fdm: dielectric-side eps on metal faces + richardson-refine cli paths#144
Closed
UnsignedChad wants to merge 1 commit into
Closed
fdm: dielectric-side eps on metal faces + richardson-refine cli paths#144UnsignedChad wants to merge 1 commit into
UnsignedChad wants to merge 1 commit into
Conversation
…i paths Conductor cells store a placeholder eps_r that was harmonic-meaned into every metal/dielectric face coefficient, wrapping traces in a fake low-eps skin half a cell thick. eps_eff converged ~25% low at the default cell size and the fdm/closed-form engines disagreed 15% on the demo board. Metal faces now take the dielectric cell's eps alone (stencil + Gauss surface stay consistent), production impedance paths use the existing richardson-refined solve, and a regression test pins eps_eff against Hammerstad. Engines now agree to 0.1% on demo_board USB_DP.
Owner
Author
|
landed on main directly as dabc979, closing |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
conductor cells store a placeholder eps that was getting harmonic-meaned into every metal/dielectric face coefficient, so every trace wore a fake low-eps skin half a cell thick. eps_eff came out ~25% low at the default cell size and fdm disagreed with the closed form by 15% on the demo board.
fix: metal-adjacent faces take the dielectric cell eps alone, in both the SOR stencil and the gauss-law charge integral so they stay consistent. the cli impedance paths now call the richardson-refined solve that was previously test-only. diff-pair path gets the same h/2h extrapolation with a fallback if it goes non-physical.
new regression test pins eps_eff against hammerstad (old code lands ~2.4 and fails it). result on demo_board USB_DP: fdm 74.77 down to 64.87 ohm vs closed-form 64.91. stripline net tightened to 2.5% cross-engine.
also corrected the CrossSection.h comment that claimed neumann side walls. solve() leaves the boundary ring at V=0 so the walls are grounded dirichlet. a too-tight box is now the dominant remaining error source.
heads up: fdm Z0 shifts up to ~15% vs the old biased numbers. anything tuned against the old output moves with it.
716 tests pass.