Olli, DQ8BHA, whose blog is listed on the sidebar, writes some very interesting material. His description of the 8 September aurora is worth reading (remember to use the back arrow to return here!).
Thanks to David, GM4JJJ, for alerting me to an article by Carl, K9LA, in October 2017 QST magazine entitled "Understanding Propagation with JT65, JT9 and FT8".
This article brought me to one of those "D'oh!" moments. Suddenly things which should have been obvious fit into place and I realise that I should have worked this out myself.
The article explains that the ability of the slow WSJT-X modes to receive signals "below the noise level" means that we can work stations using these modes when the bands are closed. Now that was sweeping statement by me. SOMETIMES. Let us think about why this can happen.
It is well know that these modes (and WSPR) can successfully decode signals which are below the level of noise in our receivers. In fact, so can good CW operators using their ears. The seemingly odd outcome arises both from the fact that the superb modes devised by Joe Taylor and his merry band can makes sense of extremely weak signals in this nether region, but also the way we define noise is rather arbitrary.
When your WSJT-X software shows that you have decoded a signal at -20dB, that relates to the noise factor for the SSB filter, and in reality it is not 20dB below what you can hear. However, make no mistake, it is a lot below what you can hear, just not quite 20dB. Let us take, for the sake of argument 10dB below what you can hear. Imagine that. Think of a signal that needs to be 10 times louder for you even to hear it, and then imagine decoding the weak version. Pretty impressive.
The fact is that the minimum signal to noise ratio required for reception of SSB is higher than that required for CW, which in turn is higher than that required for, say, JT65.
You may think of it this way. You are working a nice distant dx station just as the band closes as the MUF falls. In other words, rather than being below the "maximum usable frequency" (MUF), you now find yourself operating above the MUF. Whilst on SSB the signal would have faded into the noise, and you can no longer hear it on the loudspeaker, on FT8 or JT9 you are still about to complete the QSO. Magic. Except that it happens all the time if you use FT8 or JT9 (or JT65 or WSPR). You just keep completing QSOs where you cannot hear the other station in your loudspeaker.
However, there is another way to look at this. For an SSB operator the band has closed. They cannot make a QSO as the MUF has fallen and "the band has closed". But you, as a data mode operator can still work people. What you are doing is using what is called "above the MUF" propagation.
In effect, using these data modes has made an otherwise closed band stay open for longer.
Let us have yet another take on this. When I started on data modes back in the 1970s I used RTTY. I might be appalled by that mode these days, but it was cutting edge then when nobody had a PC. RTTY was (should be past tense) a mode which basically replaced the microphone with a bulky, oil-spewing, unreliable, clattering electro-mechanical beast. You did not get any extra performance out of this contraption, it just meant that your QSO got printed out instead of spoken. Unlike RTTY, WSJT-X slow modes are not just a text-based replacement for voice or CW. They out-perform SSB by being able to be used successfully in conditions where phone, RTTY and CW would not work at all.
In the 1970s, during any opening, my RTTY success directly mirrored by SSB success. If the band was "open" I could use SSB or RTTY. If the band was "closed" then both stopped working. But the WSJT-X slow modes continue to work. So how does that happen? Surely the F-layer (or the E-layer) is either bending the signals back to earth, or it isn't. How can using different mode make a band open?
Let us go back to basics. The diagrams are my own (copyright!) handiwork, not to scale and can be enlarged by clicking on them. The QST article is based on 28MHz and F-layer propagation, but there is no reason why the same principles would not apply to other bands or E-layer propagation. For this purpose though I will stick to the same example as the article.
When the band is "open" the ionosphere bends (some of) the signal back down to Earth in the well known way. The classic diagram shows band open ...
... (above) where you would expect to find propagation between A and B, and (below) closed, when you would expect nothing to happen between A and B ...
The diagrams above show the F-layer as if it is a thin line which either reflects (band open) or refracts (band closed) the radio signal. In reality we would have no propagation at all if that was the case. The F-layer could never reflect radio signals at the angles we transmit them. The F-layer is not a mirror, it is a layer of ionised gas which has a structure of steadily varying density. This variation in density results in a very large number of small refractions of the signal, gradually bending it down until it is almost horizontal, and only at that angle is there one, small, reflection which returns the signal via a whole series more of refractions.
So almost all the work of returning our signals to Earth is done by a large number of refractions. Let us look at a diagram of how the F-layer would look - and this is definitely not to scale - when the band is open ...
