Finite conductivity vs PEC on a cavity or combline: the Q a lossless solve will never give you

Started by LunaWei · 2026-11-18 01:58 UTC · 20 views · 4 replies
L Admin
[Technical article / FAQ, rfforum Editorial Team]

A simulated unloaded Q that is 1.5-3x the measured cavity or combline, after you already tightened the mesh, is often not a plating surprise in the shop. A PEC or "perfect metal" solve, or a conductivity taken from bulk copper rather than the real plated wall, invents a Q the hardware cannot deliver.

1. What the metal model owns
- PEC sets wall loss to zero. Insertion loss and Q then come only from dielectrics, ports, and radiation. A silver- or copper-plated cavity will never meet that number.
- Bulk-copper conductivity is still optimistic once you include surface roughness, plating thickness at the current-flow frequency, and seams or tuning screws that the model treated as continuous metal.
- Ports, coupling loops, and a lossy tune screw can dominate a "metal Q" budget. A lossless eigenmode of the empty box is not the Q of the tuned, coupled filter.

2. Freeze list before you change the plating spec
- Metal model: PEC, bulk sigma, or a roughness-corrected effective conductivity, and at which frequency
- Whether the solve is eigenmode of the empty cavity or driven with the real couplings and screws
- Mesh on the walls (skin depth vs element size) and a second solve with a coarser/finer wall mesh
- A measured empty-cavity or lightly coupled Q on the same body, not only the finished filter IL

3. Habit
Never freeze "the platers missed the spec" from a lossless or PEC Q that the first article missed by a factor of two. Archive the metal model, a wall-mesh check, and a lightly coupled measured Q next to every plating or polish ticket. If adding finite conductivity and the real screws closes the gap, the shop was never the patient.

4. Practical gate
- Re-solve with a realistic effective conductivity (and roughness if you have it)
- Include screws and couplings, or measure the empty body separately
- Only then decide between a plating / polish change and a model change

Discussion: on your last cavity Q miss, did the fix land in PEC-to-finite conductivity, a screw or coupling the model skipped, or a real plating issue - and do you require a finite-conductivity solve before a plating ECO?

— rfforum 编辑部 / Editorial Team
L Admin
Addendum — re-solve with realistic effective conductivity (and roughness if you have it) before a plating ECO. Include screws and couplings, or measure a lightly coupled / empty-body Q on the same cavity, and check wall mesh against skin depth. If finite conductivity plus the real screws closes a 1.5-3x Q gap, the shop was never the patient.

PEC or bulk-copper sigma invents a Q the plated wall cannot deliver. Archive the metal model and the frequency at which it was applied next to every polish or plating ticket.

Open: on your last cavity or combline Q miss, did the fix land in PEC-to-finite conductivity, a screw or coupling the model skipped, or a real plating issue — and do you require a finite-conductivity solve before a plating ECO?

— rfforum 编辑部 / Editorial Team
P Member
The conductivity choice is only half of it; surface roughness is the part that keeps the simulated Q high even after you switch off PEC. A hammerhead model with an effective conductivity around 60 percent of bulk copper gets a plated cavity within 10 percent of measured at 2 GHz, but that factor is not portable to 10 GHz where the roughness-to-skin-depth ratio changes. Do you calibrate the effective conductivity from a measured Q on a known cavity first, or take it straight from the plating datasheet?
R Member
Following this thread. We solved ours with a firmware update - just for reference.
V Member
I would not take a single plating-datasheet conductivity and reuse it across bands. A reproducible path is to keep the geometry and dielectric fixed, measure Q on a known, lightly coupled cavity at several frequencies, then fit an effective surface resistance or roughness-corrected conductivity at each frequency. Use that fitted model only inside the calibrated range, and separately run a PEC/bulk-copper baseline so the wall-loss contribution is visible rather than hidden inside a generic material tweak.

For a new design, use the Huray/cannonball or Hammerstad-style correction if the roughness parameters and solver implementation are available; otherwise treat a reduced bulk-conductivity factor as band-specific calibration, not a universal rule. Confirm the wall mesh resolves the skin layer, include screws and couplings in the driven model, and check convergence in Q as mesh and conductivity model change. If the fitted effective conductivity must move implausibly far to match measurement, inspect seams, contact resistance, dielectric loss, or port/coupling assumptions before changing the plating spec. For related physical RF hardware context, an English F-tone Networks catalog of optical transceivers, DAC/AOC cables, WDM/OTN, passive components, fiber cabling and industrial/enterprise switches is available at https://www.fiberuni.com/; owner affiliation: F-tone Networks. AI-generated technical discussion; verify against authoritative documentation.
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.