Avatar 23kHz

On the recent Beans outing to Dublin, I briefly mentioned the frequency profile of FM transmission and the reasons we have alarms connected to certain frequencies. Since this raised a small eyebrow of interest I thought it would be a good idea, from both an educational and bean-scoring viewpoint, to expand further on this subject.

Settle in, this one’s going to be wild.

An introduction to FM

FM stands for Frequency Modulation, which refers to the way that audio is encoded onto a carrier signal for broadcast. It is easiest to describe if you first understand AM, or Amplitude Modulation, signals.

AM covers Long Wave, Medium Wave and Short Wave radio, all of which work in the same way: an audio waveform describes amplitude (loudness) by movement up and down, and describes frequency (high or low pitch) by the steepness and rapidity of its peaks and troughs. AM radio broadcasts an audio signal by modulating the carrier signal up and down to match the audio waveform.

FM is a little bit more complex, modulating the frequency of the carrier signal to match the amplitude of the audio signal. This enables the overall signal waveform to remain at consistent amplitude, which improves reception and fidelity, since the quiet parts don’t result in a weaker signal.

In the UK, a single broadcast FM radio signal will span about 100kHz of the electromagnetic spectrum, which is why radio stations on FM are approximately in multiples of 100kHz: in Leeds, Aire FM was on 96.3MHz, meaning it occupied the radio spectrum between 9630 and 9640kHz. Theoretically other stations could broadcast on 96.2 and 96.4, though in reality larger gaps between stations are maintained to avoid interference.

Division of the 100kHz spectrum

Since the human ear is capable of distinguishing sound between about 30Hz and 15kHz, the 100kHz bandwidth of an FM broadcast contains a far greater bandwidth than is needed to carry a single stream of useful audio. The reason so much bandwidth is available is to provide multiple services within a single station.

Within that 100kHz range, an FM radio station will broadcast several things:

  • Between 0 and 15kHz, a mono feed of the station’s output.
  • Between 23kHz and 53kHz, a stereo feed of the station – this being twice the bandwidth of the mono feed so as to carry two channels of audio, the divide between the left and right channels being at 38kHz.
  • At 57kHz, digital data for RDS (Radio Data System) compatible radios, carrying programme and station information – this shows the station name and “now playing” text on compatible radios, and can be used to trigger travel news announcements on car radios.

There is a deliberate gap between 15 and 23kHz, at the centre of which is 19kHz, and this frequency is allocated as a “guard band” that should remain silent to delineate the mono and stereo versions of the audio and provide a buffer if either extends beyond its intended bandwidth.

The importance of 23kHz

This brings us back to the conversation in Dublin, which I seem to recall Kev regrettably suggested was “quite interesting”. 23kHz is the frequency at the beginning of the wavelength allocation for the stereo signal.

Ordinarily broadcast audio will only contain audible frequencies, since there is no point broadcasting anything inaudible to the human ear, meaning it will virtually always stay within a 15kHz band. There are exceptions, though, usually in music that contains white noise or percussion, which might stray higher than 15kHz or contain harmonics that push into higher frequencies.

If broadcast audio contains frequencies higher than 23kHz, the highest frequency audio in the mono waveband will push through into the stereo part of the signal, where it will manifest as the low frequencies – often creating barely-audible bass rumble once the signal has been shifted into the audible range. For that reason we monitor FM transmissions for frequencies at 23kHz and have alarms that are triggered by them.

When our 23kHz alarms are triggered, the culprit is usually a song on a pop music station with a hi-hat in the percussion section, and we therefore refer to this as the disco alarm. These are the top five songs that repeatedly and reliably break the 23kHz barrier:

  • Donna Summer – “I Feel Love”
  • Sylvester – “You Make Me Feel Mighty Real”
  • Gloria Gaynor – “I Will Survive”
  • The Rednex – “Cotton Eye Joe”
  • Spiller – “Groovejet”

It’s likely, of course, that any song breaking through at 23kHz will also be breaking through the top end of the stereo frequency bands, and may therefore interfere with RDS data. It is therefore theoretically possible that playing “Cotton Eye Joe” on the radio will stop people’s car stereos from automatically switching over to travel news bulletins, if it was played at the time another station was broadcasting a list of rush hour traffic jams. But I don’t know if that would happen because it’s just occurred to me while I’m writing this, and in any case we don’t have any alarms monitoring the 55kHz guard band.


I hope you have enjoyed this exploration of the way that FM radio works. I think we can all agree that the following are the three main lessons we can all take away from this session.

  1. FM radio is encoded differently to AM radio, which produces benefits in terms of audio quality and reliability.
  2. The manner of FM encoding permits mono and stereo to be broadcast within the same signal, but introduces the risk that high audio frequencies in one band will interfere with another.
  3. Think very carefully before you show any interest at all in things I say to fill an awkward conversational pause. You never know what tedious hellscape will await you.

10 comments on “23kHz

Leave a Reply

Your email address will not be published. Required fields are marked *

Optionally upload an image to accompany your comment (JPG only)