What you see in the picture at the top, is a mostly hidden gem of HF propagation. I took the carrier of Sofia-Kostinbrod transmitter form Bulgaria (250kW) on 9400kHz and observed it for three hours. In the upper window you see the frequency wihtin a window of 2Hz height only. You see two strong carriers: one nearly in parallel to the x-axis, the other snaking some fraction of one Hertz below it.
With one transmitter only on this frequency: How does this happen?
It’s multipath propagation. The signal takes one way via a groundwave-like way, the upper trace. It reveals a very slight drift downwards. As I use a GNSS-controlled receiver, the FDM-S3 from Elad, this miniscule drift should be happen within the transmitter, not the receiver. The snaking trace stems from a second way, most likely via the F2 layer of the ionosphere. As the ionosphere is prone to winds and an ever dynamic change of its ionization, it is moving. And with all moving objects, also this causes a Doppler effect to waves. This is exactly what we see – the angular speed of the ionosphere, relative to the “groundwave-like” signal. You may also see at least two weaker traces, caused by two further ways, hence showing other Doppler shift.
In the diagram at the bottom, you see the combined level of all traces. Because they reach the reeiver at different time and, hence, different phases, their addition leads to an ever changing signal level, called: fading.
I hope to continue this work with some other examples in the future, also taking fade-in and fade-out into account.
Working on a project which will focus on Doppler spread of HF channels (see at the bottom) and other impairements, I also bumped into some more prominent Doppler catches, namely on the VHF aero band. I took the AM carrier of nearby Hannover VOLMET on 127.4 MHz and observed doppler traces about plus/minus 200Hz the carrier frequency. Following the acitvity in the airspace via Flightradar24 in parallel, it is easy to match traces and aircrafts. In this case, I nailed cargo flight NH8406 from Frankfurt to Narita/Tokyo. It is important to remember what is shown in left part of the screenshot: it is the signal of Hannover VOLMET, reflected by this moving Boeing 777-F. Thus, the reflected frequency shows a Doppler frequency shift – depending on the relative speed in respect to transmitter and receiver. A positive Doppler frequency signals that the aircraft is approaching my location. When it turns to the lower frequencies, I see the aircraft passing.
Things get more complex wen it comes to the Doppler shift at HF propagation. You will also see planes, but effects from high winds in the upper atmosphere, coming and fading of ionospheric layers and the influences of the geomagnetic field are prevailing. Due to the much lower frequencies, the effects are just about a tenth compared to thie above example on VHF.
See below a result from my observations on HF as a preview.
We all know that propagation largely follows solar activity – on a diurnal scale, as well as along the seasons and, above all, the solar cycle – see diagram at the top of the page. We just had entered solar cylce #25, and again some unchartered waters, filled with high hopes as well as some shallows. This is a good opportunity to have a look into the rear view mirror, as in general terms we can see part of the future in the past.
To do so, I did some explorative data analysis, available for free from the following sources:
The following diagram shows the correlation of the daily sunspot number with the maximum, the mean, and the minimum MUF as measured that specific day. Therefrom, you see a decent correlation between solar activity and the MUFmean, whereas the impact of solar activity onto the MUFmax splits up at sunspot numbers above ca. 125. Values above in some cases do enhance the MUFmax, and in some cases just to the opposite. The good news, however, is that even at low solar activity, there may be experienced a MUFmax well above 30MHz, the threshold from HF to VHF.
One of the reasons that “more isn’t more in each case” lays geomagnetic activity, also triggered by the sun. From the data trove, I took seven days of November 2012: A stream of days with good propagation is interrupted by one day where the MUFmax in knocked down from more than 30MHz to just under 15MHz. The reason is a coronal mass ejection (CME), resulting in a magnetic storm, disturbing the earth’s magnetic field and, hence, ionospheric propagation. Already on the next day, propagation is recovering. Please have also a look at the “silence before the storm”, namely November 13, 2012. There you see a slight enhancement of the MUFmax of about ten percent. The geomagnetic storm is lagging behind one to three days the solar activity, as calculated from the sunspots. So, you should use this information surfin’ HF just after a massive enhancement of solar activity – before propagation recedes for a day or so. By the way, a geomagnetic storm doesn’t hit the earth in each case. This has to to with the distorted magnetic field which is no plane but resembling the tutu of a ballerina (not my association …).
