The trade-off in one sentence
With phase alone, the constellation states are spread around a circle of fixed radius. With phase and amplitude, they are spread over several circles or a grid. This changes how many bits are sent per symbol, how far apart the points are and what is demanded of the transmitter.
Phase modulation
- Amplitude is constant, so the power amplifier can work close to saturation without distorting the information, with good energy efficiency.
- It withstands noise and channel amplitude variations well, because the information is not in the amplitude.
- As the number of phases increases, the points move closer together on the circle. The spacing between neighbors shrinks quickly, and a higher signal-to-noise ratio and an oscillator with low phase jitter are needed.
Combined amplitude and phase modulation
- By using several amplitude levels in addition to phase, more points can be accommodated with a greater minimum spacing than if they were all on a single circle, which improves spectral efficiency.
- The signal envelope varies, so the amplifier must be linear. If it compresses, the outer points flatten and errors and leakage into neighboring channels increase.
- It is more sensitive to channel gain variations, so the receiver needs more careful gain control and channel estimation.
How to decide
| Situation | What tends to suit it |
|---|---|
| Weak or noisy link, or low-power equipment where efficiency is critical | Pure phase, few states |
| Link with a good signal-to-noise ratio and a need for high speed | Amplitude and phase, many states |
| Channel that varies over time | Scheme that adapts to the state of the link |
That is why current systems negotiate the scheme dynamically: they use the densest one while the signal-to-noise ratio allows it and drop to a robust one when it worsens.
How it looks in practice
On a vector analyzer, a dense scheme with an amplifier at its limit shows compressed outer points. Backing off the power or linearizing recovers them.