FMCW vs Pulse Radar | Architecture & PCB Impact

FMCW vs Pulse Radar: The Core Difference

FMCW (Frequency-Modulated Continuous Wave) radar transmits a continuous chirp that sweeps across a bandwidth, measuring range from the beat frequency between transmitted and received signals. Pulse radar transmits short, high-power bursts and measures range from the round-trip time of each pulse. This fundamental waveform difference cascades into every hardware decision—from the power amplifier and T/R switch to the PCB laminate, stackup and isolation strategy. Understanding how radar works at the signal level is the starting point for choosing the right architecture.

Waveform Principles Side by Side

FMCW: Continuous Chirp

An FMCW transmitter sweeps linearly from a start frequency f₁ to a stop frequency f₂ over a chirp period T_c. The reflected signal arrives with a time delay τ proportional to target range. Mixing the TX and RX signals produces a beat frequency f_b = (B / T_c) × τ, where B = f₂ − f₁ is the sweep bandwidth. Range resolution equals c / (2B), so a 4 GHz sweep (e.g., 76–80 GHz) yields roughly 3.75 cm resolution. Because the transmitter is always on, peak power is low—typically 10–13 dBm for automotive ICs like the TI AWR series—and the entire RF chain operates in a relatively benign, continuous-duty regime.

For a deeper look at chirp parameter selection and IF filter sizing, see the FMCW radar design guide.

Pulse Radar: Burst and Listen

A pulse radar transmits a short burst of RF energy (pulse width τ_p, typically nanoseconds to microseconds) and then switches to receive mode. Range is R = cΔt / 2, where Δt is the measured round-trip delay. Range resolution depends on pulse width: ΔR = cτ_p / 2. To get fine resolution without sacrificing energy on target, modern pulse radars use intra-pulse modulation (pulse compression), often a linear FM chirp inside the pulse—blurring the line with FMCW at the waveform level but retaining the pulsed duty cycle. Peak transmit power can reach kilowatts or more in military and weather systems, while average power stays modest thanks to a low duty cycle (often 1–10 %). This creates very different PCB thermal and isolation demands compared with FMCW. The pulse Doppler radar design page covers coherent pulse-train processing in detail.

Key Performance Parameter Comparison

When to Choose FMCW vs Pulse

Choose FMCW when you need compact size, low cost, fine range resolution from wide bandwidth, and the target scene is at short-to-medium range (< 300 m for automotive, < 30 m for industrial). FMCW dominates automotive radar at 77–81 GHz, gesture sensing at 60 GHz, and level measurement at 24 GHz. The bandwidth differences between 77 GHz and 79 GHz allocations directly set the resolution ceiling for these systems.

Choose pulse radar when you need long-range detection (tens to hundreds of kilometers), very high peak power, or must operate in bands where regulatory rules favor pulsed emissions. Weather surveillance (S-band, ~3 GHz), air-traffic control (L/S-band), marine navigation (X-band), and military fire-control radars are classic pulse domains. Pulse-Doppler variants add coherent processing for simultaneous range and velocity measurement.

For sensor-level comparisons beyond radar, the radar vs LiDAR selection guide covers complementary modalities.

PCB Architecture Consequences

FMCW Front-End PCB

Because FMCW radar transmits and receives simultaneously, the dominant PCB challenge is TX-to-RX isolation. At 77 GHz, even 0.1 mm of uncontrolled coupling path can degrade the noise floor. Key design features include:

• Separate TX and RX antenna arrays with physical spacing ≥ λ/2 (about 1.9 mm at 77 GHz) and ground-plane fencing between them.
• Tight impedance control on microstrip feed lines—trace width tolerance within ± 0.5 mil (0.013 mm) on low-loss laminates.
• Low-Dk, low-Df substrates such as Rogers RO3003 (Dk 3.00, Df 0.0010 at 10 GHz) for the antenna layer, often combined with FR-4 or mid-loss cores for digital layers in a hybrid stackup. See radar PCB stackup design for hybrid build-up options.
• Short, direct routing from the radar SoC ball pads to the antenna feed points—every millimeter of excess trace adds loss and phase error.

