ARRL 160-Meter Contest - 2019
Call: K1LT
Operator(s): K1LT
Station: K1LT
Class: Single Op HP
QTH: Ohio EM89ps
Operating Time (hrs): 23
Radios: SO2R
Summary:
Total: QSOs = 1289 Sections = 81 Countries = 39 Total Score = 348,960
Club: Mad River Radio Club
Comments:
My goal for this issue of the ARRL 160 contest was to debut the new
phased array receiving system that has been in development here for
the last few years. To accomplish this goal, I had to integrate the
new system with the old system and the rest of my station.
First, some background. The old phased array system is fed by 4 pairs
of short verticals configured as an end-fire array with a relay to
switch directions. The 4 end-fire arrays are configured as a
broadside array using software beam forming technology. The software
runs on a PC using a pro sound card to supply the analog-to-digital
converters. Because of the relays, the array can receive from either
the east or the west but not both at the same time. The ADCs are
driven by a set of Softrock-type direct-conversion receivers.
To solve the east versus west problem, I built a second phased array
(only 6 elements) and added 3 more SDRs and another pro-audio sound
card. The PC hosting the software now runs 2 copies of all of the
software and the first receiver GUI instance repeats the tuning and
filter selection to the second receiver. An outboard audio switch
hanging from an Arduino handles combining the audio from 3 receivers
to feed into the station audio. While it all works, there is no room
for expansion and the pro-audio sound cards require PCI which is
becoming rare as time goes on.
The new phased array system dispenses with the relays thus requiring
twice as much feed line and twice as many receivers. The prototype so
far has only 4 receivers and I built a new array of short verticals
aligned north and south, to provide a test platform in a useful
direction without disrupting the existing system. The receivers are
my own design: a fully double balanced Tayloe mixer feeding a TI
PCM4220 pro-audio ADC chip that delivers the claimed 123 db of dynamic
range. The ADCs feed into a Xilinx Zynq-7000 "system on chip" type
single board computer that has a built-in FPGA (programmable hardware)
and the beam forming is done in hardware. Both the old system and the
new system use the same locally modified SDR software that others
wrote. Putting the beam-steering algorithm in hardware removes the
limitation on the maximum number of receivers.
The plan is to locate the receivers outdoors in the middle of the
array to reduce feed-line costs. The computer would be "headless"
(no
keyboard or monitor) and the control and audio would be delivered by
Ethernet. However it might be cheaper to buy a lot of RG6 rather than
trying to keep the weather off of $1000 of electronics.
A longer term plan is to have the software listen in all directions
and present the operator with the strongest signal, eliminating the
need to hunt through all of the receivers and antennas to find a
signal. This reason explains the elimination of the relays.
The new antenna and receiver system has been running on a pair of
short verticals since last winter. I have been monitoring the system
to watch for instability and also track the behavior of some of the
local noise sources. One problem with the Zynq computer is that the
video also uses the FPGA and seems to be somewhat unstable. Over the
summer I upgraded the operating system (Linux) and the stability
improved greatly, although the video would still freeze after a few
days to a few weeks.
When I built the north/south array, I cannibalized the 2 short
verticals that I had been using for my spotting radio. These 2
verticals are aligned to point a null at my transmitting antenna. An
in-the-shack control allows tweaking the position of the null to
maximize its depth. This antenna permits receiving within a few
kilohertz of my transmit frequency while transmitting full power.
While testing the 4-element broadside / end-fire array, I observed
that a symmetrical "anti-broadside" pattern (the plot looks like a
figure 8) reduced local noise much more than the normal BS/EF pattern
and 2 elements does just as well in that pattern as 4. So I stole 2
elements from the new phased-array system to use for the spotting
radio. The figure-8 north-south pattern would still fill in the weak
spots of the other 2 arrays and I could leave it on continuously since
the noise was very low.
The week of Thanksgiving I took my usual vacation and made another
effort to track down the video problems. After 3 days I determined I
could not find the problem and a headless computer doesn't need video
anyway. I began another round of software changes to allow the user
interface on the old system to control the receiver GUI in the new
system, so the 2 systems would act as one. Despite family and school
pressures, I finished all of the integration issues several hours
before the start of the contest. The new phased array box sat across
the room being remotely controlled by the existing user interface.
The first hour of the contest sounded like great conditions. Signals
were loud, noise was very low and Europe called almost immediately. I
could receive from northeast, northwest and south simultaneously so all
of those hard to hear TN, GA, AL and WCF stations were easier to copy.