Tacitely, I had used the term “MUF” in the sense what is exactly dubbed “MUF(3000)F2”, or: “MUF of the F2 layer when communicating between two stations at a distance of 3000 kilometers”. Ionosondes mostly measure the vertical MUF with transmitter and receiver at the same location, which has to be multiplied by a specifc factor. This factor of about 1,5 to 4 largely depends on the distance of the two stations, and the height of the layer. The following illustration shows propagation between my location and Funchal/Madeira, just 3000 kilometers away in the Atlantic Ocean. At a given height of the refracting layer, one hop gets longer with lowering the elevation of the antenna beam. The lower this angle, the larger the hop. It is a difficult and often expensive task to get this angle under 10° or so, especially at lower bands.
The illustration below shows the MUFs of ten consecutive days for bridging different distances from 0 (critical frequency, measured directly at the ionosonde) to 4000km. The last distance reckons with a layer height of 330km, being quite optimistic in the years of low solar activity but will be reached easily in other times. Nowadays, the standard is to use MUF(3000), wheras you often find MUF(4000) in legacy literature.
In the last paragraph, I mentioned the height of the refracting layer, and this changes also with solar activity. How, is illustrated by the figure below. There you can see the heights of the four ionospheric layers F2 (highest), F1, E and Es, or Sporadic-E. At the bottom, you find the smoothed sunspot number. You can see that this mostly influences the F2 layer – raising it at high solar activity, and, hence, allowing for larger 1-hop-distances. The height of all other layers follow a much more regular yearly pattern, and are not that much dependant on solar activity.
This, in turn, leads to the fact that the critical MUFs of each layer, by and large, is matching solar activity. F2 and F layer are mostly influenced by solar activity, as all layers are by the season. See illustration below.
Just a look on how the MUF changes from day to day under more or less stable solar conditions. With the overall pattern remaining nearly the same, the MUFmax may change considerably, ranging from 20 MHz to well above 30MHz. Nevertheless, at each day you will see a sharp rise before local sunrise (fast iononization) and a flat slope (slower re-combination) directly from around sunset.
The following figure shows that there is a high probability to expect today’s propagation also tomorrow – to about 75 percent. There, I used each day of the full twelve years’ period to cover the whole solar cycle.
Did you ever asked yourself, why important contests and DXpeditions are taking place in spring and autumn, preferably in years of high solar activity? The two following figures will give an answer: the MUF peaks just in these seasons. The peaks are prominent in a year of high solar activity, and not that pronounced in a year of less solar activity.
Last, but not least, we now will leave the MUF, heading for power, or, in this case, field strength. The figure below shows hourly measured fieldstrengths on six HF channels from 22.5MHz (top) to 4.3MHz (bottom), normalized to 1kW EIRP from June 1, 1980 to December, 1993, after Deutsche Bundespost had cancelled this project. I added daily smoothed sunspot numbers to the top and the bottom figure. 22.5 MHz has even not been used at all during a time of low solar activity, whereas they stopped using 4.3 MHz from the beginning of a new solar cycle. You can easily spot that some solar activitiy greatly enhances propagation on the channels up from 6.4MHz. In this case, the daily time where you can use the higher channels up from 13.0MHz , also increases. In years of lower solar activity, the time of propagation on the lower channels, down from 8.6MHz, is increased. But this effect is by far not outweighed by the effect on higher frequencies. For these data, we are indebted mainly to Dr. Thomas Damboldt, DJ5DT (1941-2015).
To make use of the future you have to know the past. This is easy with cyclic, physical processes. So this look into history is also a look forward into same aspects of solar cycle #25. It will bring better propagation on higher bands, where we may use smaller antennas, face less atmospheric noise and have larger frequency allocations. I am sure that F2 propagation will even take miniscule signals (QRPP) around the world, allowing for daily contacts from Europe to Australia even on 6m – as I had experienced at the peak of cylce #23. Be prepared!
P.S. All calculations/diagrams had been made with Matlab. You may also try it with free Python, matplotbib and Seabornor other software, you have at hand. With millions of data fields needed, spreadsheet software must be avoided.
