New H rx under test
http://abelian.org/vlf/tmp/171216a.jpg
with the box on the left injecting a calibrated current in
parallel with the dummy antenna. This is an end-to-end test:
antenna to sound card, with 150m cat5 and the isolating
transformers.
Most of the testing is automatic: the VLF calibrator works
through a time/frequency schedule and a vlfrx-tools script
does the measurements to the same schedule.
System noise measurement is a bit trickier. The output noise
of the rx is only a little above the sound card's noise floor
(in fact is lower at higher frequencies). I allow for this
by subtracting the sound card noise power but to make it more
accurate I set a 10dB higher rx gain than when deployed.
That higher gain can't be used in the field because it would
overload on the 50Hz and strong MSK signals. This illustrates
the tough dynamic range requirement of a VLF rx, even when
using a high quality sound card such as the 24 bit M-Audio 192.
Intermod measurement is another tricky one. The sound card
generates a 2-tone test at some high frequency, through a
passive high pass filter into the rx. Rx output through
a passive low pass back to the sound card input to measure
the level of the difference frequency. Often a receiver
will look good on this test but do badly on the antenna.
The culprit is usually the strong 50 Hz signal distorting
very slightly as it goes through the isolating transformers
and modulating the whole band as it does so. The cure is
to tailor the rx HP response so that the 50Hz is low enough
not to cause noticeable intermod. What exactly 'noticeable'
means depends on whether you're listening to whistlers and
tweeks, or trying to dig a weak carrier out of the noise.
In a bad case you can hear the intermod - the background noise
has a sizzling buzz to it instead of a pure hiss. A not so
bad case will not be audible but will raise the noise floor a
few dB. The whole system: rx, transformers, and sound card,
have to be impeccably linear to actually achieve the noise
floor capability of the front-end device.
The final thing to do is set the gain, which is fixed by
components. I try to get within a factor of two either way
by bench testing. When it's connected to the antenna I'll do
some measurements then will probably have to bring it back
in for final adjustment to the gain. As it stands, the 2
channels are matched to 0.52% at 8.3 kHz.
Hopefully next week it will go outside for a real test.
Only one pair of loops and feeder cable so can't run the old
and new H rxs for a side-by-side comparison. I'll be making
good use of some regular distant amateur signals.
--
Paul Nicholson
--
|