Pulse Radar Front-End PCB

Pulse radar boards face a different stress profile: high peak power and fast switching. PCB consequences include:

• High-power handling—wider traces, thicker copper (often 2 oz / 70 µm or more), and careful thermal vias under the PA to manage instantaneous dissipation.
• T/R switch or circulator integration—ferrite circulators are bulky and need precise mounting; solid-state T/R switches require controlled impedance launchers and fast bias sequencing routed on the same board.
• Timing and control routing—nanosecond-edge trigger lines for the pulse modulator must be treated as controlled-impedance transmission lines even if they carry digital-level signals.
• Thicker, more robust laminates—X-band radar boards (8–12 GHz) commonly use PTFE-based substrates 20–30 mil (0.5–0.76 mm) thick for the RF layers, with higher voltage-breakdown ratings than the thin cores used in mmWave FMCW modules.

Material Selection and Dielectric Loss

Both architectures are sensitive to dielectric loss (Df), but the impact scales with frequency and line length. At 77 GHz, a Df increase from 0.001 to 0.004 can add over 1 dB/cm of excess insertion loss on a microstrip—enough to wipe out the link budget margin of a short-range FMCW sensor. At X-band (10 GHz), the same Df shift costs roughly 0.15 dB/cm, which matters less per centimeter but accumulates on the longer feed networks typical of phased-array pulse radars.

Use the estimator below to see how a change in Df affects insertion loss for your frequency and trace length. Enter your operating frequency, substrate Df, and line length to get the additional loss in dB.

For example, a 77 GHz FMCW antenna feed that is 15 mm long on RO3003 (Df ≈ 0.001) shows roughly 0.9 dB total dielectric loss. Switching to a lower-grade laminate with Df 0.004 pushes that to about 3.5 dB—a 2.6 dB penalty that directly reduces detection range by ~25 %. In a pulse radar at 10 GHz with a 60 mm corporate feed network, the same Df shift adds approximately 0.9 dB, still significant but more manageable. This is why FMCW mmWave boards almost universally specify premium PTFE laminates, while lower-frequency pulse radars can sometimes use thermoset laminates like RO4350B or RO4003C and still meet loss budgets.

Component and Alternate-Part Considerations

FMCW radar modules increasingly integrate the entire RF front end into a single SoC (e.g., TI AWR and IWR families), so the BOM is short but each part is sole-sourced and critical. Pulse radar boards, by contrast, have longer BOMs with discrete PAs, LNAs, mixers, circulators and timing ICs—giving more flexibility but also more risk that an alternate part changes the impedance environment or thermal profile of the board.

Before swapping any component on a radar front-end board, check the impact on impedance matching, thermal pad footprint, and bias-network compatibility. The tool below helps quantify the risk of an alternate-part substitution on board-level performance.

A common scenario: substituting a GaAs LNA with a pin-compatible SiGe part in a pulse radar receiver. Even if the gain spec matches, the input return loss may shift from −18 dB to −12 dB, degrading the noise figure by 0.3–0.5 dB and creating a standing wave on the microstrip feed that ripples across the operating band. Always re-simulate the matching network after any active-device swap, and re-measure S-parameters on the first prototype.

Isolation, Shielding and Layout

TX-to-RX isolation is the single most layout-sensitive parameter in FMCW radar. Because the transmitter never turns off, any leakage into the receiver compresses dynamic range. Achieving 50+ dB of on-board isolation at 77 GHz requires:

• Continuous ground-plane stitching vias at ≤ λ/20 spacing (~0.2 mm at 77 GHz) between TX and RX sections.
• Metal shield cans or embedded cavity structures over the SoC and feed transitions.
• Careful power-supply decoupling to prevent coupling through shared VCC planes.

