Constellation Diagram Simulator
Also known as Constellation Diagram Analyser
See what each RF impairment does to a constellation diagram, and which one dominates the EVM.
Free to use. Sign up free to save, download and print.
Worked example: 256-QAM through a compressing amplifier
Total EVM: 3.66 %
Loading the Constellation Diagram Simulator…
How it works
Constellation diagram impairments and EVM
A constellation diagram plots each received symbol by its in-phase (I) and quadrature (Q) parts, over the ideal points of the modulation. Each impairment moves the symbols in its own way: noise blurs every point into a round cloud, phase noise stretches the clouds along arcs, IQ gain imbalance scales I and Q differently, quadrature skew leans the grid into a parallelogram, a DC offset shifts the whole constellation, an untracked frequency offset spins it into rings, amplifier AM-AM compression pulls the outer points in and AM-PM twists them.
The error is measured as the EVM definitions in the standards measure it, after taking out one best-fit complex gain, so a constant gain or carrier phase costs nothing. The fixed distortions (amplifier, IQ imbalance and DC offset) are worked out together over the ideal points, because they interact, and the random ones (noise, phase noise and a frequency offset) add to them in power. The two amplifier lines come from one curve, with the twist set by the compressed amplitude: AM-AM is the compression alone and AM-PM is what the twist adds on top of it, so together they add in power to the amplifier's own error. The analyser then names the impairment with the largest share of the error power.
ĝ = Σ y x* / Σ |x|²
EVM = 100 √(Σ |y/ĝ − x|² / Σ |x|²)
A(r) = r / √(1 + β r²), Φ = k A²
EVM = √(EVMd² + EVMn² + EVMφ² + …), EVMn = 100 / √SNR
MER = −20 log₁₀(EVM / 100)
- x, y
- ideal and received symbol, as complex numbers, with the ideal points at unit mean power
- ĝ
- the best-fit complex gain (amplitude and phase) taken out before the error is measured
- EVM
- RMS error vector magnitude, in per cent
- r, A
- symbol amplitude into and out of the amplifier
- β
- compression factor of the Rapp (p = 1) curve; the output flattens towards 1/√β
- Φ, k
- AM-PM phase shift in degrees, and its coefficient in degrees per unit of A²
- EVMd, EVMn, EVMφ
- EVM of the fixed distortions taken together, of the noise and of the phase noise, in per cent
- SNR
- signal to noise ratio, Es/N0, as a power ratio
- MER
- modulation error ratio, in dB
Standards
- ETSI EN 300 392-2 (TETRA) RMS vector error limit
- 3GPP TS 36.104 and TS 38.104 base station EVM limits
- ETSI TS 102 361-1 (DMR) 4FSK root raised cosine filter
Worked example
Worked example: 256-QAM through a compressing amplifier
A 256-QAM signal at unit mean power through an amplifier with Rapp compression β = 0.15 and no AM-PM, received at a 50 dB SNR.
| Step | Working | Result |
|---|---|---|
| Corner point | 1.63 / √(1 + 0.15 × 1.63²) | 1.38, down 1.45 dB |
| Point at the mean power | 1 / √(1 + 0.15 × 1²) | 0.93, down 0.61 dB |
| Compression, all 256 points | error after the best-fit gain | 3.65 % |
| Noise | 100 / √(10⁵) | 0.32 % |
| Total EVM | √(3.65² + 0.32²) | 3.66 % |
The corners lose 1.45 dB and the points at the mean power 0.61 dB, so after the best-fit gain the outer points sit inside their ideal positions: the pulled-in corners of a compressing amplifier. At 3.66 %, a 28.73 dB MER, the signal misses the 3.5 % limit for 256-QAM in 3GPP TS 36.104 Table 6.5.2-1 and TS 38.104 Table 6.5.2.2-1 by 0.39 dB. Halving β, the same as backing the drive off by 3.01 dB, cuts the compression term to 1.99 %. These are the model figures; the EVM measured on the plotted symbols carries their random noise and moves a little with each draw.
FAQ
Frequently asked questions
Not here? Ask us
Calculations run in your browser and nothing you enter is sent to a server.
What does a constellation diagram show?
It plots each received symbol as a point, with its in-phase (I) part across and its quadrature (Q) part up, over the ideal points of the modulation. The line from an ideal point to the received symbol is the error vector, and its RMS size over many symbols, relative to the reference, is the EVM. A clean signal shows tight dots on the ideal points, and each impairment moves them in its own recognisable way.
Which modulation formats are supported?
The analyser covers BPSK, QPSK and 8-PSK, the differential formats π/4-DQPSK and π/8-D8PSK, 2FSK and 4FSK, square QAM from 16-QAM to 4096-QAM, and the 32, 128 and 512-QAM cross constellations. The cross formats are built as square grids with the corner blocks removed, as used by point-to-point microwave radios.
Are TETRA and DMR waveforms supported?
Yes. π/4-DQPSK is the TETRA waveform and π/8-D8PSK is used by TETRA TEDS, while 4FSK covers DMR. FSK formats are shown in the deviation domain, and because an FM transmitter has no IQ modulator, IQ imbalance, DC offset and similar impairments are switched off for them. The TETRA vector error limit is included; no DMR limit is shown because the DMR standard specifies deviation tolerances rather than an EVM style limit. P25 is not included.
What about HF modem waveforms?
HF serial tone modems transmit PSK and QAM constellations that the analyser already covers, and a preset loads 8-PSK at 2.4 ksym/s with noise and phase jitter. Multi-tone waveforms such as ALE have no IQ constellation to display, so they are better viewed on a spectrum, for example in the FFT Spectrum Analyser.
How does the impairment chain work?
For a single carrier, impairments are applied in the order they arise in hardware: amplifier nonlinearity first, then the IQ modulator gain and phase errors, carrier phase and DC offset, then local oscillator phase noise and finally receiver noise. In OFDM mode the clipping and amplifier are modelled statistically, as a gain and a Gaussian distortion (the Bussgang method) applied after the IQ impairments. Each impairment can be set to zero independently, so its effect can be seen in isolation or in combination.
Is the EVM limit check a compliance test?
No. It compares the calculated EVM with the LTE and NR base station limits and the TETRA vector error limit for the selected modulation. It is useful as a pre-compliance indication, but a real transmitter still needs to be measured against the relevant specification.
Sign up free to keep your work.
Free to use. Saving, downloading and printing need a free account.