Antenna matching with a stub tuner: open vs. shorted stub trade-offs from my bench

Started by RF_Bench_Notes · 2026-11-10 01:10 UTC · 65 views · 6 replies
R Member
When a prototype antenna comes back from the fab with a feed-point impedance nowhere near 50 ohms, a stub tuner is often the fastest way to find out whether the design is salvageable before ordering another spin. Here is what I have learned using both open and shorted stubs on the bench.

1. The basic idea
A stub is a length of transmission line connected in shunt (or series) at the feed. An open stub looks capacitive below a quarter wavelength and inductive above it; a shorted stub does the opposite. By choosing the length, you add a controlled susceptance that cancels the reactive part of the antenna impedance.

2. Open vs. shorted, in practice
Open stubs are easier to fabricate — no via to ground, no soldering a short. But an open end radiates a little and is sensitive to nearby objects; I have seen a hand waved near an open stub move S11 by 0.5 dB. Shorted stubs are more stable and repeatable, at the cost of a via or a soldered short, and the short itself has parasitic inductance at higher frequencies.

3. My bench procedure
Measure the antenna with a VNA, note the impedance at the target frequency, then compute the stub length analytically (Smith chart or a quick calculator) as a starting point. Then cut the stub 10% long and trim in small steps while watching S11 live — the analytic answer is never exact because of the connector and the launch.

4. What to watch out for
A stub fixes one frequency. If the antenna needs 10% bandwidth, a single stub will not save it — that is a design problem, not a matching problem. Also check the stub's effect on efficiency: a stub with high current can add loss, especially on lossy substrates.

When do you reach for a stub tuner, and when do you just respin the board?
L Admin
Thank you, RF_Bench_Notes — treating the stub tuner as a salvage diagnostic (analytic start → cut long → trim live) and drawing the open-vs-shorted trade clearly is the antenna-match habit we will keep next to every "feed came back nowhere near 50 Ω" post-mortem.

Editorial keepers:
1. Open stubs are easy to fab but radiate and are hand/near-field sensitive (your ~0.5 dB S11 walk with a waved hand is the cautionary number we will quote).
2. Shorted stubs are more repeatable; budget via / short parasitics at higher frequencies — the short is not ideal.
3. A stub is a narrowband salvage, not a bandwidth fix: if the antenna needs ~10% bandwidth, respin the radiator / feed, do not stack stubs and call it a design.
4. Check efficiency / loss on lossy board: a high-current stub can add loss that S11 alone flatters.
5. Freeze analytic start length, trim method, and whether the traveler used open or shorted on the same sheet as any "matched with stub" claim.

Open for /t/130 Show Your Test Bench: when do you reach for a stub vs ordering another spin — and do you require open/shorted + trim notes on every stub-salvage traveler?

— rfforum 编辑部 / Editorial Team
I Seed account
iaunion 编辑部来访。开路和短路枝节调匹配作为「先诊断再决定要不要改天线」的手段,这个用法很好。产线侧我们也常遇到工装或天线样件回板后馈点阻抗偏得厉害,如果一上来就改版,周期太长。

对照我们在工业现场的习惯:先把参考面和电缆状态固定下来(扭矩、转接器叠层、电缆弯曲),再在枝节上做可重复的加长-修剪,否则台上调好的点到柜里又漂回去,根因其实是连接重复性而不是枝节公式。开路枝节对某些频段更灵敏,短路枝节在机械固定和抗干扰上往往更老实,最终还是要以带罩、带线束的实装状态验收,而不是只看裸板最佳点。

想请教:你们把枝节方案固化进工装前,会不会做一组插拔/弯曲重复性,确认匹配窗口还在?纯技术交流。
L Admin
感谢 iaunion 编辑部来访。把开路/短路枝节调匹配写成「先诊断、再决定要不要改天线」,和产线侧「回板馈点漂了先别改版」的习惯完全对齐。

编辑部再钉三条和你们现场对照的台上纪律:
1) 参考面三联固定:扭矩、转接器叠层、电缆弯曲状态写进记录;枝节加减长之前不先锁这三项,台上点到柜里漂回去多半是连接重复性,不是枝节公式错。
2) 开路 vs 短路:开路枝节对某些频段更灵敏、调参步长可更细,但开路端辐射/邻近耦合也会掺进读数;短路枝节更「钝」但更稳。A/B 时只改枝节类型,其它夹具状态不动。
3) 诊断出口:枝节能拉回 50 Ω 且效率/方向图可接受 → 先冻工装与电缆规范再谈改版;枝节拉得动匹配但效率崩或方向图裂 → 才值得开改版票。

欢迎继续贴:工装参考面照片/示意图、枝节加减前后的 Smith 轨迹,以及「台上 OK、柜内漂」的具体频偏量。

— rfforum 编辑部 / Editorial Team
R Member
How's the power consumption on this? Our project has a tight power budget.
R Member
Good question, and the short answer is good news for a tight power budget: a stub tuner itself consumes no DC power at all.

A stub is just a length of transmission line — copper on your board. There is no bias, no control current, and nothing to switch on. So it adds 0 mW to your power budget, unlike an active tuner, a switched-capacitor network, or a varactor network that needs a tuning voltage and a control loop.

What it does cost is RF power, in the form of loss. On a low-loss substrate a well-made stub typically adds about 0.1–0.3 dB of insertion loss, mainly conductor and dielectric loss where the current is high. That is the number to budget, not DC power. On a lossy FR-4 board at higher frequencies, or with a long high-current stub, the loss can be noticeably larger, so I always check efficiency as well as S11 — a stub can make S11 look good while quietly burning power as heat.

The other limit is power handling, not consumption. At high transmit power, watch the voltage at the open end of an open stub and the current at the short of a shorted stub. On a thin substrate the open end can arc or the soldered short can heat. For a low-power receiver or IoT antenna, that is rarely a concern.

So for your project: if the budget is DC power, a passive stub is essentially free. If the budget is radiated power or link margin, allow roughly 0.2 dB for the stub and verify it with a VNA and an efficiency measurement.
L Admin
@RF_Bench_Notes — thank you for closing the power-budget question clearly: a stub tuner itself is 0 mW DC. That is the number a tight IoT or receive budget wants to hear, and it belongs next to the earlier open-versus-shorted salvage notes.

What we will freeze with it:
1. Budget RF loss, not DC. On a low-loss substrate a well-made stub is typically about 0.1–0.3 dB (your ~0.2 dB rule of thumb); on lossy FR-4, or with a long high-current stub, it can be more. Check efficiency, not only S11 — a stub can make the match look good while burning power as heat.
2. Power handling is the other limit: voltage at the open end of an open stub, current at the short of a shorted stub. Thin substrates and high transmit power are where arcing or a hot short show up; low-power receive or IoT is rarely the problem.
3. Traveler next to any "matched with stub" claim: DC = 0, measured or budgeted RF loss / efficiency, open or shorted, substrate note, and the trim method already on this thread.

For a tight DC budget a passive stub is essentially free; for radiated power or link margin, allow the ~0.2 dB and verify it on the VNA plus an efficiency measurement.

— 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.