Wednesday, February 26, 2014

The Importance of Good Modulation

Searching for a new DX target got me to thinking about something I have noticed about a number of Shortwave Broadcast Stations. Many have good signals but are all but unreadable or unlistenable because of poor audio. I am not sure if this is coming about because of a lot of automation and lack of monitoring, or something else.

This came about because it appears that my new DX target will have pretty good audio. The posts that have me looking have been telling of Horizon FM in Tenerife ( Canary Islands) being relayed on 5780 kHz. I am not sure if its the station itself doing this or some hobby group. The posts seem to indicate they are starting with 75 watts and an inverted V antenna up about fifty feet. One post quoted the broadcasters as saying they would be processing the audio with an Optimod audio processor.

This gave me hope of hearing them for two reasons. First the Optimod in all its versions is a pretty good device for, as it name implies “optimizing the modulation”. And second, because this indicates that those involved in putting the transmitter on the air “ get it”. You have to get good audio into the transmitter for the listener to get good audio out of the receiver!

With Amplitude Modulation (AM) there are two parts to the signal: The Carrier and the Sidebands. The sidebands carrier the intelligence. In regular AM there are two, upper and lower, with the bandwidth being equal to the highest audio frequency transmitted. In other words, a tone of 5000 Hz will result in sidebands 5000 Hz above and below the carrier wave.

The more power in the sidebands, the more power can be recovered by the receiver. At one hundred percent modulation, the maximum power radiated is FOUR TIMES the unmodulated carrier power. Therefore it is important to keep the modulation at as high a level as possible. This is made somewhat difficult because complex audio is made up of different frequencies with different levels. Each instant of audio is not at the same level as it is the next instant. In order to keep the transmitter from being overmodulated, the overall level must be set so the highest peak does not go over 100 percent.

It might seem a simple matter just to keep the level going into the transmitter turned down to the point that the highest peak does not go over 100 percent. But if that is done, the other peaks would not reach that high and some effective power would be lost. There is also the average power in the audio to be considered. It is actually that average power that gives the impression of “loudness” and provides the real secret to having the station well heard above the noise and the clutter on the band.

To further complicate the issue, different forms of audio are complex in another way. There is what is known as the “peak-to-average” ratio. Human speak particularly has a much wider divergence between its peak power and its average power, so even if one were to be able to keep one's hand on the level control and somehow constantly juggle it to keep peaks and valleys near 100 percent, some program content would have less average power than others. The music or prerecorded material would always sound louder than the plain speech.

The answer is in forms of automatic level control. This is done with devices at two different levels. The first is called “ compression”. The compressor works very much like a human hand on the gain control, riding the gain and keeping the general program level fairly constant. It usually has a fairly slow attack and recovery time and follows the average program levels.

Then there is the peak limiter that follows it. This device actually works on that “ peak-to-average” ratio mentioned earlier. It works much faster and takes the errant high peaks and pulls them down to the general level of everything else. This is especially important with speech.

There are two other areas that have to be dealt with in what is termed in the trade (audio processing”. One is the fact that some forms of audio, particularly human speech, is not symmetric in its wave form. The positive and negative swings of the wave are not always equal. In Amplitude Modulation it is particularly important not to go over 100 percent modulation in the negative direction because this can cause extreme distortion and “splatter”...interference with stations on adjacent frequencies. It is also a waste of power that could be used to make the signal stronger for the desired listener. Modulation in the positive direction is less concerning in this regard. If a peak goes over what is termed 100 percent in the positive direction, the peak just continues to rise and actually delivers more power to the receiver.

The trick is that one never knows on which side this asymmetry will occur. Some peak limiters have been designed to sense this asymmetry and to electronically invert or “ flip over” if you will, the wave form so that the highest peak is always positive. If memory serves me right I believe the CBS Volumax limiter was the first widely distributed device to do this, with others following suit during the 1960's and 70's.

With both peak limiting and compression, time of recovery and attack determine just how “dense” the audio becomes. There is a limit, however, to how far you can take this. On low frequency notes, if you recover too fast, you actually cause gain recovery on part of the cycle itself and the device tries to turn the pretty sine waves of a bass into a square wave. On the other end of the spectrum, short duration percussion sounds would be so high and so short in duration that they would force the compressor or limiter to audibly “ push down” the mid range frequencies giving the impression of punching holes in the audio or giving a “pumping” impression.

During the 70's Mike Dorrough developed a multi band compressor that took care of that problem, by splitting the audio spectrum up into sections or bands and processing each separately, with slower attack and recovery times for the low frequencies, faster for mid range, and really, really fast for the highs. Many of the better processors today use this idea.

