WHY RADIO STATIONS SOUNDED DIFFERENT
- AND LOUDER - IN THE 1970's
If you listened to AM radio in the 1970's, then
you know the unique sound it had - from both the programming and technical
point of view. 590/CKEY is here for you to listen to today!
We here at 590/CKEY are proud of using the exact
equipment that was used when 590/CKEY was the top station in Toronto in
the 1970's. Our competition was 1050 CHUM with the Drake/PAMS Top
40 format with its tight programming and jingles, and the very high power
in the audio sidebands that brought out every watt the transmitter was
Competing radio stations scrambled in finding
the best equipment for being the loudest sounding station. An added
benefit at the time was the fact that tube amplifiers and transformers
were in use in transmitters, which by their very nature having a warm sound
responded very well to attempts at driving them harder to produce louder
signals even when some distortion appeared. Attempts today to use
multi-band digital audio processing with the transistor switching modulators
now used in transmitters, while clean sounding, are not successful in reproducing
that classic loud sound, even when playing the exact songs from the 1970's,
because they cannot be driven beyond the digital clipping level, which
is fixed in absolute voltage level with no headroom to go beyond it, and
the multi-band processing has its own artificial artifacts that makes the
songs contain a somewhat unnatural frequency distribution - a sound similar
to CD's versus vinyl recordings: clean, but lacking something in its character
because of the signal approximation as a result of the limited digital
resolution 16 bits and low sampling rate 44.1 kHz = artifacts of omission
and sampling and aliasing errors, and a coldness due to the entire absence
of beating of >20kHz frequency components to lower, audible frequencies,
a natural phenomenon of mixing multiple instrumentation harmonics.
There is much confusion about how the classic
analog loud sound was achieved, in what order the equipment was connected,
and why it sounded the way it did. Here are the secrets revealed.
In an AM transmitter, unlike an FM transmitter,
the power of the carrier is modulated , or added to, by the audio signal.
The more modulation power that is added, the louder the signal, for the
same carrier power and antenna radiation pattern. In theory there
is no limit. In practice there are limits - but there are many tricks
that can be used to get around them. Engineers found these by trial
and error, and through experimentation.
WHY USE SIGNAL PROCESSING?
One: Radio is different than listening
to a CD, record, or tape at home. At home, the environment is quiet,
the CD plays directly into an amplifier and loudspeaker system.
Radio uses electromagnetic waves radiating outwards
from an antenna...every doubling of distance from the transmitting antenna
results in 1/4 the transmitted power available. There is natural
and man made noise present in the atmosphere which the receiving antenna
picks up. There is also noise in the radio receiver. As the
transmitted signal gets weaker, these noises become greater in proportion,
ie. the received signal to noise ratio gets poorer.
Two: A transmitter can only handle
a certain peak power output before distorting or clipping or having voltage
breakdown. Natural audio signals have a rather high peak to
average ratio. Turning down the transmitted audio to accomodate the
peaks so that they will just drive the transmitter to maximum peak output
would mean the average signal would be much weaker...further reducing the
transmitted signal to noise ratio discussed in reason one.
The psychology of listening is such that listener
fatigue will set in while listening to a poor signal to noise ratio.
The listener will soon tune to another radio station which has a better
signal to noise ratio, especially if similar programming is available from
another station. From a marketing point of view, if the transmitted
signal to noise ratio can be increased, more listeners will be attracted
-- this is the basic premise of the "loudness war" in radio.
Broadcasting stations in north america have a
limit of 50,000 watts carrier power output. The best way to situate
the transmitting antenna is outside of a metropolitan area, preferrably
near the ocean or a lake, and then use multiple antenna towers to beam
the signal in the direction of the metropolitan area (at the expense of
other directions). This will result in the greatest possible transmitted
field strength to the major listening area.
After the obvious basic things that can be done
to improve signal to noise ratio, transmitter location, power output, and
antenna gain, the only thing left is signal processing.
THE AUDIO PROCESSING CHAIN SECRETS REVEALED
Firstly, the high frequencies can be boosted.
Because most AM radios have IF amplifiers with limited bandwidth, audio
above 5kHz is severely attenuated. The FCC/DOC allows the transmission
of up to 11 kHz sidebands, or a total of 22 kHz occupied bandwidth, since
adjacent channels are not assigned in the same geographical area. The lack
of high frequencies in a received signal in the presence of atmospheric
noise in an AM receiver results in a poor signal to noise ratio.
Thus transmitter audio pre-emphasis boosts the high frequencies above the
noise level to some degree. Since the IF amplifier in an AM receiver
attenuates high frequencies, including noise, the resultant signal will
sound natural again, yet the noise has been reduced. This gives the
impression that the signal is stronger, and also sounds more high fidelity
at the same time. So the signal from the mixing console first passes
through a high frequency booster or equalizer.
Secondly, to ensure that the AVERAGE audio level
is as high as possible regardless of the program material, a compressor
is used. This compressor must have a medium attack time, around 28
msec, to avoid a sudden loud sound from being obviously suddenly reduced
in level. It must also have a slow decay time > 10 seconds to prevent
the average loudness from noticeably changing too quickly, which would
cause an unpleasant effect called gain pumping. All of this ensures
that the average audio level is consistently as high as possible without
leaving any obvious artifacts that the gain is actually changing.
The Gates Sta-Level is such a compressor, using tubes, and increases the
audio level by 2-15 dB compared to not using it. The actual benefit
depends on the program material, the original loudness variations, and
the rate of syllabic variation.
