77 GHz RADAR / BEFORE ANTENNA LAYOUT
77 GHz radar PCB requirements should connect the sensing task to a specific device configuration, RF construction, mechanical envelope and acceptance test before antenna artwork begins. Start with measurable range and angle requirements, then identify which assumptions belong to firmware, the PCB, assembly and the installed sensor. A laminate name and a 50 Ω routing rule do not complete that specification.
This worksheet turns those decisions into a reviewable handoff. Its original example shows why a nominal 4 GHz chirp does not necessarily deliver 4 GHz of sampled bandwidth. All numerical results are ideal calculations from selected teaching inputs; they are not measurements, a validated device configuration or a claim about production radar performance.
1. Write the sensing requirement before choosing the antenna
Separate maximum detection range, minimum operating range, range resolution and range accuracy. Record the target type, orientation, environment and detection criterion beside each value. Two targets that must be distinguished at the same angle create a different requirement from estimating the distance of one isolated reflector.
Do the same for angle: field of view, angular resolution and angular accuracy need separate entries for azimuth and elevation. Texas Instruments’ range and angular resolution brief, SWRA841A, January 2026, distinguishes two-object separation from single-object localization accuracy. An angular-resolution figure also depends on the array and processing assumptions. A wide field of view does not by itself establish fine angular resolution.
Create one row per requirement, with these five fields: target, conditions, responsible owner, controlled input and verification evidence. For example, assign sampled bandwidth to the radar configuration owner, antenna coordinates to RF engineering, and mounting tolerances to mechanical engineering. Leave unresolved values visibly open. Do not let “to be confirmed” become an invisible default in released artwork.
2. Freeze the device and reference-design boundary
Record the exact sensor, package, silicon/document revision, SDK or firmware version, and reference-board revision. Identify what is being reused: a power architecture, a package transition, an antenna layout or a complete assembly. Each has different dependencies. Keep a change list for the parts of the reference design that the product cannot preserve.
As a concrete device example, the AWR1843 data sheet, SWRS222D, September 2024, lists 76–81 GHz coverage, 4 GHz available bandwidth, three transmitters and four receivers. Coverage is not the same quantity as usable chirp bandwidth. Nor does the channel count alone establish twelve distinct, uniformly spaced azimuth positions. This example identifies a documentation boundary; it is not a recommendation to use that device for every radar.
Attach the actual TX/RX coordinate map and channel schedule. State which channels contribute to azimuth and elevation, which measurements occur at different times, and how calibration is retained. The array PCB review provides related questions about feed geometry and channel consistency. A MIMO virtual array and an electronically steered transmit array should not be treated as interchangeable architectures.
3. Calculate resolution from the sampled sweep
For an ideal linear FMCW sweep, use ΔR = c/(2B), with c = 299,792,458 m/s and B in hertz. Here B is the frequency span observed during the useful sampling interval. For constant slope S, the worksheet uses B = S × Tsample. TI’s Programming Chirp Parameters, SWRA553A, February 2020, describes the resolution relationship and separates ADC start, sampling and excess ramp time. These intervals must be considered before using the full ramp span in a resolution estimate.
Select a teaching slope of 80 MHz/µs and a 50 µs linear ramp. The full span is 4,000 MHz, or 4 GHz. If useful sampling occupies only 40 µs, the observed span is 3.2 GHz. If it occupies 30 µs, the span falls to 2.4 GHz. The timing margins here are invented examples, not prescribed settling times for the AWR1843.
| Interval used | Bandwidth | Calculated ΔR | 50 mm ideal target |
|---|---|---|---|
| Full 50 µs ramp | 4.0 GHz | 37.474 mm | Only if fully usable |
| 40 µs sampled | 3.2 GHz | 46.843 mm | Passes this screen |
| 30 µs sampled | 2.4 GHz | 62.457 mm | Fails this screen |
To reproduce the 40 µs row, multiply 80 × 40 = 3,200 MHz, then compute 299,792,458/(2 × 3.2 × 109) = 0.0468425716 m. A selected 50 mm ideal target requires at least 2.99792458 GHz. This arithmetic screen catches an inconsistent requirement before layout; it does not guarantee that two real targets will be separated.
Windowing, signal-to-noise ratio, target contrast, clutter and the detection algorithm affect the observable result. FFT zero-padding can create a finer displayed range grid without supplying missing sweep bandwidth. Maximum range also requires a separate IF/sampling and signal-budget check. Preserve the exported device configuration and processing settings alongside the worksheet.
Check the bandwidth assumption. Enter 3.2 and 2.4 GHz in the ideal FMCW range-resolution tool. Expect 46.843 and 62.457 mm. Record how the usable bandwidth was obtained; the tool does not validate chirp timing, device limits or target detection.
