lines module#
The 50 ohm line’s layers, tags and rules. The run that produced this board also drew them on a launch-thru calibration fixture (two jacks back to back), from these same functions, so its line, gap, fence and stitching were the board’s by construction.
Rules (line_rules; main adds its branch rules to them):
[2]
RFLineTagon a signal net appliesfenced_rs50(), a copy of the substrate’sRS_50with a one-row via fence on each side (the substrate class is not ours to edit). The copy keeps RS_50’s width, velocity and loss.[2]
(RFLineTag, CoplanarGroundTag)sets the coplanar gap to the top pour. The runtime stands a pourPOUR_GAP_OFFSET(0.01 mm, measured inlayout_cpwg_two_row_fence) past the winning clearance, so the copy and this rule ask for 0.12 mm and the realized gap is RS_50’s 0.13 mm.[3]
(LaunchPadTag, CoplanarGroundTag): the pour standsLAUNCH_PAD_GAP(the line’s 0.13 mm) off each jack’s centre pad.[4]
ExitNeckTagon each exit’s neck route appliesneck_structure()(EXIT_NECK_WIDTHin the line’s slot), and(ExitNeckTag, CoplanarGroundTag)its gap, beating the net’sRS_50and line gap.StitchTagon the top pour: aStdViagrid stitching it to the bottom plane.
- RF_LAYER = 0#
the RF lines and the coplanar ground.
- Type:
Top copper
- PLANE_LAYER = -1#
the solid ground plane.
- Type:
Bottom copper
- FLOOR_MARGIN = 0.02#
Hold-off above a fabrication floor wherever copper is placed at one, mm: the export draws curves as chords up to the realization tolerance (0.02) inside the true arc, so copper drawn exactly at a floor can measure under it.
- FENCE_PITCH = 1.0#
at most 1.5).
- Type:
Fence via pitch along each 50 ohm line (the run’s fence assumption
- STITCH_PITCH = 1.5#
General stitch grid pitch over the top pour.
- STITCH_INSET = 0.25#
Stitch via pad edge to the computed top pour boundary.
- LAUNCH_PAD_GAP = 0.13#
the line’s own gap. Launch-tune S1 tried 0.33 mm; the S0/S1 contrast placed that change on the pad (27.9 ps from the thru’s centre, the pad’s 29 ps) at -0.142 pF per launch, but the baseline’s excess C sits nearer the line (8 to 19 ps), so the pad compensation helped 2.4 GHz and hurt 5-6 GHz. Launch-tune S2 restores the line’s gap here and compensates on the exit (
EXIT_NECK_WIDTH).- Type:
Realized gap from each jack’s centre pad to the top pour, mm
- EXIT_NECK_WIDTH = 0.38#
the exit’s last
EXIT_NECK_LENGTHbefore the launch point, inside the line’s own slot (pour edge unchanged at 0.285 + 0.13 = 0.415 mm off the axis, so the neck’s gap is 0.225 mm). A 2D quasi-static solve (0.762 RO4350B, the export’s mask and gap fill) gives the neck 67 to 69 ohm and a net excess (C - L/50^2) of -61 to -65 pF/m against RS_50, -0.116 pF over its 1.83 mm of narrow copper (-0.142 with the 1.23 ratio the S0/S1 contrast measured against the 2D pad solve). The S0 thru’s first-order (Born) response is best cancelled by -0.12 to -0.13 pF spread 9 to 19 ps from the thru’s centre, which is where the narrow copper sits (9.6 to 19.5 ps).- Type:
Trace width of the exit’s neck, mm
- EXIT_NECK_LENGTH = 2.4#
from 0.60 mm past the pad’s inner end to the launch point (3.0 mm exit), x 2.57 to 4.97 from the board edge. The full-width routes’ end caps (half of 0.57 each) reach into it, so the copper is neck-wide over about 1.83 mm, x 2.86 to 4.69.
- Type:
Neck route length, mm
- class CoplanarGroundTag[source]#
Bases:
TagThe top ground pour, which the coplanar gap is measured to.
- class ExitNeckTag[source]#
Bases:
TagThe neck
Routeof each exit: it takesneck_structure()and the neck’s coplanar gap.
- class LaunchPadTag[source]#
Bases:
TagEach jack’s centre pad: the top pour keeps
LAUNCH_PAD_GAPfrom it.
- corridor(a, b, half_width)[source]#
The band
half_widtheither side of the segment fromatob, flat at both ends.
- via_keepoff(centre, via)[source]#
A via keepout round a code-placed via: a generated via is kept out when its pad touches it, so the disc’s radius is the drill plus
min_hole_to_holeless the generated via’s pad radius, which leaves the two drillsmin_hole_to_holeapart.
- gap_rule()[source]#
The clearance the structure copy and the gap rules ask for: the runtime stands a pour
POUR_GAP_OFFSETpast the winning clearance (layout_cpwg_two_row_fence, measured), so they ask for that much less and the realized gap isRS_50’s own.- Return type:
- launch_pad_gap_rule()[source]#
The clearance the pad’s gap rule asks for:
LAUNCH_PAD_GAPless the runtime’sPOUR_GAP_OFFSET(asgap_rule).- Return type:
- rs50_fence_row()[source]#
The line’s fence row from its trace edge, mm: the via pad edge at the coplanar gap plus the minimum copper space, and the drill at least
min_copper_hole_spacefrom the trace edge, each plusFLOOR_MARGIN.- Return type:
- neck_gap()[source]#
The neck’s realized coplanar gap, mm: the line’s slot edge less half the neck width, so the pour edge does not move.
- Return type:
- fenced_rs50()[source]#
RS_50with one fence row each side: the first row’s via pad edge at the coplanar gap plus the minimum copper space, and its drill at leastmin_copper_hole_spacefrom the trace edge, each plusFLOOR_MARGIN.- Return type:
- neck_structure()[source]#
The exit neck:
EXIT_NECK_WIDTHwide atneck_gap(), fenced on the line’s own via row (offset from the neck’s edge by the width it gives up, so the row stays where the line’s is). Its 0 ohm is a label that keeps impedance lookups onRS_50, not an impedance.- Return type:
- line_rules(structure, gnd)[source]#
The line’s rules (module docstring): the fenced structure on
RFLineTag, the coplanar gap toCoplanarGroundTagfrom the line, the launch gap from eachLaunchPadTagpad, the neck structure and gap on eachExitNeckTagroute, and theStitchTaggrid.- Return type:
- Parameters:
structure (RoutingStructure)
gnd (Net)