However, I had neglected one issue: there was no mechanism to mute the
new receiver while transmitting. That's fine, I used the original
phased array system that way for several seasons before implementing
an automatic muting scheme.
After an hour, I noticed that strong signals from due south had 2
tones. The new phase array system had crashed: the video froze but
the audio kept going without frequency synchronization. So while
running stations at 2 per minute, I rebooted the new stuff and reset
the software. I should have written a script. After several more
crashes after decreasing useful periods, I left it off. Somehow the
video freezing problem was made worse by the network traffic to
deliver remote commands.
The rest of Friday evening played out very similarly to 2018. That
year, I stayed up all night but this year I just couldn't. I took a
3-hour nap at 0830z. After a couple of hours in the morning (no JA or
V6 heard) I had about 50 QSOs fewer than 2018. I'll just work those
guys Saturday evening. Ha!
After another short nap Saturday afternoon I figured that the headless
box wasn't supposed to need video so I changed the software to turn it
off. After that, it ran non-stop without issue, except one which I
will discuss below.
Late Saturday afternoon I had parent duty at the kid's school, which
kept me away from the radio before sunset. When I got back at 2130z,
the band was already fairly full. The evening started nicely, perhaps
because I was late, but it ran slower than last year.
After another hour or so, the two-tone problem came back. I am
guessing that without video to give the FPGA some extra work to do,
the Zynq runs a bit cooler, and the clock chip drifts a few Hz. Since
this clock chip is currently the sync for the receiver local
oscillator, the receive frequency drifts. The audio from the new
receiver is not on the same frequency as the audio from the old
receiver. The different tone isn't the problem, but the 3 Hz beat
frquency made CW hard to copy. So again I turned off the new
receiver.
The surprise multiplier was OX3XR. However no one spotted OX3XR this
weekend and they did spot OH3XR. I suspect my surprise at being
called by juicy DX with an instantly recognizable call got the better
of me. The log checkers will remove my busted QSO and unearned
multiplier.
Around midnight the Alpha made an electrolytic capacitor venting noise
and placed itself in standby without activating the fault LED. After
some sniffing (no smells) and futzing around I power cycled the
amplifier and it seemed to work OK. Maybe the noise was an arc.
Don't know why I'd get an arc. Maybe the dust has returned in just 1
year. The fault circuit seems to have more fault than it is supposed
to. The amplifier no longer faults on overdrive, so I have much less
protection than I thought. Time to troubleshoot Alpha problem #5.
The second night got very slow, even before Eu sunrise. By 0845z I
was ready for a 3-hour nap. After the 3-hour nap, I had really severe
abdominal cramps. I even considered calling 911. Every time I got up
from laying down, the cramps came back. I tried 3 times to place butt
in chair and on the third attempt I turned some stuff off and gave up.
That was very disappointing. Fortunately, the cramps stayed away
after a couple more hours laying down.
DX worked: C6, CM, CT, D4, DL (14), EA (6), EU, F (7), FS, G (7), GM,
GW, HA (4), HB, HI, I (3), LU, LY (2), LZ, OH (5), OK (6), OM (4), ON
(4), OX, PA (3), PJ2 (1), S5 (2), SM (7), SP, SV, TF, TI, UA (2), UR
(9), V3, XE, YL (2), YU, ZF for a total of 39 entities and 110 5 point
QSOs, way down from last year.
I missed VE5 and NT. I worked 3 KL7s and 4 KH6s. I always miss a
couple of Caribbeans / South Americans that everyone else works. This
year it was HC and YV. There were probably others.
It was very pleasing to work D4C and a couple of new European
multipliers using the spotting antenna on the 2nd radio.
Equipment: K3S, P3, Alpha 8410 loafing at 1000 watts. K3 and P3
second receiver. 61-foot (formerly 66-foot) "Tee" top vertical over
75 or fewer radials. 2x4 BS-EF phased array and receiver contraption
and another 2x3 BS-EF contraption. 2x2 BS-EF phased array and new and
improved receiver contraption. 6 2-wire Beverages ranging from 500 to
900 feet, some of which even work but none were used. I use the 150
degree Beverage to watch for signals on the P3 from the south that the
phased arrays often miss.
So despite the disappointing finale, I hope the trial run of the new
hardware and software was useful. I am going to try to get the main
Eu array attached to the new hardware in time for the Stew Perry test.
Posted using 3830 Score Submittal Forms at: http://www.3830scores.com/
______________________________________________
3830 mailing list
3830@contesting.com
http://lists.contesting.com/mailman/listinfo/3830
|