Beam Steering in Combat
or
How to Use Software Radio and an Array of Short Verticals in a 160 Meter Contest
This past weekend, I used a new secret weapon in the 12th Stew Perry
Topband Distance Challenge Warm-Up Event. This secret weapon consists
of 8 short verticals, 4 Softrock v6 receivers, and a Delta 1010LT
sound card plus computer running my Phasor module, DttSP and
SDR-Shell. The verticals were configured as a 4 element broadside
array of end-fire arrayed verticals. See http://k1lt.com.
Since the phased array must be used with a computer to form an entire
receiving system, I had to integrate this new technology with a
traditional contest station. I did this by routing the audio from the
computer/radio/antenna system to the left channel of my headphones and
the audio from the main station to the right channel. Since I didn't
have time to build a real SO2R (single-op, 2 radio) control box, I
used a Ramsey 25B stereo FM transmitter and a Radio Shack Stereo FM
Headset receiver. This headset has separate level controls for each
ear (for balance) besides an overall volume control (nominally, the
volume control). Thus I could use the audio control on the IC765 for
"traditional" (Beverage antennas and a triple conversion receiver)
audio, and either KDE's kmix audio level pop-up control or the left
earpiece level control on the headset to control "high technology"
(phased array and software radio) audio.
I set up my station a few hours before the first expected signals.
The contest is already on-going at this point, but nobody bothers with
ground-wave contacts in this event. As I putter about the shack, I
can monitor the waterfall display provided by SDR-Shell for signals.
Since my monitor is 1600 pixels wide, I can see about 36 kHz of
spectrum at a time, which is plenty for most 160 meter contests. (It
would not be enough for the ARRL 160 or the CQ WW 160 contests, which
occupy most of the bottom 60-80 kHz of the band.) (SDR-Shell must use
2 pixels per FFT bin.) I guess I should buy a 1920 pixel monitor
before then :-)
The first problem becomes immediately apparent when I call the first
station: the software radio provides the monitor of my transmitter,
since its always receiving, but the audio is delayed by 0.021
(2048/96000) seconds, which is enough to make sending via the paddle
difficult. I thought about quickly hacking the antenna phasing module
to block the audio in the presence of a signal stronger than -40 dBm,
but the thought of hacking software during a contest was a bit too
much like work. Also, blocking the audio in the phasing module would
likely add severe clicks and would disable the waterfall display. The
waterfall is useful in that I can see where I'm transmitting in
relation to the signal I was just receiving.
Several solutions suggest themselves: mute the audio in DttSP using
the push-to-talk signal from the transmitter, or even better, use an
additional pair of ADC channels to sample the transmitted signal and
subtract it from the received signal. Alternately, separate
processing could delay the audio from the traditional radio by 21
milliseconds, although this would not provide for the monitoring of
one's own sending, even though is does cure the weird echoing sound
when listening to the same signal through both ears.
In years past, I've used SDR-Shell and DttSP to monitor the 160 meter
band during a contest. Although there are almost always too many
signals visible at one time to tell if one of them is new, there are a
couple of uses for the waterfall display. First, a pile-up is usually
an indicator of an interesting station, and a pile-up is plainly
visible. Second, holes stand out, which makes finding a new run
frequency much easier than tuning the band. However, my own
transmitter usually paints the entire display white with the
distortion products from overload. Therefore, one has to stop
transmitting for a while to make use of the display.
With the phased array, I noticed that I can steer the array for a nice
compromise between listening in the direction of Europe, and
attenuating my own transmitted signal. Of course, the transmit
antenna is much too close to the array for a proper null. But by
steering the array a mere 6 degrees north of the ideal European
bearing, I could reduce the apparent strength of my own transmitter
from -40 dBm to -48 dBm, which is enough to reduce the extent of the
waterfall "white-washing" to about 25 kHz and allow me to see some
other signals despite my own transmitter.
Thus is born the idea of subtracting the transmitted signal from each
of the array receivers. This trick requires another pair of ADCs
phase (word) locked to the existing 8 ADCs (by using another Delta
1010) and the same software that I'm already trying to use to
"calibrate" the phased array. So this is an area of future
experimentation.