Pulse radar has a different isolation problem: the T/R switch must protect the receiver from kilowatt-level TX pulses and recover within nanoseconds. On the PCB, this means the limiter and switch circuitry sits physically close to the antenna port, with very short, well-matched transmission lines. Any via transition or connector in this path must handle high peak voltage without arcing—a concern that does not exist in low-power FMCW designs.

Fabrication and Inspection Differences

FMCW mmWave boards (77 GHz) demand the tightest fabrication tolerances in commercial radar: etch tolerance ± 0.5 mil, dielectric thickness tolerance ± 0.5 mil, and plating uniformity within ± 2 µm on antenna patches. These boards are almost always built to IPC Class 3 or beyond. Automated optical inspection (AOI) catches trace geometry issues, but X-ray inspection is needed to verify via fill and BGA solder joints under the radar SoC.

Pulse radar boards at lower frequencies (L through X band) have wider traces and thicker substrates, so etch tolerance is less critical in absolute terms—but the boards are often larger (up to 300 × 400 mm for phased-array panels), which introduces panel-level warpage and registration challenges during lamination. Hybrid stackups mixing PTFE RF layers with FR-4 structural layers require careful bonding-film selection from the laminate vendor to avoid delamination at solder-reflow temperatures.

Architecture Decision Flowchart

Use this sequence to narrow the choice:

1. Required detection range? Under ~300 m → FMCW is simpler and cheaper. Over 1 km → pulse or pulse-Doppler is likely necessary.
2. Operating frequency? 24 / 60 / 77 GHz → FMCW SoCs dominate. L / S / C / X band → pulse architectures are mature and well-supported.
3. Range resolution needed? Sub-10 cm → wide-bandwidth FMCW chirp. Meters-class → pulse width or modest pulse compression suffices.
4. Size and cost constraints? Must fit in a bumper or drone → FMCW module. Rack-mount or ship-mount acceptable → pulse.
5. Regulatory band? Check ITU and national allocations; some bands mandate pulsed operation with specific duty-cycle limits.

For range-tier-specific PCB guidance, the short vs long range radar comparison maps detection distance to laminate and stackup choices.

Frequently Asked Questions

Can FMCW radar detect velocity as well as range?

Yes. By measuring the phase shift of the beat signal across successive chirps, FMCW radar extracts Doppler velocity with resolution inversely proportional to the coherent processing interval. Modern 77 GHz automotive radars resolve velocities to ± 0.1 m/s.

Why do pulse radars need higher peak power?

Because energy on target is concentrated in a short burst rather than spread over a continuous waveform. To achieve the same signal-to-noise ratio at long range, peak power must compensate for the low duty cycle. The radar range equation shows that detection range scales with (P_peak × τ_p × N)^1/4, where N is the number of integrated pulses.

Is pulse compression the same as FMCW?

Not exactly. Pulse compression uses a modulated waveform (often a linear FM chirp) within a pulsed envelope, combining the energy advantage of a long pulse with the resolution of a short one. The RF chain still switches between transmit and receive, unlike FMCW where the transmitter runs continuously. The matched-filter processing is mathematically similar, but the hardware timing and isolation requirements differ.

What Df value is acceptable for 77 GHz FMCW radar?

A Df ≤ 0.002 at 77 GHz is the practical ceiling for antenna-layer laminates. Most production designs target Df ≤ 0.001 using PTFE-based materials like Rogers RO3003 or RO3003G2. Above Df 0.003, feed-line losses become difficult to budget around without shortening the feed network to the point where antenna aperture shrinks.

Can one PCB support both FMCW and pulse modes?

In principle, yes—some military and research radars use “multi-mode” waveforms on the same hardware. The PCB must then satisfy the stricter of the two requirement sets: FMCW-grade TX/RX isolation and pulse-grade peak-power handling. This typically means a larger, more expensive board with additional shielding and heavier copper, so it is uncommon in cost-sensitive commercial products.