RF PCB stackup for 77 GHz radar - material choice?

Started by NVIDIA_GPU_David · 2026-11-09 05:02 UTC · 57 views · 4 replies
N Member
77 GHz radar boards punish the wrong material choice. What stackup and material do you use, and what did the loss measurements show?
L Admin
77 GHz 雷达板材,错选一次就够痛。编辑部观察到的可复现点:

1) 介质:常用低损耗碳氢/PTFE 混压(如 RO3003 类或其对标),薄芯 + 连续地;不要用通用 FR4 硬扛毫米波走线。Dk/Df 用供应商实测 stackup,而不是目录标称。
2) 工艺:线宽/线距公差、铜箔粗糙度、阻焊开窗对插损和相位都很敏感;同一设计换厂后 S21 差几 dB 并不罕见。
3) 验证顺序:先单线/差分校准结构量插损与相位,再上天线阵列;via 过渡和过渡波导/连接器单独做 coupon。

欢迎贴叠层厚度、材料牌号、走线长度、插损 dB/cm,以及换材料前后对比。有 77 vs 79 GHz 同板差异故事也好。

— rfforum 编辑部 / Editorial Team
R Member
For 77 GHz radar, the material choice is really a loss and tolerance budget, not a brand choice. A thin low-loss hydrocarbon or PTFE-based laminate with a stable Dk around 3.0 and low Df is the usual starting point, with the RF traces on a thin core bonded to a thicker, cheaper FR-4-like mechanical layer below a solid ground plane. That hybrid keeps the RF layer thin enough for controlled impedance and low dispersion while keeping the board affordable and stiff. A full PTFE board is easier electrically but harder to handle in fabrication, so I only use it when the routing is long or the loss budget is very tight.

The numbers that mattered most in our measurements were not the datasheet Dk and Df, but the effective values at 77 GHz on our own test coupons, including copper roughness. Rough copper can add a noticeable fraction of a dB per centimetre at this frequency, and a shop that changes foil type or finish without telling you will move both loss and phase. I now ask for the measured Dk/Df and the foil profile on the actual stackup, and I measure a short and a long thru line on every panel so I can separate the connector and launch loss from the trace loss.

One more thing to check early: the via transition to the antenna or to the chip. A via that looks fine at 10 GHz can be a real impedance discontinuity at 77 GHz, and it will dominate the loss if the stackup is otherwise good. What trace lengths are you running, and have you measured insertion loss per centimetre on a coupon from your fab yet?
R Member
Interesting direction, will keep watching. Please update if there's progress.
N Member
Interesting direction, will keep watching. Please update if there's progress.
广告 AdRenesas 瑞萨微控制器芯片产品照片Renesas Electronics微控制器、模拟与射频/时钟方案 · MCUs, analog, RF and timing访问官网 →免费展示 · 资料来自公开信息 · 如需撤下请联系 ads@rfforum.com
广告 Ad广告位招商 / Advertise hereads@rfforum.com · 刊例 Rate cardT2 · 728×90
Log in or register to reply.