Free tool · OFDM fundamentals · TS 38.211 §5.3

OFDM, generated live

No canned screenshots — your browser runs a real IFFT on randomly modulated subcarriers and plots the constellation, the time-domain waveform with its cyclic prefix, the spectrum and the measured PAPR. Regenerate as often as you like.

Waveform parameters

Random data → QAM map → IFFT → cyclic prefix. Everything below is computed, not drawn.

TX constellation

Time domain — Re{s(t)}, CP shaded

Power spectrum (avg of 8 symbols)

Under the hood

What the IFFT actually does

s(t) = Σk Xk · ej2πkΔf·t   ⟺   s[n] = IFFT{Xk}

Each subcarrier is one QAM symbol parked on one tone. The IFFT sums them into a time signal whose tones stay orthogonal because each completes a whole number of cycles per symbol. The receiver's FFT undoes the sum perfectly — as long as the symbol boundaries line up.

The CP buys that alignment: copying the symbol's tail to its front turns multipath's linear convolution into a circular one, which the FFT diagonalises into a single complex gain per subcarrier — the whole reason equalisation in OFDM is one tap.

The price is PAPR: hundreds of independent tones occasionally add in phase, producing peaks ~8–12 dB above average (watch the measured value as you regenerate). That headroom is what power amplifiers must reserve — and why uplink offers DFT-s-OFDM.

Want the full story behind OFDM and the NR waveform?It's taught scene-by-scene in the 5G NR PHY Advanced masterclass.
See the course →
FAQ

Common questions

How does an OFDM signal get generated?
Data bits are mapped to QAM symbols, each assigned to one subcarrier of an IFFT input vector (unused carriers and DC set to zero). The IFFT produces the time-domain symbol, the last samples are copied in front as the cyclic prefix, and symbols are concatenated. This page performs exactly those steps in JavaScript.
Why does the OFDM spectrum look flat with steep edges?
Each subcarrier is a sinc in frequency; hundreds of overlapping sincs sum to a near-flat plateau across the used band, falling off at the edges where carriers stop. The slow sinc skirts you see are why real systems add windowing/filtering to meet spectrum masks.
What is PAPR and why does it matter?
Peak-to-average power ratio measures how much the waveform’s peaks exceed its mean power. With N independent tones, occasional in-phase addition produces ~8–12 dB PAPR, forcing the power amplifier to back off and lose efficiency — the main drawback of OFDM and the reason 5G uplink can use DFT-s-OFDM instead.
What FFT sizes does real 5G use?
It depends on bandwidth and SCS: a 100 MHz / 30 kHz carrier has 3276 subcarriers, typically realised with a 4096-point FFT at 122.88 Msps; a 20 MHz LTE-style carrier uses 2048 at 30.72 Msps. The generator’s sizes are scaled-down but mathematically identical.
● Go deeper — flagship course

This tool is a free taste of the 5G NR PHY Masterclass

From these subcarriers to CP-OFDM vs DFT-s-OFDM, windowing and why PAPR hurts amplifiers — the masterclass derives the waveform visually before a single formula appears.

48 cinematic lessonsaudio-video syncedreal 3GPP specs — no fluffcertificate included
Start the masterclass → Browse all courses