I recently fired up an old Pentium III motherboard to support an ISA GPIB controller.
The fastest ISA board I have is 800 MHz. (If you've priced a PCI bus GPIB controller,
you will understand my interest in an ISA bus motherboard.)

I downloaded the WSPR source code and compiled it under Fedora 16.
It actually ran. However, the CPU utilization was rather high.
One time I observed WSJT skipping a decode because it had to make a
transmission.

I was hoping I could run WSPR on the old box but it looks like GPIB
will be its function.

This afternoon's transmissions at 14.097022 are going out at
one half milliwatt or 500 microwatts.

I set transmit power to 1 watt on my Bendix 263 and 30 db
attenuation on my R-9000. I then switched out the attenuators
and dropped audio gain to compensate. Finally I dropped audio
further, another 3 db or so.

I use CAT (ICOM 756 Pro ci-v) for PTT and QSY.

The rigctl sub program fails to run
complaining about a missing libusb0.dll

Ham radio Deluxe 5.?? does not seem to need this
DLL to talk to the radio.

uninstalling 2.10 and reinstalling the old version
gets WSPR back on the air.

Rawhide is the development branch of Fedora on which new
ideas are tried out before general release. At the moment
Rawhide is working with the Nvidia driver so I can run
Compiz and all that good stuff.

I used SVN to get the current WSPR code. After installing
required packages, WSPR compiled. I had to change permissions
on ttyS0 to allow regular users to use it with WSPR.

Only bug so far is a recurring one - the rig selection menu
does not display anything. I had to note the rig number from
the Windows version and type that number.

Propagation was very good last Tuesday morning.
I was able to hit Antarctica with about one milliwatt using my Thunderbird 5 mk 2 at 60 feet.


Timestamp Call MHz SNR Drift Grid Pwr Reporter RGrid km az
2010-08-31 14:06 WA7KGX 14.097121 -19 0 CN85no 0.001 VP8DMH FC52wk 13390 159
2010-08-31 14:08 WA7KGX 14.097122 -20 0 CN85no 0.001 VP8DMH FC52wk 13390 159
2010-08-31 14:16 WA7KGX 14.097140 -23 0 CN85no 0.001 VP8DMH FC52wk 13390 159
2010-08-31 14:18 WA7KGX 14.097140 -24 0 CN85no 0.001 VP8DMH FC52wk 13390 159
2010-08-31 15:02 WA7KGX 14.097007 -28 0 CN85no 0.001 VP8DMH FC52wk 13390 159


Other than this, I have not been able to get past the East Coast with 0 dbm.

Up to an hour or so ago, reported power on transmissions
of 5 watts or less were understated by 2-3 db. This was caused
by my MFJ tuner reading 2-3 db low in the one watt range.

While MFJ does not spec the accuracy of the wattmeter in the
993B, it is inelegant to show two significant digits when
the measurement accuracy is less than one significant digit.

I decided to try something fancier than the Minimalist
sound card interface I described previously.

For the 2.0 version I bought two Radio Shack 600 ohm
isolation transformers. These sell for about $4 and
come with leads. I kept the 10-turn trimmer I used
with the previous interface and added a forward facing
pot to the circuit. Now the transmit audio from the sound
card goes to one side of an isolation transformer. The other
side goes to the trimmer pot. The arm of the trimmer goes to
the front panel pot. Set the sound card output to full
as well as the front facing pot. Adjust the trimmer to get
max output from the radio with little or no ALC reading.
Between the radio's RF OUTPUT control and the front pot
it is reasonably easy to set the RF output anywhere from
100 watts to one milliwatt.

It may be necessary to add a parallel combination of a
10k resistor and several mfd cap between the sound card
output and the isolation transformer if the sound card
will not output to the transformer directly.

The receive audio goes through the other transformer
to the audio card's input.

Keep the grounds separate to avoid ground loops.

