From Marcus DF6NM via the RSGB LF reflector earlier today:
"Today between 7:30 and 14:00, Lubos OK2BVG and I successfully completed a two-way contact on 8.97 kHz. We believe that this is the first international QSO ever on VLF.
The distance between Breclav (JN88KS) and Nuernberg (JN59NJ 69es) is approximately 424 km. The weekend before, Lubos and I had found that we could see each other's VLF transmissions from our small home antennas in a slow spectrogram, using about 0.45 millihertz FFT resolution ("DFCW-6000"). This long integration requires about 2000 seconds of continuous non-interrupted carrier to reach full sensitivity.
We decided to use absolute frequency encoding of characters ("MFSK-37" mode), which has a simple structure but is significantly more efficient than two-frequency DFCW. The software signal generators in SpecLab were used to create 30 minute dashes by editing the frequency in 1 mHz steps every half hour. This can be automated by opening a textfile which is read by the "periodic actions" function. Between transmissions, additional half-hour gaps were inserted to let the FFT ring down, and allow for TX-RX reconnections.
Characters are identified by reading the frequencies above 8970.000 Hz, with 8970.000 to .009 assigned to the numbers, .010 idle or space, and .011 to .036 the letters A-Z. Lubos used a Rubidium standard to lock his transmit samplerate, while both receivers and my TX were synchronized by military MSK signals from GQD (22.1 kHz) and DHO (23.4 kHz).
With about 80 watts from an audio amplifier, I could get up to 0.38 A antenna current into my top-loaded vertical 9 m above the roof, radiating on the order of 10 to 15 microwatts. I believe that Lubos is using similar equipment, perhaps a couple of dB stronger. For receive, my soundcard was connected straight to the loading coil and antenna, whereas Lubos has an active probe at a quiet site remote from his TX location.
We wanted to go for a "full QSO" format with reports and confirmations, beyond the rudimentary three-dash "micro-QSO" format which was used in June 2009 between DJ2LF and myself. With single-letter suffixes, we ended up proceeding as follows:
07:30-08:30 "NB" ;Lubos' call: df6Nm de ok2Bvg
09:00-10:30 "BNM" ;my reply and report: Bvg Nm M
11:00-12:00 "RO" ;his confirmation and report
12:30-14:00 "RTU" ;my confirmation and thank you
Although today the noise was not as low as it had been before, we managed to exchange the essential information in 5.5 hours until 13 UT, after which QRN from lightning in southeast Europe became strong enough to obliterate further copy.
Attached image contains captures from Lubos' grabber at Apollons temple (top) and my receiver (bottom), with screenshots from the DK7FC and OE3GHB grabbers pasted inbetween. The original captures are at http://df6nm.bplaced.net/VLF/ok2bvg_df6nm_120325/. All spectrograms were stretched to the same timescale (5 min/pixel). On the left hand side, you can see a MFSK pre-test from Lubos, sending his call to hs own grabber at very low power. There was also a long dash from OE3GHB on 8970.030, and a carrier from DJ8WX on 8970.022 is visible at DK7FC.
The rightmost part of the captures shows the essential eight dashes from our QSO. You can see that I had almost lost Lubos' "B" when my noise blanker triggered heavily on local QRM, inducing me to give an "M" rather than "O" report. But hovering the cursor to 8:15 UT does show the peak at 8970.012 Hz in the spectrum graph. My final "R" is just barely visible at .028, whereas the following "TU" at .030 and .031 ended up drowning in the increasing noise.
During the lowest noise period (9 - 12 UT), the dashes were visible (though not decodable) in 4.5 mHz "600" spectrograms. At that time, our symbols also appeared on Paul Nicholson's spectrogram http://abelian.org/vlf/fbins.shtml#p=1332712800&b=110&s=sp
Thanks again to Lubos for the effort and patience! Although exchanging half-hour symbols is tedious and may sound boring to some, both of us enjoyed it and actually found it quite exciting.