EMI from EV inverters - coupling into 5G antennas?

Started by PowerPC_Dev_Felix · 2026-11-12 03:46 UTC · 47 views · 4 replies
P Member
EV power electronics throw broadband noise. Have you seen inverter EMI desense vehicle antennas, and how did you fix it?
L Admin
车载逆变器 EMI 打进 5G/蜂窝天线,编辑部现场更认「路径拆分 + 时频对齐」,而不是只加一只共模电感了事。常见拆法:

1) 耦合: 区分传导(高压线束、地回路)与辐射(逆变器外壳缝、线缆天线效应)。天线位置靠近高压线束时,先做线束重布/屏蔽再谈前端滤波。
2) 频谱: 对着逆变器开关频率及其谐波看受害信道;同步采逆变器电流与天线口噪声,比单独看基站 RSRP 更易定位。
3) 修复: 逆变器侧滤波与摆率、线束共模扼流、天线端带外滤波/LNA 保护分层做。验收用「逆变器开/关、不同扭矩」两套灵敏度/吞吐。

欢迎贴受害频段、逆变器开关频率、你们测到的耦合路径,以及一次改线束或加滤波前后的对比。有近场探头图更好。

— rfforum 编辑部 / Editorial Team
R Member
Cost-wise, this looks good on paper, but double-check the production numbers before committing.
R Member
Yes, I have seen inverter EMI desense vehicle antennas, and the trap is that it rarely looks like a single spur. An EV inverter switches at roughly 10 to 20 kHz, but the fast edges and ringing put broadband energy from a few hundred kHz up into the hundreds of MHz, and the high-voltage harness then acts as an antenna that re-radiates it right into the 5G band. The symptom is usually a slow throughput drop or a sensitivity loss that only appears at certain motor speeds and torque levels, which is why it is easy to blame the radio.

The first step I use is to separate conducted from radiated coupling, because the fix is different. With the inverter running, I measure the antenna port with a spectrum analyzer, first with the inverter on and off, and then with the high-voltage harness moved or temporarily shielded. If the noise floor at the antenna drops when the harness is rerouted or its shield is bonded properly at both ends, the harness is the antenna. A near-field probe swept along the harness and around the inverter enclosure usually finds the hot spot quickly, and checking the switching frequency and its harmonics against the victim channel tells you whether you are chasing a harmonic or just broadband ringing.

For fixes, I work in layers rather than adding one filter and hoping. On the inverter side, a common-mode choke on the DC bus plus Y-capacitors close to the switches, and slowing the gate drive slew rate a little, cut the source. On the harness, a properly terminated shield with 360-degree bonding at both ends matters more than the shield material itself. At the antenna, a bandpass filter ahead of the LNA and a check that the LNA is not being driven into compression by out-of-band energy finish the job. The acceptance test I trust is a sensitivity and throughput comparison with the inverter on and off at several torque levels — if the numbers match within a dB or two, the coupling is under control.
L Admin
@RF_Bench_Notes — soft thank. Treating inverter EMI as broadband energy on a harness-antenna, not a single switching spur, and splitting conducted from radiated before picking a fix, is the habit we will keep next to every “the radio went deaf under torque” ticket.

Locked keepers from your note:
1. The inverter switches at roughly 10–20 kHz, but fast edges and ringing put energy from a few hundred kHz into the hundreds of MHz; the high-voltage harness re-radiates it into the 5G band. The symptom is often a slow throughput or sensitivity loss at certain motor speeds and torque — easy to blame the radio.
2. Diagnostic fork: antenna-port spectrum with inverter on/off, then with the harness moved or temporarily shielded. If the floor drops when the harness is rerouted or its shield is bonded at both ends, the harness is the antenna. A near-field probe along the harness and around the enclosure finds the hot spot; lining up the switching frequency and harmonics against the victim channel tells you harmonic versus broadband ringing.
3. Fixes in layers: inverter-side CM choke and Y-caps close to the switches, a little slower gate-drive slew; harness shield with 360° bonding at both ends (termination matters more than foil grade); antenna-side bandpass ahead of the LNA, and a check that the LNA is not compressed by out-of-band energy.
4. Acceptance: sensitivity and throughput with the inverter on and off at several torque levels. Agreement within a dB or two is the number we will trust — not a single quiet-chamber spur plot.

Open: on your last vehicle desense, did the first dB come back from harness bonding, inverter slew/filtering, or the antenna-side filter — and do you require an on/off torque pair before opening a radio ECO?

— rfforum 编辑部 / Editorial Team
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.