[Please note:  This material was published in the May-June 2005 issue of 
The Canadian Amateur]

Transatlantic Reception of QRP 1600 metre signal

(Believed to be the first transatlantic reception of a signal 
in the 160-190 kHz band and the first QRP transatlantic LF 
signal) 

Joe Craig VO1NA, Alan Melia G3NYK and Hartmut Wolff

In October, following discussions with Alan G3NYK, a renewal of the 
letter of authorisation originally granted in 1992 to VO1NA was received
from Industry Canada.  Preliminary experiments with the aerial used
on 2200 metres showed very promising results.  Alan and I decided 
to try a transatlantic attempt on 1600 metres and Alan advised me that 
detection of the signals may be possible, despite the very high QRM 
from the multi-megawatt commercial stations in Europe that all but 
completely covered the 160-190 kHz band. 

A severe ice and wind storm destroyed the 25 metre tower which used to support 
the LF aerial on 24 January, but the 100 metre wire was raised again
to an average height of about 10 metres in an inverted L configuration.
Alan described this aerial most aptly as a 10 metre vertical with 100 metres
of top loading wire. Most of the ground current flows to the AC mains
grid earth connection. About 10 percent goes to the station ground. 

The transmitter was a class E designed using Alan's spreadsheet of Sokal's
equations and had an output of 5 watts forward and about 0.5 watts reverse
(2:1 SWR) with no serious effort being made to effect a resistive match. The 
antenna was tuned for minimum SWR with a 0.35-1.38 mH variometer.  About 
0.6 mH was required to tune the wire and the aerial current was 0.4 amps. The 
ground loss was estimated as R=P/I^2=30 Ohms. The 189.81 kHz 
carrier was generated by dividing a CMOS 4001 crystal oscillator
by 16 using a 4020 whose output was directly connected to the gate 
of the IRF640 final with a Vdd of 13.8 volts.  Because the signals 
appeared quite strong even at 25 km distance, reports were 
solicited on the Long Wave Club of America reflector with the expectation 
that perhaps something might be detected off the island of Newfoundland.  
Traces of reception were received from Warren K2ORS on 21 February
and copies were reported by Jay W1VD and John W1TAG on the 22nd using
ARGO Fourier transform software developed by I2PHD.  
On the morning of 22 Feb. a very clear copy was received by John.
A distinctive drift of the carrier was evident in the spectra, no doubt 
due to the crude oscillator design.  At the time the transmitter was keyed 
at QRSS120 with a computer to send strings of "NA". 

On the afternoon of the 22nd, a 555, 4017 and diodes were breadboarded to 
send a 10 bit message conisting of an "N" and a space of 5 bits at about 
10 WPM. A 4700 uF capacitor was put across the 4.7 uF timing capacitor to 
reduce the speed to about 0.01 WPM corresponding to QRSS120.

The transmissions were not announced in Europe over the Radio Society of Great 
Britain reflector. It was planned to do this when a relatively clear frequency 
was found by Alan, the carrier drift could be rectified and a serious effort at
crossing the pond on 1600m could begin.  It was intended to follow Alan's 
advice to use a 1 Hz FSK with a period of several minutes.  What happened on 
the morning of the 23rd February therefore caught us completely by surprise.  
Hartmut Wolff in Germany 4400 km away sent a spectrum of signals received from 
0200-0300 showing what appeared to be two "N"s and a report of a partial copy 
of the beacon.  He was not aware that the CMOS keyer had replaced the computer 
ID. The timing of the N's was consistent with the 10 bit message and I was 
pleased to confirm to Hartmut the signals were mine, almost beyond any doubt. 

Hartmut was using a JRC NRD-525 receiver with a 250 Hz filter and a 30 dB 
preamp. Despite the proximity of the Iceland AM broadcast station  
on 189kHz, the receiver was running with AGC off and RF Gain on maximum.
The receive aerial was a double wire K9AY loop.

Hartmut provided the settings of the DL4YHF Spectrum Lab software he used to 
copy the signals. He was using FFT with a Hanning window centred on  
709.68 Hz with a 4Hz range around the center frequency.  The FFT bin size 
was 0.0053 Hz and the sampling frequency was 11025 Hz.  The waterfall scroll 
interval was 30,000 ms.  From the latter, the period of the ID cycle was 
calculated and determined to be within 5 percent of the actual measured 
cycle period, confirming the reception. 

Conditions at the time were favourable as reported by Alan.
"Interestingly the Dst was low last night moving into occasional positive
figures for the first time in many days. Steve's (W3EEE) plot of DCF39 (in
Germany on 138.83 kHz) showed reasonable but not outstanding received levels."

All those who provided signal reports and the collective advice and expertise 
of the LF experimenters world-wide are gratefully acknowledged.