What nobody told us in radio school was that every time one of those refractions takes place, there is also a reflection. This has been known for hundreds of years in optics, and light is just a different wavelength of electro-magnetic energy from radio, so the same principles apply. The Fresnel equations can calculate the relative strengths of the reflection and the refraction. The other basic optical principles apply too - so the angle of the diffraction will depend on the relative difference in density, but the angle of the reflection will still be the same as the angle of incidence. Which means that actually we get something like this ...
In fact, given the noise handling ability of your radio you might never hear them. But JT65 can.
And JT9, FT8 and WSPR can hear them too. WSJT-X slow modes can successfully decode signals well below what we can hear. You might correctly take that to mean that they can hear weaker stations when the band is open, but it can also mean you can work stations when otherwise the band is closed and you can hear nothing but noise on the loudspeaker.
Moving on from the time the band is open until when it is closed we would get this diagram for the ionosphere ...
The result of this is that before the bands open, and after they close, the weak signal modes should be able to decode signals we cannot hear above the noise. We need to re-think our existing assumptions. Most MUF predictions are made on the basis of a conventional SSB radio with about 100W and a dipole or small beam. The QST article suggests that a path of almost 3,000km, a single F-layer hop, would be open for an SSB contact at 28MHz (obviously) with the MUF of 28MHz. In fact this could be done on low power, with 100mW of CW doing the trick. But of course once the MUF falls below 28MHz this path is lost and the band is considered to be closed.
The article goes on to suggest that CW using a narrow filter could keep the path open at 28MHz at 10W even if the MUF falls to 25MHz. So, operators are already using "above the MUF" propagation. However, using FT8, JT65 or JT9 this path would be open with the MUF of around 23MHz. So the ten metre band would sound dead, SSB would be possible on fifteen metres, but data operators could operate on the otherwise "closed" ten metres. CW operators might get away with the WARC band on twelve metres.
The significance of this is that the MUF rises to 23 MHz far more often than it reaches 28MHz.
The fact that WSJT-X data operators are making stacks of contacts when the band is open is already clearly demonstrated. But this other fact shows what some of us had already noticed - these modes can make contacts possible on an otherwise "closed" band.
It could be said that there is confusion over our own figure "maximum usable frequency". For practical reasons we have set this figure by taking the measured critical frequency using near vertical incidence reflections and multiplying it by a constant which produces a figure which works for SSB and the receivers we all normally use. However, the better sensitivity of these WSJT-X modes alters the constant to be applied. What we call "MUF" is really the "Maximum workable frequency for easy SSB contacts".
In a sense it is silly to talk about "above the MUF" contacts as F-layer propagation should be impossible above the MUF by definition if it is really the maximum usable frequency. However, I bet that the term MUF will continue in use to mean the frequency at which those easy F-layer QSOs start happening.
I avoid the easy contacts and go for the difficult ones. But you knew that already.
So what happens if the MUF is much lower, and the F-layer basically disappears? Does this "above the MUF" propagation disappear? Not totally. At that point although reflection more or less stops, even at a very low level, scattering from the atmospheric molecules will still occur and produce just the sort of weak signals which JT modes love. Ionoscatter has been known for years too, but it usually requires high power as it produces weak signals - something which JT modes are ready to help with.
I should have seen this coming. I knew that these modes can receive much weaker signals than the human ear. What I had not thought about was that they could in effect outwit the conventional calculation of "maximum" usable frequency. None of this is new - Isaac Newton (1642-?1726) knew about reflections during refraction in light. The people who made my camera or my glasses spend a lot of time trying to minimise the effect by applying coatings to the lenses. However, our radio educators thought fit not to remind us about it. As so often, the standard diagram in the radio text books is over simplified. Oh yes, it was over simplified on this blog too ... mea culpa.
When I started using WSPR I told some old timers about the results I was getting. Their immediate reaction was that it is was impossible and somehow WSPR must be using the internet rather than radio. When I assured them that WSPR was all radio over the whole route from my antenna to the other station's antenna they were very sceptical. Now I know what was happening. More recently as WSJT-X modes became more popular on 6m, several of us have been finding paths open when the band is otherwise "closed".
As always, more investigation is required.
This is my stumbling attempt to explain this. I encourage you to look up the much clearer explanation by K9LA in QST if you can. I hope to put it to more use soon.
And thanks again to GM4JJJ for putting me on to this. It explains a lot of what I have been experiencing but not understanding.
I wonder how often stations turn on, listen to the band, hear nothing, and switch off. What would happen if they tried calling CQ on FT8?