EVM
Understanding EVM
EVM is the most comprehensive single-number metric for digital signal quality. Unlike SNR (which only measures noise) or phase noise (which only measures phase variations), EVM captures all impairments simultaneously, making it the preferred transmitter quality metric.
EVM Sources
- Phase noise: Rotates constellation points, especially affects high-order QAM.
- Amplitude compression: Compresses outer constellation points, especially with OFDM (high PAPR).
- I/Q imbalance: Amplitude and phase mismatch between I and Q paths distorts the constellation.
- Frequency error: Rotates constellation over time.
- Thermal noise: Distributes constellation points randomly around ideal positions.
EVM Requirements
| Modulation | Max EVM |
|---|---|
| BPSK | -9 dB (~35%) |
| QPSK | -13 dB (~22%) |
| 16-QAM | -19 dB (~11%) |
| 64-QAM | -25 dB (~5.6%) |
| 256-QAM | -32 dB (~2.5%) |
| 1024-QAM | -35 dB (~1.8%) |
Frequently Asked Questions
What is EVM?
EVM measures the deviation of transmitted/received symbols from their ideal constellation positions. It captures all signal impairments in a single number. Lower EVM (in dB) means better signal quality. Each modulation scheme has a maximum allowable EVM.
What is the relationship between EVM and SNR?
In a noise-limited system, EVM (dB) approximately equals -SNR (dB). For example, 30 dB SNR gives about -30 dB EVM. Other impairments (phase noise, compression, I/Q errors) degrade EVM beyond what noise alone would cause.
Why does higher-order modulation require lower EVM?
Higher-order modulation packs constellation points closer together. The symbol spacing relative to the symbol magnitude is smaller, so less error is tolerable before a symbol is decoded incorrectly. 256-QAM requires 15x better EVM than QPSK.