During my expeditions into the thicket of mediumwave offsets, I bumped into pictures like that at the top. In the lower part of the screenshot, you see two carriers mit seahorse-like structures looking to the right. In the evening, they look towards the West.
This is one of the several effects which can be seen at local sunrise/sunset. Here, the carrier gets “clouded” and show frequency changes. These effects are associated with Doppler shift (moving of ionospheric patches/layers) as well as scattering caused by irregularities of the ionosphere, most notably Travelling Ionospheric Disturbances, or TID. Whereas the Doppler shift, by vertical moving of reflecting layers like combining of F1- and F2-layer to one and lower F-layer when approaching darkness, is comparatively small, high wind speeds in these regions can cause a much faster horizontal movement of such regions. This, in turn, may cause a Doppler shift of about 1Hz or even higher in the medium wave range.
The Figure at the top demonstrates this effect at two transmitters on 1422kHz, namely SRR Radio România Actualități from Râmnicu Vâlcea/Olănești (sunrise 05:55 UTC/sunset 15:12 UTC; distance 1433km) and Radio Coran/Radio UFC/Radio Culture/Chaîne 3 from Ouled Fayet/Algeria (sunrise 06:58 UTC/sunset 17:00 UTC; distance 1840 km). Seen from midnight, sunrise first occurs at the Romanian transmitter, followed by the Algerian one with the seahorse-like pattern of the scatter towards the higher frequencies. Around each local sunset, first Romania sees darkness, followed by Algeria. Here, the scatter pattern turns towards the lower frequencies. In the insert at the right, contrast has been sharpened to additionally reveal a split-up of these carriers due to propagation into two paths.
This effect often helps to determine the local sunrise/sunset of a carrier. I marked what presumably is the carrier of MBC Radio 1 from Matiya/Malawi, sunrise 03:22 UTC; listed 02:00 to 22:00 UTC, but obviously on a 24 hours’ service this Tuesday.
Both Figures at the bottom try for some detective work without knowing specific offsets (because not available) but relying only on schedule and the above mentioned propagational effect. Crime scene takes place on 1233kHz, where we want to scrutinize two channels, one on 1232,9937 kHz, the other on 1232,9951kHz.
The s/off- and the s/on pattern match that of Chinese National Radio #17’s Kazakh service. Incidentally, sunrise takes place in Qinghe at 01:42 UTC, and in Boertala at 02:04UTC – next Figure. Boertala is listed with 10kW (stronger signal), Qinghe with 1kW. Unfortunately, the f/out time of other CNR17 transmitters on this channel is mostly covered by phase noise from Rádio Dechovka in the Czech Republic and Absolute Radio in the United Kingdom.
Here I am indebted to Jens Mielich, Head of the ionosonde at Juliusruh/Germany, who was so kind to comment on this observation. According to him, the observed Doppler shift of 1Hz on 1422kHz should have been caused by a refracting medium, moving at an (angular) speed of roughly 105m/s. At Juliusruh, he observed e.g., an ionospheric drift of 311m/s±93m/s from East towards West on January 19, 2021 at 04:19 UTC: “You will get a positive Doppler shift during a West/North drift, and a negative one at East/South drift.” He adds that further investigations on a more longer time series are needed.
Recently, I came across an upgraded version of Ivan Monogarov’s PskovNDB software, already having collected all laurels available as being the Gold Standard for chasing non-directional beacon, or NDBs. Recently, Ivan had expanded his tool with some as unique as exciting features for the avid medium wave DXer.
At a first view, it converts recorded WAV files (also: RF64 format, done with SDRC V3 software) into spectrograms of high resolution in which you can easily see the number of stations, measure their precise offset and see their signal strength.
A second view reveals the smart feature of producing diagrams of each signal – plus noise level and the combined power of the whole window. You can see both in the screenshots on top of this page.
A third view almost exactly helps to distinguish between signals where you can here music, listen at least to some words or phrases, or which do provide full audio.
Nothing more? Yes. Under the hood, there is much more. So, you can do automatically recordings each day and also automatically send them to PskovNDB software for showing the spectrograms, one after the other, like on a film roll. This enables you to pick the recording of the most promising day(s) for further inspection.
I wrote a short introduction to the beta version of this free software, and Ivan was so kind to add some most helping notes to this. You can download it here. It contains also some additional information, i.e. a link for downloading the software.
