Hello Stefan,
I wondered if peak voltages or peak currents from your parasitic capacitances
and parasitic inductances (lab load and/or earth load) could be contributing to
the current limit, via:
[(a) transistor heating from I/V peaks outside Safe Operating Area (SOA); or
(b) combination of heating and spectral losses from harmonics caused by
parasitic-induced peak currents/voltages outside the transistor linear region].
I have such effects in my VLF large-loop transmitters [but I usually operate
Class D, which is more sensitive to parasitic (out-of-phase, out-of-SOA) peaks].
If the above applies, one solution might be to characterize the transformer and
load with a swept-sine (or impulse) signal, and then compensate with lumped L's
and C's; but I have not done that with large VLF earth-loops or with large VLF
buried copper-wire loops because I suspect that some of the parasitic L/C
effects that I have seen with oscilloscope connected to the PA and matching
network in those cases might be from distributed capacitances and 3-D
distributed inductances in the earth; and compensation might therefore be
difficult.
Whether or not the above applies, I wondered if swamping parasitic-L/C-induced
I/V peaks at the PA output with a lower-impedance PA (and perhaps some parallel
resistance) might keep the PA transistors in their SOA, and enable 90%
efficient operation closer to your goal of ~ 50W PA power. Using a 500W linear
PA to provide 50W (at 90% efficiency) would be expensive, and inexpensive Class
D is more susceptible to I/V peaks, but I thought I would mention this
possibility just in case.
Also guessing that you had previously rejected the concept of lower-impedance
PA-output because SWAP is prohibitive for your remote application; so I'm
hoping that you or another can identify a limiting element that can be
compensated.
At the moment I am at $3/watt to address my similar-sounding issue using
phase-angle-robust 5-milliohm PAs, but this is beyond my hobby budget for the
larger earth loops and buried loops, so I've been settling for lower power into
the larger VLF loops. The $3/watt phase-angle-robust PAs can be paralleled,
which will eventually be helpful if the problem can be reduced to average phase
angles (at the PA output) less than 30 degrees, beyond which the VA to watt
ratio at $3/VA would be hard to justify.
I think I will characterize a large buried VLF loop anyway; prior oscilloscope
measurements with large buried VLF loops seemed to suggest a relatively simple
L/C-earth-network model; it would be nice to know if this is the case, even if
it doesn’t provide an immediate solution for PA SWAP problems with VLF earth
loops. Perhaps such a model will show that earth-distributed L and C are a
limiting factor at VLF; in which case earth-distributed-C might be less of a
problem at 970 Hz. If distributed reactance is primarily from L or C and not
both, it could perhaps be compensated at the PA. Perhaps your 970 Hz test will
provide a clue.
Best compliments on your latest series of fascinating experiments.
73,
Jim AA5BW
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of DK7FC
Sent: Wednesday, August 1, 2018 1:00 PM
To: [email protected]
Subject: VLF:Ground loop antennas, how to tune best?
Hi all,
Today i played a bit with my linear VLF/ULF PA, some output transformers, a 50
Ohm dummy load and a 3 mH air cored coil (a smal one, from LF). I thought this
could represent the ground loop antenna during tests in the shack.
With the switchable C-network, i managed to tune to resonance between
8.27 kHz down to 0.97 kHz. A usual R-L-C network. The goal was to proove that
the system can tune to and run 1 A antenna current on each of the desired
frequencies.
However, i did not manage to reach more than 500 mA 'antenna' current.
Above, there appeared significant distortions / harmonics. Why?
When i connect the dummy load directly to the output transformer (just R
instead of RLC), i can easily tune to 1 A 'antenna' current and the sine wave
almost looks perfectly. So it cannot be an issue of a saturating transformer.
Also the coil cannot saturate.
So, where are the distortions coming from?
The main component of the overlayed distortion voltage is maybe 10x higher in
frequency.
BTW in the first test with the 900m ground loop the voltage showed some
distortions as well and i wondered how it comes from. Then i thought it has to
do with saturation and a far-from-ideal matching of the PA output impedance...
Back to the test in the shack:
Eventually there are further resonance frequencies, coming from the stray
inductance of the transformer and the 'internal' capacity of the coil.
Now if this can happen with discrete elements in the shack, it can happen on
the ground loop anyway! This antenna will certainly have an interesting
frequency response, TX-wise.
So my thought is: Maybe it is even better not to series resonate the loop but
to parallel resonate it! This will lead to a higher feed point impedance, which
will be frequency dependent, so it is a more complex thing. But the parallel C
should help to avoid transmitting on harmonics.
Or, i could series resonate the antenna on the frequency of interest and then
add a parallel resonated LC circuit. This is easier to do because the parallel
LC can be tuned before connecting the antenna. Then the antenna can be
connected...
Or, even something like a T-filter (2 series L and one parallel C) but that
will be complex to align when sitting in the car which is full of electronis
stuff anyway... Also it is a challenge on ULF :-)
And certainly it helps to minimise the stray reactances, e.g. by using just as
less primary turns on the output transformer as possible. But then it is not
usable on all frequencies and will need taps on the primary and secondary side.
Any ideas / comments? :-)
73, Stefan
|