I picked up a pair of fancy 1/8 inch mono Radio Shack
phone plugs for the latest iteration of my rig interface.
Everything worked normally until Thursday afternoon
when the stations stopped receiving my WSPR signals.
The fishfinder also showed no output. Adjusting audio
and RF gain for full output yielded nothing. But the
radio's microphone got out normally.

After a few more checks I located the problem. The
connection between the solder lug and the tip of the
Radio Shack connector was loose. The factory riveting
wad defective. A hard squeeze with pliers helped things,
but I will be looking for another connector source.

I changed from DTR to CAT for PTT and gained rig control in the bargain.
With the 756pro antenna on Auto, the antenna switch remembers the setting
for each band. Nice.

Using a dedicated sound card for both send and receive seems to make WSPR
stable under Ubuntu.

I had to download, configure, and compile the current hamlib to get the
756pro to play under WSPR. With Fedora, the list of rigs on station setup
comes up blank. Entering the rig number manually makes it happy.

Under Ubuntu, WSPR would not save and restore the 756pro selection.
I suspect entering the rig number manually would solve that.

A little while ago I got tired of XP wasting more than half
my office computer's memory. I coughed up $125 to upgrade
to Windows 7 64 bit premium. If there justice, the cost of
W7 should be subsidized by hardware makers. Let me count the
ways W7 forced hardware upgrades:

Video board - The current video drivers for a Nvidia 9600gt
caused the status messages for WSPR to disappear. Upgrade
to ATI.

Sound board - I had been using a SoundBlaster Live! with WSPR.
No support. I got a $13 board that worked for a while, sort of.
It quit working when I added isolation transformers to the radio interface.
Now I am using a $90 Xonar DX board which is reputed to have W7 drivers.

Oh, the motherboard clock does not work well W7. Loses time like an
old wind-up wristwatch. The W7 sync doesn't work well enough for WSPR.
Windows 7 protection bollixes Dimension 4.

And there are several other perfectly good devices that don't work with W7.
People bitch about Linux not supporting devices! Offhand I'd say W7 is worse.

I've been in Ham Radio since the 1960s.
I was active on RTTY in the 1970s with a
microprocessor based bit banging terminal
using modified Sidereal Muicronet II hardware.

The unique tools provided by the WSPR system
have awakened my interest in HF propagation
and antenna issues. I finally put up a 130
foot Van Gordon dipole I bought some ten years ago,
and put my Dentron SuperTuner back together.

Now I understand why hitting Europe from the
Oregon Rainforest is so tricky - it's the
auroral oval that is in the way.

Connect a short ground strap between the transceiver and computer. Connect radio's aux audio output to computer's line in, line out to radio's aux audio in. Use a pot at the radio end of the cable to adjust audio level for transmit. (Accessory plug in many ICOM radios). Connect 220k resistor from serial port RTS to base of NPN transistor. Add a .01 or larger cap between base and emitter. Emitter to ground. Collector to PushToTalk. Adjust audio levels to suit. Voila. The sound card audio output level should be as high as possible to allow the pot to reduce noise as well as control transmit level.

I added a used Sound Blaster Live Value card to the computer so I could run WSPR concurrently with normal ops.

I decided to participate in the 160m test.
I strung up a 135 foot doublet which I bought
15 years ago but had never opened. Right now
the ends are up about 30 feet and the center
is about 10 feet. The ladder line comes in
the window sill wrapped in a plastic shopping bag.
Most of the 100 feet lies on the floor in a
scramble.

I tried matching it on 160 with a Nye Viking
PI network matchbox but it wouldn't tune below
about 3 mHz. So I put my Dentron Super Tuner
back together and that did load up on 160.
The Dentron was working hard to match the doublet,
with the matching coil becoming warm after 110 seconds of
100 watt RF. I could not detect any warming of the
balun coil.

This setup seems to get out on 160 about the same
as some of the other stations adjusting for power.
Unfortunately Microsoft rebooted the computer after
Patch Tuesday, so WA7KGX missed the wee hours of
the morning when conditions on 160 might have been
the best.