Between the console and the FM transmitter sits a signal few people ever look at: the multiplex, or MPX. It carries mono, stereo and RDS in a single baseband. Understand its structure and you understand why FM radio sounds the way it does — and why it stays mono-compatible.
In frequency modulation, the transmitter doesn’t broadcast your audio tracks directly: it modulates its carrier with a baseband signal called the composite signal, or multiplex (MPX). This single signal stacks several pieces of information at different frequencies, so that both an old mono set and a modern stereo system find what they need. The whole art of the FM system lies in that stacking, designed from the start to stay backward-compatible.
Mono first: the audio baseband
In a mono broadcast, the composite is just the audio itself, limited to 30 Hz – 15 kHz. The transmitter applies a frequency deviation proportional to that signal, with a standardised maximum excursion of ±75 kHz. That ceiling isn’t a formality: exceeding it spills onto the adjacent channel and interferes with another station. That is why on-air processing polices levels so strictly before the transmitter.
Stereo without breaking mono
When FM stereo arrived, left/right separation had to be added without silencing existing mono receivers. The solution, still in use, is clever:
- The main channel (L+R) occupies the same 30 Hz – 15 kHz range. It is the sum of both channels: a mono set reads it as-is, just like before.
- A 19 kHz pilot tone is inserted, just above the audible audio. It serves as a frequency and phase reference.
- The difference channel (L−R) is amplitude-modulated with a suppressed carrier (DSB-SC) around a 38 kHz subcarrier — exactly twice the pilot. It occupies 23 – 53 kHz.
At the receiver, the decoder regenerates the 38 kHz subcarrier from the 19 kHz pilot, demodulates the L−R, then does the arithmetic: (L+R) + (L−R) gives 2L, (L+R) − (L−R) gives 2R. You recover both channels from two signals that, on their own, held neither. The pilot is injected at a modest level (around 8 to 10% of total deviation): too low and the receiver won’t switch to stereo; too high and it steals room from the other components.
RDS, at 57 kHz
Above the stereo channel comes RDS (Radio Data System), the data that shows the station name, the current title or triggers traffic announcements. It rides a subcarrier at 57 kHz — the third harmonic of the 19 kHz pilot, which keeps it synchronised. The data rate is deliberately slow (around 1.2 kbit/s) and injection is low, because RDS too consumes a share of the available deviation. It carries, among others, the identification code (PI), the service name (PS), radiotext (RT), programme type (PTY), traffic flags (TA/TP), clock time (CT) and alternative frequencies (AF) for following a station in the car.
Some stations add further auxiliary subcarriers (historically SCA, for example at 67 or 92 kHz) for extra services: reading for the visually impaired, data, background music. Each addition nibbles at the deviation budget, hence a constant trade-off.
Pre-emphasis and generating the MPX
FM applies high-frequency pre-emphasis before transmission, offset by matching de-emphasis at the receiver: this improves the signal-to-noise ratio in the top end. The time constant differs by region — 50 µs in Europe, 75 µs in North America and Japan — and every broadcast processor has to account for it. It is precisely the broadcast processor that generates the composite today: the stereo generator, once a separate box, is now built into the processing chain, right after the final clipper. The goal is to make the most of the permitted ±75 kHz without ever exceeding them — which is where much of a station’s sense of loudness is won.
Getting the composite to the transmitter
The MPX still has to travel from the studio (or the tech room) to the transmitter, sometimes miles away. Historically, the analog composite was carried over coax. Today, the studio-to-transmitter link often does it digitally: composite over AES, or IP transport using compressed schemes such as µMPX, which squeeze a full multiplex into a fraction of the bandwidth. Online radio skips MPX entirely, since it modulates no FM carrier: streaming online and broadcasting on FM don’t answer to the same constraints.
My take
FM multiplex is a small feat of engineering, more than sixty years old and still holding up: slipping stereo and data into a signal even a 1960s set can read is elegant. For anyone tuning a broadcast chain, the real practical point fits in one word: deviation is a budget. Pilot, L−R, RDS and SCA all share the same ±75 kHz, and every decibel gained on one is paid for on the others. Grasp that stacking and you stop setting things by guesswork.