Spassiba, Ivan, for another software breakthrough!
With the new Elad FDM-S3 and its OCXO/GNSS-stabilized clock, I did a 24h recording of the whole medium wave band on January 19, 2021 in Northern Germany; plus longwave on Januar 21, 2021. Free software SDRC V3 enabled me to make up a spectrogram of each channel within a window of 50Hz width, and at a frequency raster of 9kHz on medium wave. You can easily see:
accurate and stable frequency offset over full 24h down to a millihertz
frequency control of the transmitter’s oscillator (stable, drift, sinus, sawtooth …)
propagational effects (doppler, scatter …)
The format is PDF, DIN-A4, landscape, resolution 300dpi – see screenshot at the bottom. This allows you to zoom to a factor of about 400% to search for details and better read out of the time/frequency scale. It weighs 865MB. You can download it here, and open it with your PDF reader (you can also point your mouse cursor onto the link, click right mouse key, and choose “Save under …”). Leafing from one page to another gives an interesting overview.
A similar Atlas showing a raster of 10kHz is also available for free – just scroll to the previous post of this blog. It is also planned to publish a general article about the background, about what to do with such a tool, and how to do this by yourself.
I am sure that it will open some new horizons on Medium Wave DXing, including accurate offsets over up to 24h.
With Elad’s FDM-S3 SDR now hitting the market, we have a receiver at hand which is supported by an OCXO/ GNSS frequency reference. This combines short-time accuracy with long-time stability and allows for precise frequency measurement in the millihertz range (under 30MHz). Exploiting this feature is as exciting as it is innovative. With this new tool, also a new kind of DXing is evolving. One example is propagation analysis. See below the 24h spectrogram of Radio Gotel from Jabura/Nigeria on its exclusive channel of 917kHz:
What surprises, is both, the late fade out at around 07:30UTC and the early fade-in as early as 15:20UTC. It is important to note that you here see the carrier with a resolution bandwidth of 0.0009Hz, roughly just one millihertz. The gain, compared to a listening bandwidth of 10 kHz, is a whopping 70dB, allowing extreme DX. Audio starts to emerge only from around 18:00UTC. As DX Atlas shows, the whole path between my location and Radio Gotel is under daylight at the palpable fade-in at around 15:20UTC, see screenshot below.
Broadcasting on medium wave still is a very active part of using the electromagnetic spectrum. An unique and outstanding source of information is supplied for free by MWLIST, a team of smart DXers. They provide tons of up-to-date and precise information – down to exact locations and even offsets from the nominal channel.
By visualizing those data, you get an even better insight. Here, free Tableau Public software is (for me) the tool of choice to do just that – please see the screenshot on top of this page. You simply download the free Tableau app, and – also for free – sign up, and you are done.
For me, most striking is visualizing the spatial data, i.e. to show the transmitters at their proper place on a map. Another welcome feature is filtering the data to answer specific questions like: How are traffic broadcast stations above 1.6 MHz spread over Pennsylvania? Or: What can I expect listening on 1521kHz on a late winter afternoon in Europe? Or: Where are Chinese stations located, carrying the CNR1 programme of China National Radio? You will find screenshots illustrating these examples below.
Those are just screenshots, not active maps. If you want active maps, there is an option (WP-TAB, Tableau Public Viz Block) available for WordPress’ business version which I don’t have at hand. But there is a simple solution: go to my Tableau Public page, download my TWBX-map “Medium Wave Station [Copyright MW List]”, and it will automatically be loaded into your Tableau Public app – after you have installed this. Then the map comes into live, and you can do all filtering, zooming etc.
[My profile photo shows a fisher’s deity in Japan, seen in October 2019 in Tokyo’s Kappabashi street. As a DXer and hobby cook, I thought location and statue being quite appropriate – thanks for asking …]
Surely, you immediately will find other ideas to realize, e.g. marking heard/verified signals by just a flag in your list and combining this with a special color on that station on your Tableau map.
