Hi Paul, VLF, >second order... Like classic Luxembourg effect, the crossmodulation amplitude scales with the power of the heater(s) (ie 2nd order), and also linear with the victim. So overall it's a third order effect. In this case we look at the product appearing at f1 - f2 + f3 (ie. DHO - HWU + HGA). The same relation holds for classic AM heating, where f1 and f2 correspond to the carrier and an AM sideband of the heater. > 2 * 1.65kHz... No. Heater power density scales with (E1+E2)^2, and the modulation comes only from the mixed term 2*E1*E2', which is purely sinusoidal with the difference frequency. I am fascinated by this effect, and have been pursuing it for some time. I am quite sure that it's not an instrumental artefact, for several reasons: - the ICM product appears only at night, and only when there is also Luxembourg effect from other AM broadcasters, - I've observed it on two different passive resonant LF antennas. It is also picked up by the crossed loop DF-channel and lines up with the expected direction, - I do not see other possible combinations with high difference frequencies. These are suppressed by thermal inertia, but would be present in case of rx or antenna nonlinearities. Nothing is observed when HWU is on its alternate frequency 18.3 kHz, - the impressed depth of modulation is approximately right for the heater power and distance, - the ICM bands have sometimes appeared on other receivers as well (DK7FC LF-DX grabber, Twente WebSDR), - the relative phase of the crossmodulation is not constant but varies with propagation (see below). To study the effect further, I have set up a correlation experiment, using SpecLab to simultaneously receive the heaters, the victims and the ICM product. The right stereo channel receives VLF directly from two series-connected loopstick antennas, which are resonated and pointed to HWU 21.75 kHz and DHO 23.4 kHz. After some software filtering, and a small delay to compensate for path differences, the VLF channel is AM-demodulated in software to generate the 1.65 kHz difference frequency. This is further processed by an AGC and an FFT filter. On the left channel, I process the audio from my LF SSB receiver which is tuned to 135.5 kHz USB. After noise blanking, the signal passes a bandfilter for the ICM bands around 137.1 kHz. Note that to this stage the channels are completely independent, and the noise blanker knows nothing of the VLF signal. Finally the two outputs are multiplied. This removes the MSK spread from the LF ICM band, and transfers it back to the original victim frequency. The multiplier output feeds the primary channel of a phase-senstitive colour spectrogram, with display slots centered on the DCF39 (3330.6 Hz) and HGA (-70 Hz) idle carriers. The secondary channel is taken from the LF input before the bandpass, resulting in a colour which encodes the phase lag for the ICM path, relative to the direct paths from the VLF and LF transmitters. To visualize the procedure and results, I have uploaded some files to http://www.df6nm.bplaced.net/LF/vlf_lf_icm_2013/ - df6nm-grabber_090221_0100_annotated.png is an older colour-DF spectrogram from Feb 2009, which nicely shows the two "ghost bands" in different colours, - HWU_DHO38_DCF39_HGA22_DF6NM.jpg shows the geographical locations of the heaters, victims and receiver site, - Twente_LF_VLF_ICM_130210_0010_annotated.png shows the ICM products as seen in Holland yesterday night. Note that there are also 1.3 kHz bands which are presumably caused by ICM from DHO (23.4) and GQD (22.1 kHz), - VLF_loopsticks.jpg shows the loopsticks for DHO and HWU, - icm_130210_0200.jpg is an example of despread output from last night, - icm_plot_130210_0808.png is a level plot for last night. Despread ICM products are shown bright red (137.08 kHz) and bright purple (137.18 kHz). Strong red and purple are the direct levels of the LF victims, HGA22 and DCF39. The blue ones are the direct waves from the heaters, DHO38 and HWU, - SpecLab_settings_vlf-lf-icm_130210.ini is a SpecLab configuration file for the correlation experiment, for anyone wishing to try it at home. Best 73, Markus (DF6NM) ___________________________________ Re: [VLF_Group] Ghost bands created by VLF ionospheric heating Posted By: paul@... Send Email Sun Feb 10, 2013 10:57 am | Markus wrote: > DHO38 on 23.4 kHz and HWU ... 21.75 kHz > difference frequency of 1.65 kHz, > beat frequency between the two transmitters modulates the electron temperature > creates crossmodulation on the carriers of HGA22 (135.43 + 1.65 kHz) and DCF39 (138.83 - 1.65 kHz). That would be a remarkable kind of 2nd order Luxemburg effect. One minor flaw in this suggestion is that the heating effect would be at 2 * 1.65kHz since the heating doesn't care what polarity the signal has. Perhaps it is simply the LF receiver itself picking up a bit of DHO and HWU and mixing them. I am inspired to feed a little bit of signal from prx-e1 into an LF scanner or my BT878 card that samples at 448k/sec. -- Paul Nicholson -- ___________________________________ From: Markus Vester To: rsgb_lf_group@blacksheep.org ; rsgb_lf_group@yahoogroups.co.uk VLF_Group@yahoogroups.com Sent: Sunday, February 10, 2013 1:07 AM Subject: LF: Ghost bands created by VLF ionospheric heating On my LF grabber http://www.df6nm.de/grabber/Grabber.htm there is again a ~ 200 Hz wide red noise band centered on 137.08 kHz. It has been observed several times before but has been absent for a few months. Sometimes it is accompanied by a weaker "brother", centered on 137.18 kHz with purple (ie. northerly) colour. The structure is reminiscent of an MSK signal ("the ghost of CFH"). The bands were seen regularly but are stronger in nights with strong Luxemboug effect. After a lot of head scratching, a while ago I have found a really astonishing explanation for this effect: It is ionospheric cross modulation, with a pair of strong VLF transmitters acting together as a modulated heater. DHO38 on 23.4 kHz and HWU (which has recently reappeared on 21.75 kHz) have a difference frequency of 1.65 kHz, with the sum of the two FSK shifts. Similar to an AM demodulator, the beat frequency between the two transmitters modulates the electron temperature in the D-layer, and the variable absorption creates crossmodulation on the carriers of HGA22 (135.43 + 1.65 kHz) and DCF39 (138.83 - 1.65 kHz). This constellation is rather unusual not only in that the heating is done by VLF signals. What is even more peculiar is that unlike a simple AM transmitter, the two non-colocated heaters must produce an inhomogeneous and fast moving heating pattern. If you look at the Twente WebSDR http://websdr.ewi.utwente.nl:8901/ now, you can currently see and hear prominent 1.65 kHz FSK sidebands around DCF39 and also DCF49 (128.93 kHz). However HGA seems less affected at Twente, presumably because the easterly reflection area is not so well illuminated for that path. Best 73, Markus (DF6NM)