Thirdly, in 1970, CBS Laboratories produced a
revolutionary product. The Volumax 4300. The first analog signal
processor for maximum loudness. It contains two circuits. A
special compressor, and a peak limiter. This device is connected
after the main slow acting compressor. The special compressor is
a FAST acting compressor. Unlike the main compressor, this one has
a medium-fast attack time, around 8.5 msec, and a relatively fast release
time, around 260 msec. Percussive sounds faster than 8 msec do not
trigger the compressor and go straight through with their original volume
level. The syllabic sounds are compressed and the gain is re-adjusted
very quickly due to the 260 msec. This only works well if the average
signal input is relatively constant - the job of the main slow acting compressor.
The two compressors work together in preventing gain pumping, and allow
percussive sounds to sound loud because they don't reduce the gain of the
second compressor (ie. their attack time is not altered), yet the syllabic
variations are highly compressed and are brought up to maximum gain quickly,
making the signal sound very loud, with a further increase of 6 dB or 4x
power compared to not using it.
Forthly, because the transmitter uses power supplies
and amplifiers that have a limit to their power output capability before
clipping, distorting, or causing illegal signal splatter to adjacent channels,
the PEAK audio level must be instantly limited to the level before such
clipping, distortion, and splatter occurs. In the case of an AM transmitter,
the negative going peaks are critical, since the transmitted carrier cannot
go below 0 watts or -100% modulation - if it reaches 0 for any length of
time, there is no output at all, and the non-sinusoidal character of such
negative overmodulation results in illegal splatter and distortion of the
transmitted signal. The positive peak, on the other hand, is limited
only by the power supply and the modulator amplifier capability, as well
as the voltage insulation of the transmitter components and can go as high
in power as the modulator can cleanly produce without distortion, and as
long as voltage breakdown does not occur in the transmitter carrier amplifer.
This can be much more than +100% modulation that a symmetrical AM sine-wave
signal is limited to. Thus unsymmetrical modulation of the carrier
allows more sideband power in the signal compared to symmetrical modulation.
A peak limiter is the equipment that is used for this purpose. It
is a very fast acting device, with an attack time < 1 usec. In
this application, only the negative peaks would be clipped to keep them
at -100%, while the positive peaks would be allowed to go upwards beyond
+100%. By using the two compressors discussed above, the signal is
as loud as can be made before clipping cuts off the excessive peaks - a
much easier job to do compared to that if the signal loudness was varying
all over the place. The peaks can be removed very close to the average
signal - resulting in full utilization of the volume gain due to compression.
Thus the second circuit in the Volumax 4300 is
an unsymmetrical clipping circuit which takes care of the negative peaks,
fixing them at the -100% modulation level. The 6dB improvement from
this device alone is all totally useable since at no time will the transmitter
be overmodulated in the negative direction, neither will it be undermodulated,
as the average signal will always be close to the clipping level.
Every top 40 radio station wanted one of these. One does not really
hear any distortion because the energy in the clipped peaks is very low
- and what is left can fully modulate the transmitter toward -100%
Summary of equipment used in audio chain, in order:
RCA BC-5B tube mixing console
Frequency response 30-15000 Hz +/- 1 dB ref. 1
kHz. This is the broadcasting standard.
Kahn symmetra-peak, a device that reduces the
positive/negative asymmetry especially on voice signals, to manageable
values so that at -100% modulation the positive peaks are under +130% modulation
level: average power gain +2 dB
(Without this device, the highly unsymmetrical
nature of voice signals would require the negative direction to be peak
limited to less than 100% because the positive peaks would exceed the transmitter
components insulation breakdown, which is around +130% and no more.
The alternative is to clip the positive peaks more aggressively = too much
Linear pre-emphasis of frequencies above 5 kHz,
reaching +15dB at 15 kHz: average perceived on-air power gain +3dB
Gates Sta-Level tube compressor, attack time of
28 msec and dual constant release times of 2.35 sec and 10 sec: average
power gain +3dB
Volumax 4300 volume controller/peak limiter
8.5 msec attack time, 260 msec release time: average power gain +6
< 1 usec peak limiting of negative peaks for radio transmitter
Radio Transmitter, capable of +130% positive modulation:
average power gain +2 dB
(on-air signal only)
Total on-air power gain compared to not using
this equipment: +16dB (or 40x more power - thus a 50 kW station
sounds like 1000 kW!)
SUMMARY OF EQUIPMENT
RCA 44-BX and RCA 77-DX ribbon velocity microphones
ZaraRadio software package mp3/wave file player/jingle&commercial
scheduler on HP notebook computer
RCA BC-5B tube broadcast mixer console
Kahn symmetra-peak phase rotator
Pre-emphasis from 3-11 kHz +1 to +6 dB
Gates Sta-Level tube dual time constant slow
Volumax 4300 fast compressor/negative peak limiter
Brick wall 11 kHz splatter filter
SHOUTcast mp3 streaming audio server on Packard
Bell 910C computer, Bell DSL internet line
NOTES ABOUT DIGITAL SYSTEMS SUCH AS INTERNET
Since the power level represented by a digital
signal over the internet has fixed levels determined by the fixed number
of data bits, this energy limited representation requires the positive
and negative peaks to be symmetrical +/-100% and to be represented by data
bits 0 and 255 for maximum volume.
Thus the peak limiting of positive peaks necessary
for the SHOUTcast digital audio service over the internet makes a signal
that isn't quite as loud as the on-air signal, but it is still LOUD, +13
dB or 20x more power. Most internet radio stations use some processing
- many use excessive and artificial digital processing resulting in distortion
on low frequencies and a listening fatigue due to an unpleasant spectral
energy distribution - therefore, we are confident that our signal is the
loudest - and clearest - on the internet.
Please note that the 80 kilobits/second data stream
compression will create audible artifacts that would not be heard live
on the air. Plus, some of the "rare" commercials are poor mp3s with
artifacts unfortunately...Other than these, the audio is virtually distortionless