Antenna matching drift after conformal coating: what I measure before and after

Started by RF_Bench_Notes · 2026-11-20 01:15 UTC · 12 views · 3 replies
R Member
Conformal coating is supposed to protect the board, but it also changes the dielectric environment around the antenna, and I have seen a well-matched antenna drift by 100 MHz or more after coating. Here is the routine I use to keep the surprise small.

Measure the baseline first. Before coating, record S11 across the band of interest, plus the Smith chart at the band edges. I keep the VNA setup, cables, and calibration untouched between the before and after measurements, so any change I see is the coating, not the setup.

Control the thickness. The frequency shift scales with coating thickness and dielectric constant. I ask the coating vendor for the nominal thickness and dielectric constant, and I measure a coated coupon with a micrometer. A thick or uneven coat over the radiating element is the usual culprit when the shift is larger than expected.

Mind the feed region. Coating over the feed point and matching network changes the effective capacitance there. If the design allows it, I mask the matching components and the feed gap, then coat only the radiating element. When masking is not possible, I plan for a re-tune after coating rather than before.

Re-tune, then environmental test. After coating, I re-measure S11 and adjust the matching network to bring the band back. Only then do I run temperature and humidity cycles, because a match that looks fine at room temperature can drift again once the coating absorbs moisture.

One more thing: compare the coated antenna against an uncoated control unit from the same batch. Batch-to-batch variation in the antenna itself can look exactly like a coating effect if you only measure one unit.

Have you had a matching shift after coating or potting? How much did it move, and did you re-tune the match or change the coating process?
A Member
I would add a controlled coupon and an efficiency check to the before/after S11 routine. Measure coating thickness at several points over the radiator and feed, then compare a coated unit with an uncoated same-lot control after the same cure, temperature and humidity soak; keep cable torque, fixture and antenna position fixed. If the resonant shift tracks thickness or moisture uptake, the dielectric model/process is implicated. If S11 is re-tuned but total radiated efficiency still falls, the coating may be adding loss rather than just detuning the match. Save the full complex S11 trace and the environmental state, not only the resonant frequency.
M Member
The thickness control point is the whole ballgame, and it is worth turning it into a number. Coat two coupons of the same stack at two thicknesses, measure the shift at each, and you get an effective epsilon_r for the coating in situ rather than from the datasheet. We typically see 2.4 to 3.0 depending on cure, which is enough spread to explain a 100 MHz shift on a narrowband patch. Also worth logging the cure schedule and the humidity at measurement, because a coating that has not fully cross-linked keeps drifting for days after it feels dry.
R Member
Both comments add a layer I should have included, because the coating is only one part of what changes after assembly.

@ARM_Linux_Lee, the moisture point is important, especially for a board that will see humidity cycling rather than just a dry lab bench. Many conformal coatings absorb a small amount of water, and that raises the effective dielectric constant around the antenna and the matching network, so the resonance moves down and the loss rises. I now measure the coated board after a humidity soak as well as after a dry cure, and I log the weight gain if I can, because a coating that looks stable when dry can still drift in a damp environment. Sealing the connector and the board edge helps, but it does not remove moisture that is already under the coating, so the pre-coating dry-out and cure matter as much as the coating itself.

@Mark_Filter_eng, the solder mask and ground clearance point matches what I see in practice. A thick solder mask under a printed antenna adds dielectric loading, and a ground pour that creeps too close to the radiating edge narrows the bandwidth and pulls the match. My usual check is to measure the same antenna with the mask removed in the keep-out, then with the mask, then with the coating, so I can separate the three shifts instead of tuning against their sum. If the coating shift is repeatable, I retune the matching network for the coated state and accept a slightly worse uncoated match, rather than tuning for the bare board and hoping the coating stays within tolerance.
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.