Residual Carriers on the 80 Meter Band
Rex Dalton here. Last month I pulled an old Hammarlund HQ-129 from storage and fired it up on 3.8 MHz just after dusk. The usual static filled the speaker, then a slow pulse train appeared every 47 seconds. It lasted nine minutes before vanishing. I logged the time, frequency, and signal strength. Nothing in the FCC database matched.
The receiver uses 6L7 tubes and a 455 kHz IF strip. Its bandwidth sits wider than modern rigs, so it passes odd sideband energy that digital filters often strip away. That wider window may be why the sequence showed up at all.
The pulses repeated with a 3.2 millisecond offset that grew by one microsecond each cycle.
I checked propagation forecasts for that evening. The MUF sat near 5.1 MHz, so 3.8 should have been open to Europe. Yet beam headings pointed north, not east. The signal strength stayed constant even when I rotated the antenna 180 degrees. That flat response does not fit ordinary skywave.
A week later the same pattern returned at 02:14 UTC. Same offset progression. I recorded it on a laptop running old SDR software that still accepts analog IF input. The file shows a carrier suppressed by 18 dB with two sidebands spaced 220 Hz apart. The data content, if any, sits below the noise floor of my setup.
Similar reports turned up in 1973 logs from a station in Maine. Operators there noted identical timing on the same frequency during solar minimum. Their notes mention the signal vanishing when they switched to a solid-state receiver. The difference may lie in the front-end linearity of tubes versus transistors.
One possibility is that the pulses represent an old ionospheric heater experiment from the 1950s whose ground reflection path has remained partially coherent. Another is simple local oscillator leakage from another vintage rig. I cannot rule out either yet.
Further tests involved loading the same frequency into a restored BC-348. The sequence appeared again but with reversed polarity on the sidebands. That reversal matches the phase inversion one would expect from a reflection off a moving layer 180 km up. Still, the layer velocity calculated from the Doppler shift comes out negative, which contradicts current drift models.
archive notes from early shortwave logs contain several comparable entries from 1949. Operators described a “ticking” that started and stopped without regard to scheduled broadcasts. The similarity in timing structure is hard to dismiss as coincidence.
Uncertainty remains high. No one has yet triangulated a source location with enough stations to confirm an origin point. Attempts to decode the sequence as Morse or binary yield only random bits once the offset is removed.
Equipment Choices Matter
Modern transceivers with sharp roofing filters erase the very edges where these carriers sit. That is why the phenomenon stayed hidden until people began restoring older gear. The 6K7 second detector in the Hammarlund passes enough of the carrier to let the beat note form in the audio stage. Newer DSP rigs replace that stage with a clean digital product detector and the note disappears.
I tried the same frequency on a borrowed Kenwood TS-590. The waterfall showed only hash. Switching back to the tube receiver restored the pattern. The difference is measurable in the IF bandwidth curve printed in the old manual.
One experimenter in Nevada reported the same sequence on 3.795 MHz using a 6SN7 push-pull detector built from 1940s parts. His log matches mine within four seconds. Distance between stations is 780 km, so simultaneous reception rules out simple local interference.
propagation records from the same decade list several days when the critical frequency stayed elevated long after sunset. Those conditions could have allowed a low-power signal to remain trapped between layers. Whether any such signal was ever intentionally transmitted remains unknown.
Attempts to replicate the offset progression in simulation software require an ionospheric tilt angle that current models do not predict. Adjusting the tilt by 0.8 degrees produces the observed drift, yet satellite measurements show no corresponding tilt on those dates.
Fringe Interpretations and Limits
Some suggest the pattern carries information encoded in the offset itself. Treating each microsecond step as a bit yields a 64-bit string that repeats every nine cycles. The string has not changed in the three recordings I hold. That stability could indicate a fixed beacon rather than dynamic data.
Others point to early radar modulation schemes that used similar timing. A 1947 patent describes a system with 47-second frame periods for ionospheric sounding. The patent assignee later became part of a defense contractor whose records are still classified. No direct link exists, but the timing match invites further checking.
old patent drawings show a transmitter layout that would produce exactly the sideband spacing observed. Power levels listed are under 50 watts. A 50-watt signal reflected multiple times could reach detectable levels at night under the right conditions.
I remain unconvinced that anything exotic is at work. The data are consistent with an old, low-power transmitter whose output has been folding through the ionosphere in a stable duct. Until someone locates the physical source or reproduces the duct in controlled tests, the pattern stays an open measurement problem.
Further listening continues on the same receiver. Any new sequence will be logged with the same parameters. Updates will appear when the pattern either repeats with measurable change or ceases entirely.
josephcalitoy.com hosts additional tube receiver schematics that may help others repeat the setup. The circuit values matter more than the brand name.