Sidebands

Post date: Mar 05, 2012 1:52:43 AM

In a couple of past lessons, we showed what amplitude modulation looks like in the time domain (oscilloscope). Last week, we considered it from the view of the frequency domain (spectrum analyzer), as a multiplication of two signals, producing 2 new products (a+b,a-b) in addition to the original signals. Let's examine that a little further.

Consider an AM broadcast transmitter emitting a 1 MHz (1,000,000 Hz) carrier, modulated with a pure (sine wave) 1 kHz test tone. Displayed on a spectrograph, it would look something like the picture to the left. You can see 3 distinct signals: the original carrier at 1MHz, and two other waves of lesser power, one 1kHz above, the other 1kHz below, the carrier. (Of course, the original 1kHz audio signal still exists, way down off the scale of the display, but it would never make its way off the antenna into space, even if it managed not to be filtered out by the transmitter circuitry before it got there!)Human speech, of course, consists of more than one frequency; a range of 200-3,000Hz would be nice. Frequencies below 200Hz are often used for "sub-audible tone squelch" and the like, so they are included in the band of frequencies allowed to modulate the carrier. When this mix of frequencies is mixed with the carrier, you end up with a band of frequencies from 0-3,000 Hz above the carrier (the upper side band) and one of the same size below the carrier; thus, the range of frequencies occupied by the signal would be from about 997kHz-1.003MHz -- a 6kHz-wide band. The spectrograph of such a signal would resemble the drawing below and to the right.There is a lot of "wasted" power in a standard AM radio signal like this; virtually all of the information needed to reproduce the original audio signal is contained in either one of the sidebands, and here we have a complete copy of it in the other sideband, plus all that power in the carrier that we don't actually need! Hmmm, if we could somehow get rid of one of the sidebands and the carrier, we could be putting out less than 25% of our original total power* while keeping our one remaining sideband the same -- while at the same time taking up only half the bandwidth! *or we can keep our power output constant, and have 4x the effective power in our transmission!This is the very concept behind SSB, known more verbosely as AM-SSB-SC (Amplitude Modulation, Single Sideband, Suppressed Carrier). The carrier and the other sideband aren't perfectly eliminated in the real world, but nearly so.The only real disadvantage is that by suppressing the carrier, you eliminate the ability for the receiver to automatically "lock onto" the signal -- there's no longer a base to determine the original frequency. That 1,001 kHz wave: is that supposed to be a 1kHz tone at 1MHz, or a 2kHz tone a 999kHz? Or a 3kHz tone at 998kHz? Or is it the lower sideband of a 1KHz tone at 1,002 kHz?

First off, to reconstruct the original audio tone, the receiver needs to produce its own signal to heterodyne with the received signal, to bring it back down to audio frequencies. This circuit is commonly known as the BFO, or Beat Frequency Oscillator. But there is still no automatic way to determine what the transmitter's frequency was, or whether you are receiving the upper or lower sideband; the only way to get it is by trial and error -- by adjusting your receiver until the voice sounds right. For the moment, we'll only consider the upper sideband (USB). If your receiver is set to 999.7 kHz and the transmitter was on 1,000 kHz, all of the sounds you hear will be 300 Hz too high, and the speaker will start to sound like Donald Duck or one of the Chipmunks. If your receiver is tuned too high, his voice will sound too low and be very difficult to understand. If you're listening in LSB mode and he's sending the USB, all of the frequencies will be inverted (low frequencies come out as high ones and visa-versa) and he won't even be understandable!

This is why even carefully-tuned SSB usually sounds just a little funny, and why your local broadcast station doesn't use it. It's a geek mode. It gives you way more "bang for your buck" than other modes like full AM-DS (Double Sideband) or FM, but it won't just automatically tune in and sound natural. It's hard to ignore its efficiency, however, which is why it's been used for over 90 years, and has been the standard for long-distance communication -- both "serious" and amateur -- for the past 50.

You don't have to completely eliminate the carrier to reap some benefit -- a compromise is certainly possible. You could reduce the power of the carrier (and the other sideband), but still leave enough to lock onto. This "reduced carrier sideband" would still require less bandwidth and power than the original signal -- it just wouldn't be as great a savings as more aggressive suppression would provide.

This is actually done, though not usually in amateur radio. NTSC television transmission, the analog standard used in the U.S. (though not as much now that most stations have converted to digital) uses amplitude modulation to encode the video. It takes a tremendous amount of bandwidth (4 MHz for each sideband), but it is crucial that the receiver be able to lock onto the signal, and there is a lot of low-frequency information that it is important be preserved. To trim the bandwidth from over 8 MHz to something that will fit into a 6 MHz channel, part of the lower sideband is filtered out, leaving about 1.25 MHz of it, and leaving the carrier and lower frequencies of the upper sideband pretty much untouched.