PRODUCTION NC (1/8" steel) — slug tabs banded 0.018 small / 0.033 large, none under the
machine's 3/4" drop-through; start point (= tab position) on a flat, never a corner, circles at 12:00;
cut order sequenced so the head avoids rapiding over pieces already cut and still tab-held.
Tabs: 2 small,
0 large,
19 omitted. Every slug is numbered in the CSV — report failures by number.
The material string MSO7,0.125 is a GUESS and keys the condition table: confirm it at the machine before running.
Healed DXF = cut geometry only (one layer, closed contours — this is
the file for QC or the machine). Bridge-review DXF = the same geometry plus mark-up
on separate CM_* layers (rings, measured web, proposed width, notes) — reading material,
not cut paths. Mark-up only is that annotation on its own, for opening alongside the
cut file if your CAD's layer handling gets in the way. All R2000-2002 (AC1015), $INSUNITS = 1 (inch).
Download links are versioned, so you always get the current build, never a cached one.
original (source)healed linehealed arcfit-induced jog⭕ red circle = QC anomaly (open contour / duplicate)
Raster/component test at 20 thou/px on 60 thou material:
every candidate web is cut in the material raster and the part re-labelled — a piece
that separates only because of those cuts was being held by them, so it is a
real island and those webs are its bridges. Everything else that is narrow is spacing.
36 narrow webs · 12 real bridges holding
10 islands · 0 islands with no bridge at all.
8 bridges at or above the table — left alone (client's file, in spec).