Then we get back to bandwidth. If the receiver was broad band enough to pass the entire audio spectrum, this would be the end of the story. But since regular AM broadcast receivers are not that broadband ( hearing interference from adjacent channel stations would be the result of that) any power transmitted by our station that falls outside that bandwidth is not only wasted, it causes interference to other stations nearby on the dial.

For that reason, on AM and Shortwave transmitters, it is prudent to limit the bandwidth of the transmitted signal to that which it is expected a receiver will pass. For standard medium wave stations in the US, this would be about 7500 Hz or less. For shortwave stations, it would be less...and in fact may be mandated to be less by some licensing agencies. Therefore, there is usually some form of filtering done to roll off the higher, less useful and often offending frequencies.

There is another area of power waste in the extreme low frequencies. In older transmitters that used high level plate modulation with large transformers to couple the audio from the modulator to the final amplifier, low frequencies at high levels could saturate the transformers, resulting in heating, but also in distortion and lost power. With listeners using smaller portable radios with small speakers that could not reproduce those lows anyway, this power was simply wasted. Even with todays “ boom boxes” it might be argued that the extreme lows are not that useful to the shortwave broadcaster anyway. It kind of depends on who the listener is expected to be and whether quality of music transmission or readability of speech programming is more important. The frequencies in speech that deliver understandability are not necessarily the extreme low end. For that, individual preference of the programmer comes into play. And for that, a higher grade version of the familiar graphic equalizer used in some home audio systems comes in to play.

In any event, it behooves the broadcaster to shape the baseband of his transmitted audio to fit the bandwidth and preferences of the receiver and listener where the sound will come out.

With limited bandwidth, it might seem to some audio purists that the sound would become dull and muddy, and in fact without some further enhancement it might. However, with some judicious reshaping of the frequency response of the audio chain in the upper midrange, where the consonant sounds of speech are generated and the beginning of what is know as the “presence band” begins, this can be overcome. By boosting audio frequencies in the 2500-5000 Hz range a bit with rapid rolloff above, one can put a bit of the “edge” back into the signal and the human ear and brain makes adjustments to fill in the rest. The ear is somewhat fooled into thinking it is hearing higher highs than it does.

The difference between the signal from a well processed transmitter and a non processed transmitter is not just minor...its phenomenal. The average audio is much higher, appears louder, overcomes noise and interference and becomes more listenable during less than optimal conditions.

The effect cannot be obtained just by putting a few boxes between the studio and the transmitter. The entire system must be considered. If the studio sends out audio that is distorted to begin with or is filled with hum or noise, those problems will only be magnified by the processing. The noise or hum will rise between syllables of speech or in pauses in the music. The path from the studio to the transmitter, whether it be telephone line, microwave or satellite link or internet connection must be able to handle the higher density audio and must not introduce their own distortion or noise. And the final peak limiter should be right at the transmitter input.

The transmitter itself must be in good condition. Tubes in the modulator cannot be of low emission or gassy or weak in general or there will be distortion. Tubes in the final amplifier must be able to deliver the positive peak output of the asymmetric audio or the positive peaks will be clipped off in the RF stage resulting in distortion or splatter. High voltage power supplies must be in good condition to supply the higher average currents and have voltage regulation to handle the extra current drawn during asymmetric, high positive peak modulation.

I have seen this work very well in standard medium wave broadcasting back in the dyas before FM dominated the music broadcasting world. How well does it work? In one station where I did the engineering we were on a crowded “ graveyard” channel-1340 kHz, with a power of 1000 watts day and 250 watts night. Many of our listeners in a nearby city were in an area that at night suffered some skywave and other distant station interference. With good processing, we were able to keep our audio above the clutter and gained many listeners. During the day the station was audible and listenable almost a hundred miles away. We were noticeably louder and “ bigger” sounding than our competition that ran the same power and were even noticeably louder on the dial than our other competitor that was a 5000 watt station. We had more audio coming out of the radio than they did, simply because they were not taking advantage of what they had.

It is somewhat disturbing, then, to tune across the bands and hear a large, international broadcaster burning up lots of power lighting up a 100 or 250 kilowatt transmitter and having it almost unlistenable because of low audio, poor modulation, or simply levels not set to achieve the optimum level of modulation. Given the cost of electricity and for some, the importance of getting a message out, that they would take note. And of course, for the DX-er, if they would, perhaps that station might more easily be entered into a log!

No comments:

Post a Comment