4. Treat construction changes as RF design changes
The RF cap thickness, foil, plating, solder-mask boundary and package transition belong in the same controlled construction record. TI’s RO4835 LoPro example in the RF PCB design guide (SPRACG5, May 2018) places that material between layers 1 and 2 of the AWR1642 evaluation board, with FR4 below. It also shows that changing the substrate changes the package-to-PCB transition. That historical example is construction-specific, not a universal stackup for later radar parts.
For your project, list the proposed construction next to the reference construction and highlight every difference. Ask the RF owner to disposition each change: preserve, remodel, prototype or reject. A substitute from any material supplier needs this comparison; matching a nominal Dk number does not establish matching transition or antenna behavior. Use the material qualification worksheet to record the exact grade, thickness, foil and evidence behind the selection.
Mark critical features on a separate drawing: package launches, feed widths and gaps, antenna dimensions, reference-plane openings, mask edges and relevant via locations. Assign tolerance limits using sensitivity analysis and the fabricator’s proposed process. The 50 Ω trace tolerance example explains the screening method; its simplified transmission-line calculation is not a 77 GHz antenna or package-transition signoff.
Agree how any CAM compensation will be reviewed. For an antenna feed, a small artwork change may affect more than impedance. Request an annotated disposition of RF-critical changes instead of authorizing unspecified trace resizing. Identify what will be measured on initial boards and how those dimensions will be associated with the corresponding RF results.
5. Reserve the radome, mounting and thermal envelope
Bring the enclosure into the antenna model before the PCB outline freezes. TI’s mmWave Radar Radome Design Guide, SWRA705, August 2021, discusses how material, thickness, shape, incidence angle and antenna spacing affect transmission and radiation behavior. A normal-incidence wall-thickness estimate is a starting calculation, not proof across a product’s field of view.
Give mechanical engineering a controlled envelope containing the antenna aperture, nearby fasteners, brackets, connectors and radome surfaces. Identify resin grade, coatings, thickness tolerance and antenna-to-cover spacing. Define which environmental states the product must be evaluated in, including temperature and any relevant surface contamination or water condition. Avoid approving an enclosure only from a nominal material name.
Thermal planning needs the sensor’s actual operating mode. Record active channels, frame timing, duty cycle, processing load, rails and enclosure boundary conditions. Ask the hardware owner for the associated power budget and the mechanical owner for the cooling path. Keep estimated junction behavior separate from measured board or case temperature. Do not copy a reference board’s power result into a different firmware mode.
6. Assign evidence to the correct build level
Use three acceptance records. Each should name the hardware and software revision, method, operating conditions, pass criterion and result owner. Decide before ordering whether the prototype is intended to verify geometry, exercise the sensor, or demonstrate an installed sensing task. One build can serve all three purposes only if the required access and tests were reserved.
- Bare PCB: identify material and stackup records, RF-critical dimensions, registration, finish and agreed coupon measurements. Define measurement locations and sampling. A TDR impedance result alone cannot establish antenna pattern or 77 GHz insertion loss.
- Assembled sensor: verify the intended firmware/configuration loads, rails behave as required, channels operate and the agreed calibration procedure completes. Record temperatures and mounting conditions. A successful boot does not establish detection performance.
- Installed sensor: evaluate the required targets over the specified positions and conditions with the released enclosure and processing settings. Preserve target geometry, distance, angle, clutter and calibration records. Define repeatability and uncertainty before judging a borderline result.
If a test fails, use the revision links to narrow the cause. A changed result after fitting the cover suggests a different investigation from a dimensional shift across a bare-board lot. Retain the failed configuration and its evidence before making changes; otherwise an apparent improvement may be impossible to reproduce.
7. Release a package that can be quoted and checked
Package the sensing worksheet, device/reference revisions, exported chirp configuration, antenna coordinates, stackup, RF-critical drawing, Gerbers or approved manufacturing data, BOM, mechanical model and test matrix together. Label provisional items. State who may approve material substitutions or RF artwork changes, and require a new revision when an approved boundary changes.
Request a radar PCB manufacturing review.
Send the sensor and reference design, operating band, antenna geometry, proposed stackup, mechanical envelope, quantity and acceptance plan. Identify which evidence is required from fabrication and which belongs to assembly or sensor testing. RFPCB can review the proposed build and provide a project-specific quotation.
Sources reviewed October 15, 2026. Figures are original engineering concepts and calculations, not product photos, simulation output or measured performance. The selected device documentation and approved fabrication, assembly and validation package govern the actual project.
Put this into production
Request a material, stackup and manufacturing review for your RF PCB project.