P.S.: Taking some suggestions from the fruitful discussion which follows the initial publication of this site, I like to add some more examples:
If you are looking for some challenges, a European listener may start with low-power stations in the UK (LPAM), transmitting with just 1 Watt of power, leaving 500mW from both sidebands, combined, for the audio at max. Filtering the MWLIST with Tableau Public and visualizing this by a map, leads to the screenshot below. I also attached an audio clip of Carillon Radio. Yes, reception quality of this station of the Leicester/Loughborough hospital resembles a bit the state of NHS 😉
A second example is even more challenging for European DXers, but not entirely impossible. The map shows some low-power Japanese service radio stations for parks, traffic, weather and harbours.
Recently, I came across the different sign-on ceremonies of different transmitters. The idea is to understand this workflow in which obviously several stages of the transmitter are switched on consecutively. See at the top one example, where Voice of Turkey is swithcing on their transmitter on 9880kHz in five steps within about three seconds.
The diagram was made with Simon Brown’s unique software SDRC V3. I used the Signals Analyser module, providing a (needed!) time resolution of down to just one millisecond, or 1000 values of level vs. frequency in just one second! These data (CSV) had been exported and visualized in QtiPlot software.
I would like to encourage other people to join these observations. One goal can be toi fingerprint no only a transmitter, but also the workflow of the people at the transmitter. Please refer to this website for a database of broadcasters and their transmitters plus galore of associated data.
In the meantime, I already observed a couple of different workflows/transmitters. Please keep in mind that all these measurements (better: estimations), of course, are prone to fading. You may also see some effects during sign-on in the spectrogram, see below.
In the last weeks, I had used Sporadic-E conditions to stroll a bit in the FM broadcast band in search for DX. Elad’s FDM-S3 covers the whole 20 MHz wide band, and Simon Brown’s SDRC V3 software again provides an unique and most valuable tool to dig out DX. Antenna is an active Dressler ARA-200 (R.I.P.).
This blog entry shows how to make use of short openings of only some (ten) seconds.
First step is to record the whole FM broadcast band for hours on external HD. Then you make up so-called “spectrograms” by V3’s Analyser module. This provides you with a picture of activity (signal strengths color-coded) over time and frequency – see screenshot at top of this blog.
Scrolling through this spectrogram, you can make out even the shortest openings. Just click onto one of them, and the software instantaneously tunes into it. The sensitive RDS decoder of V3 is doing the last step – showing its RDS identification.
The short video below gives one example from a recording of June 26, 2020. On 91.8 MHz, I received semi-local transmitter NDR 1 NDS at Visselhövede (5kW@67 km distance), with “Stand by me”. From the spectrogram, I saw a “blob” (see screenshot at the top of this blog), stretching over around 40 seconds. It turned out to be Algerian’s Akfadou transmitter with Chaine 2 programme, 70 kW ERP@1’810km distance! RDS did tell me. Just have a look at the short video below which was made with V3’s video recorder …
V3 software provides also an a-symmetrical tuning of bandwidth, even at wide FM/BFM. This is important to identify some stations “in the clear” – if they are prone to some spillover from a local/regional station right on an adjacent channel. The following example spots Radio Marca/Mallorca from Spain on 91.6 MHz, suffering not only from a a strong local just 100 kHz below, but also from a very short appearance “out of the blue”, to where it disappeared again after less than 30 seconds. The latter is shown in the spectrogram, made by the Analyser, where I magnified the small/short signal of Radio Marca over 1’541 km. The video at the bottom shows how to evade the interference from the channel below to get the RDS code “B002 R.MARCA” correct.
Sometimes propagtion is too short for any identifcation, neither RDS, nor by announcement. Take the next screenshots as example: The spectrogram shows some very short openings revealing similar pattern which cropping the recording (Crop – > Apply) confirms. It turns out to be an English-speaking stations for a maximum of ten seconds. Parallel listening reveals the same programme on the following eight frequencies: 88.3MHz, 88.4MHz, 88.5MHz, 88,7MHz, 88.9MHz, 89.1MHz, 89.7MHz and 89.8MHz. The only intersection turns out to be Raidió Teilifís Éireann from different locations with their Radio 1 programme.
RTE transmitter usually do have RDS onboard, but here the time with a modest signal was too short to raise the alarm. On the other hand, there are stations with RDS, but not programmed or even without RDS at all. Take Radio Tisnath/Algeria in Tamazight, a Berber language, as an example for the first and Radio Blagovestiye/Russia as an example for the latter: