we want to improve by
about 10dB. The correct value resistor is approx 0.2 ohms and it must
dissipate 15 watts, if the amp is to survive clumsy tuning into an
antenna tuner at full power. You could get away with a 5 watt device
if you insisted on only SSB (no CW or FM) and only into a matched
load. Smallest resistor I was able to find to meet this was a chip
style component, about 1/2 inch X 1/2 inch. It measured 5nH of
inductance. At 30MHz, XL=nearly one ohm. The stage gain at 30MHz was
reduced to approx 2dB, and the phase shift of this inductance reduced
the IMD benefits of the NFB to having no IMD reduction at all. At
1MHz, the solution worked very nicely - stage gain stabilized at 14dB,
and IMD measured about -42dBc (referenced to either of two incident
carriers)
If you could somehow create a 15 watt resistor that is 0.2 ohms and
under approx 0.2nH of inductance, then your proposed solution will work.
Let us know when you find that resistor.
Z
--- In ham_amplifiers@..., R L Measures <r@...> wrote:
tnx, Tony. Since adding RF-NFB to a transistor amplifier is as
simple as adding unbypassed R to the emitter leads, it puzzles me why
Ham transceiver manufacturers don't wake up and start building
pristine radios.
Since the TS-830S uses essentially a copy of the KWM-2's RF
amplifier, it isn't surprising that the 830 has a reputation for
cleanliness.
cheerz