The second problem became apparent when numerous weaker European
stations were calling: not enough (any) knobs! When trying to copy a
barely audible weak CW signal, this operator tweaks the incremental
tuning, the pass-band shape, and the antenna direction to try an find
the best copy. Often, these settings change during the course of a
difficult QSO as propagation changes (fading). So, a mouse or
keyboard based user interface does not make for rapid tweaking. Maybe
we can find a bank of rotary encoders that can be interrogated via a
USB port, and interface that to SDR-Shell. Also, I think I shall
update the phasing module to allow "push-button tuning" to select a
number of preset beam headings, perhaps by mapping the edge keys of
the numeric keypad as if they were a compose rose. I use the same
tactic to select from a set of 12 Beverage antenna direction using a
telephone style keypad.
Also, I might have to add a personal modification to SDR-Shell to
disable the escape key. (The function of the escape key is nominally
completely correct, except for my own silly situation). My station
has two IBM model M keyboards side by side, the left keyboard for the
Linux machine that runs DttSP and the rest, and the right keyboard for
the Windows machine that runs the NA logging program. When the rate
gets slow, I get up and wander around while the CQ machine runs. When
someone finally calls, I rush back to the keyboard and bang the escape
key to stop the transmitter. Several times during the contest, I hit
the wrong escape key, which shuts down a receiver when I most want to
receive! There is nothing like real user tests to find these kinds of
issues :-)
During the first two thirds of the contest, when signals may be
arriving from Europe, I kept the phased array focused on Europe, and
used the Beverages to aim west. That way, I could take calls from
either coast without having to hunt through all the Beverages as I
have in the past. I still need to check the southern Beverages for
calls from Florida and Texas, but that is still better than the old
way. In previous contests after each CQ, I would switch through 4 or
5 Beverages to make sure I didn't miss a weak signal. During the
hours from 0400-0600Z, which have the most European contacts, I kept
both the phased array and the Beverages focused on Europe. Although I
could usually hear a European caller on both antennas, the copy was
nearly always better on the phased array. The phased array seemed to
have a slightly better SNR although its very difficult to make
critical quality judgments while trying to copy a weak signal. Note
that NEC modeling suggests the phased array should be about 3 dB
better than my best European Beverage.
C52C in The Gambia, Africa, was on-the-air during the contest,
although he was not participating. I did note that the phased array
could distinguish him from the Europeans in that his bearing seemed to
be about 20 degrees to the south of the Europeans. Also, I worked
several Scandinavians, and they seemed to be about 10 degrees north of
the typical European bearings. Note that I have the phasing module
set up to change direction in 10 degree steps by default, and in 1
degree steps with extra key presses. Also note that propagation often
smears the apparent bearing of many stations outside of North America,
so take these observations with several grains of salt. But the
results seem promising.
After sunrise in Ireland, I took a nap and came back around 0930Z to
begin listening for Oceania and Japan. However, I was severely
drained of motivation by that point, and also the propagation did not
seem to be there, so I did not gain much insight about the phased
array performance to the west. DX to the west from my location has
always been difficult for several reasons, including geography, local
topography, and personal chronometry (I have a hard time being awake
very early). So, I didn't listen for long, and I may have missed DX
and certainly missed the opportunity to evaluate the phased array to
the west. My quick perception is that the array was noisy.
Finally, I need to record the output of the 4 Softrock v6's, so that I
can analyze this stuff after the contest. For example, I could go
back and figure out why I think I heard ZS0X in grid square JO99, when
ZS is a South African prefix, and JO99 is in the Baltic Sea. (Maybe
the call was 7S0X, although I swear I heard "Z" and not "7".) Or, I
could figure out why TI7H came in from the south, even though he gave
me a stateside grid square. I suspect there are JACK plug-ins that
could do multi-channel recording, and if not, such a capability would
be fairly simple to add to my Phasor module.
The bottom line is that the phased array works, even though I have not
finished optimizing it, nor have I a decent way of calibrating it.
_______________________________________________
Topband mailing list
Topband@contesting.com
http://lists.contesting.com/mailman/listinfo/topband
|