How do two targets become separable?
Switch between Range, Velocity and Angle. Each tab changes exactly the physical quantity that dominates the respective resolution.
Resolution, accuracy, maximum range and bin spacing are not the same thing
How close two targets may be in one dimension before they no longer appear as separate responses.
How close an estimate is to the true value. Good accuracy can be better than the nominal resolution.
How far or how fast can be measured without ambiguity. That is a different design problem.
Bin spacing and physical resolution are related but not always identical – especially with windowing and interpolation.
Range resolution follows from bandwidth
For an idealized FMCW signal, approximately: the larger the effective sweep bandwidth, the smaller the minimum separable range difference.
Velocity resolution needs observation time
Doppler velocity is estimated from phase evolution across several chirps. A longer coherent observation window yields finer Doppler resolution.
Angle resolution needs spatial aperture
Angular separation arises from phase differences across several spatially separated antenna channels. A larger effective aperture improves separability.
Better resolution costs resources
ΔR ↓More RF/ADC bandwidth and suitable chirp design.
Δv ↓Longer observation can affect latency and maneuver sensitivity.
Δθ ↓Larger aperture or more virtual channels increase hardware and calibration requirements.
Zero padding creates more samples – but no new physical information
A longer FFT via zero padding can interpolate peaks more smoothly and refine the display. But it does not automatically improve the fundamental ability to separate two physically non-separable targets.
Key point: More FFT points ≠ more sensor information. Resolution arises from bandwidth, observation time and aperture – not from a cosmetically finer axis.
The simple formulas are design anchors, not a complete performance forecast
SNR
Weak peaks may not be reliably separable despite nominal resolution.
Windowing
FFT windows change main-lobe width and sidelobes.
Amplitude
A strong target can mask a weak neighboring target.
Array calibration
Phase/gain errors degrade angle processing.
Off-boresight
Angle resolution is not constant across the entire field of view.
Waveform
MIMO and chirp planning couple resolution to further system parameters.
Frequently asked questions about radar resolution
Is 0.2 m range resolution the same as ±0.2 m accuracy?
No. Resolution describes the separability of two targets; accuracy describes the deviation of an estimate from the true value.
Does a larger FFT improve resolution?
Zero padding can refine the spectral display and peak interpolation, but it does not create additional physical information.
Why does angle resolution get worse away from broadside?
The spatial frequency depends on sin(θ); this changes the angular scale and the separation with viewing angle.
Sources
- Texas Instruments. Intro to mmWave Sensing: FMCW Radars – Module 1 (Range Estimation).
- Texas Instruments. Intro to mmWave Sensing: FMCW Radars – Module 2 (Velocity Estimation).
- Texas Instruments. Intro to mmWave Sensing: FMCW Radars – Module 5 (Angle Estimation).