pa0r

There is a heated debate going on over which is the best digital mode. This debate gets fuzzier and louder every time a new mode is introduced. Those new modes always have more carriers, larger bandwidth and doubtful increases in performance.

This increase in performance is argued to be the very thin line between copy and no copy. The differences are measured in dB with reference to the noise floor, and for some the throughput speed of the system is the main cryterion. Meanwhile, the appliance operators (is your MixW version up to date?) cannot wait to apply the new mode and whine about not being able to tell them apart in their waterfall anymore. Yes, we are now turning the waterfall into some sort of TV where the 90% clueless can actually read with their own eyes that this is a PSK31 signal! I am amost certain I will witness the day when voice output will be added ("You are looking at a PSK31 signal, courtesy of the Microsoft Company"). (Sorry for that one, but I had to...)

Now you could argue about what is copy, and what not. And when have you last heard the noise floor in the digital portion of the band? In person-to-person communication we are used to signals with holes in them because of QSB or QRM. That's why we have learnt to repeat the most important information ("My name is Xilantriugharius, I spell X, i, l, ...). This has been practise since the advent of ham radio, and at least my generation thinks this is perfectly normal.

As soon as the machines (RTTY, TOR) took over the generation of the signal stream, and the operator did not get tired anymore when a long message had to be sent, as his brain did not have to do the de'coding', some clever people though that it might be advantageous to send everything twice, as a preventive measure. This marked the birth of FEC. FEC was used in one-tx-to-many-rx messages, the so-called bulletins. In station-to-station communication, however, not FEC was used but ARQ.

The difference between ARQ and FEC is simple. FEC allows to reduce the number of errors generated at the receiving end, and ARQ allows to kill them completely. For bulletins ARQ is not possible because you need 2 concrete entities for it, and bulletins have many receivers which would create chaos. The interesting thing about it is that when ARQ is used in a clever way it  has a lot less overhead than FEC, and wastes less power per symbol. Which is a nice contribution to ecology.

In other words, FEC is just a stop-gap which should be used in situations where ARQ is not possible.

Now, why is it that all these new fancy, qrm generating bandwidth eating modes are conceived? I think it is an outcome of the primitive way the male brain works. If more developers were women, you would have more emphasis on intelligent solutions than on 'wider, bigger, stronger', in other words the brute force method. I know that a lot of people are very much in favour of the bulldozer approach, but could we plase do it outside of ham radio?

OFDM modes work after the principle of: 'when you qrm one of my frequencies, I will transmit simultaneously on two frequencies'.  And 'when you qrm 8 of my frequencies' I will transmit on 8 more frequencies. 'And moreover, I will make sure I  transmit everything twice, 4 times or whatever I need to get the message through'. All this sounds very much like war to me. And frankly, some days I have the feeling I am already in the middle of one.

Now, as the free lunch (TM) is still 'pie in the sky', all these simultaneous carriers and send-twice-with-a-time-gap need lots of power (the larger the bandwidth and the length of the message, the more power), and all the unnecessary redundancy produces a lot of qrm on our crowded bands.

ARQ, however works more like the radio operator every large ship or airplane used to have - when they could still afford them. If some part of the message does not come through, it can ask for repeat of just that part... If my single carrier is knocked out for 10 seconds because of qsb, it will repeat just the part which was missing. And if my single carrier is suddenly disappearing because a bloating OFDM signal comes on top of me the ARQ system can wait until the brag tape is sent, and pick it up from there...

And now an easy puzzle: If it takes 20 Watts to produce an S9 signal with 1 carrier, how much power do you need for 16 carriers? Or 64?

Now let's see... if I am 5 carriers above the noise floor...