The single most challenging tech feat a RS can deploy: SFN!

Greece has some issues regarding radio stations operating within the country’s territory.

First, they are not licensed. True, no one is! All licenses have either expired and never renewed since 1990, or issued on 2001, meant to last until 2005, never newed since.

"But then, isn’t it illegal to operate a broadcast radio station in Greece?" one would ask. Nope! not if you were operaating illegaly at a specific date! the November 11th, 1999, to be precise(!).

That is because no political party ever came to be goverment dared to bear the political cost to actually issue a proper number of licenses or auction them under terms like financial prognosis or technical and content completeness.

The long prologue is to show that here in Greece, owning a radio station isn’t exactly the revenue creating business one could expect it to be, and investments on the area won’t ever be funded either by any bank loan (as banks properly require the business to be licensed to operate on whatever field they would, ie a food processing business needs to have a health and foods dept. license and HACCP european food processing standard certificate) or by any serious investor who targets at profit.

The cherry atop the pie, is that radio stations are obliged by law to use a single frequency (the one used back to the date mentioned earlier) , and to cover just their home prefecture.

But Greece has a unique terrain morphology that includes several thousand feet high mountains, coastal cities, towns, villages and tourist resorts.

Atop the highest of the mountains, which a prefecture could be easily be covered with one transmitter,

-either the presence of ancient civilisation ruins makes the installation of antennas an act punished like a capital crime(!) (thank god CP is forbidden by constitution in GR lol)
-or, in the rare case that no ruins exist, power companies won’t install PL where a paved road doesn’t exist.

So, radio stations need to use cliffs, rather than mountains, to cover the prefecture, but this is impossible due to the mountains between the cliffs(!). That leaves enormous gaps in coverage, but the only way to use a repeater is to use the same frequency on both sites.

In most cases, a severe percent of the coverage of the two sites overlap. This calls for extreme measures, like the ones used in digital television for reuse of the frequency. No matter how extreme this seems to look like, it is the only way a radio station here could cover the whole prefecture without using a pair or triad of frequencies.

Well, the technical issues regarding the problem are two-dimensional.

-First, you need to sync all transmitters down to cycle. This is easily achieved with unmodulated carrier, by using external PLL references synced to a reference bound to a GPS clock. Many new exciters have the relevant input, so a pair of GPS 10MHz reference clock generators is all that is needed. They are cheap, prices starting about 130$. The antennas are the standard car-type GPS antennas (they need just 3 satellites to extract the clock, anyway.)

-Second and hardest, is the aural sync. As I said before, the cost needs to be minimal. That by default excludes all AES-EBU digital STL’s and exciters, because their price range simply cannot be afforded by radio stations with maximum investment target 300.000$. It is half the whole invested value on STL’s and exciters. (What about offices, building and such?!?). We need to work with what we already have.

Of course, I understand that calibrating of spectral response curves should be incredibly accurate, precorrection would be needed in both sites, and extreme accuracy of analog levels should be achieved. But let’s skip this part for now (far easier compared to what follows)

I made some thoughts regarding WHAT should be used to achieve the delay needed (around 166 uS for a trip from main transmitter to repeater and half the way back, assuming a delay of a few uS from STL antenna to FM antenna (actually several phase shifts between the amplification circuits), and given a distance of 50km (~30miles) between the main and the secondary site. (75km the trip of secondary signal, 25km the primary signal which is already under way, 50km target delay [(1/300000)X50]plus the transmitter’s delay. The amount of the delay equals half a period of a sine wave near 6 KHz, which explains why bass usually escapes the delay and the phase difference but mid and highs sound muffled and thudded. When this comes to multiplex, you can not imagine the mess, you should hear it with your own ears.

Well, for the academic part, I made some thoughts of how to eliminate this part of the problem. An adjustable IF delay line could be made, and installed on a second identical discriminator stage inside the STL receiver. The delayed signal would then feed the primary transmitter, when the originally-timed signal would carry on to the secondary transmitter. But it can’t be done – SAW delay lines should be made application – specific, as noone builds delay lines @ 10,7MHz. SAW delay lines on order are usually fixed timed, and come in packs of 1000’s – bounces the cost sky-high.

Variable delay lines exist, but they are also frequency – specific, and maximum delay obtainable is in the range of 1000’s of nS – too little delay. We need to make a whole cluster of those to achieve the req’d delay.

Can we do it in software? If the software isn’t the “windows kernel mixer” of DX9, sure. But we need an enormous sampling rate and sample – accurate adjustable FIFO buffering to do that. A regular PC with the interrupts cannot do it – or CAN IT?

This is today’s question for you, Leif, the lord of audio DSP. Can one by any means set up a PC with whatever needed OS in it to run one card in loop or two cards in 384 or higher sampling rate with sample-wise adjustable FIFO buffering lenght?

I know that I can connect an ADC-DAC 768KHz pair through a serial memory and adjust the pick-up point for throughput for $20 parts and make big $$$ with it, but the question is, can it be done in software and save those poor guys of the development of PCB, the minimum order requirements, the P&P and the MTBF of such a device?