Handing a placed board to someone else to route is an act of trust. Footprints were placed for mechanical reasons you may not have written down; the board outline matches an enclosure; the title block, fab notes and design rules were set on purpose. A routed board that comes back with any of that silently changed is worse than an unrouted one, because you may not notice until the boards arrive.
So every routed board we deliver can be compared with the board you uploaded, directly on the order page. Here is what that comparison checks and how it works.
What the comparison checks
| Question | How it is answered |
|---|---|
| Did any component move? | Every footprint is matched by reference designator and its position, rotation and side are compared (to 0.1 µm and 0.001°). |
| Did anything except copper change? | Every top-level element that is not routed copper — footprints, outline, graphics, text, nets, board setup and design rules, title block — is compared byte for byte with your file. Elements that differ only in formatting are reported as equivalent, anything else as changed. |
| What copper was added? | Tracks, vias and filled zones are compared as sets per layer: added length, number of segments, pours and anything removed. |
| How far did routing get? | Nets that carry tracks before and after. |
The result is shown as three verdicts at the top — for example Placement unchanged, Only copper was added: 50 of 50 non-copper elements are byte-identical and 189.0 mm of copper routed — above two linked 3D views of the board before and after, with the new copper highlighted. Both views share one camera, so you can orbit, zoom and inspect a net in either and see the same spot in the other.
Why byte-level, not “looks the same”
A visual comparison would miss the things that bite later: a changed solder-mask expansion on one footprint, a design-rule value, a text field in the title block. KiCad’s S-expression format is text, and a careful tool can rewrite a board while leaving every element it did not intend to change exactly as it was. That is the standard we hold our own tooling to — a lossless round-trip — and the comparison is how you verify it rather than take our word for it.
Reproduce it with Git
If your project lives in Git, you can see the same thing in a diff:
git checkout -b routed cp ~/Downloads/board-routed.kicad_pcb hardware/board.kicad_pcb git diff --stat git diff hardware/board.kicad_pcb | grep '^[-+]' | grep -v '^[-+][[:space:]]*(\(segment\|arc\|via\|zone\)' | head
The last command lists changed lines that are not tracks, arcs, vias or zones. Because zones and tracks span several lines, a handful of their inner lines may still appear, but footprint, outline or rule changes stand out immediately. On a board routed by us you should see new (segment …), (via …) and zone fills, and nothing else.
What it does not tell you
- It does not judge routing quality. That is what the DRC/DFM results, the constrained-net tables and your own review are for; the evidence dossier covers them.
- Filled zones are compared by outline and net. Refilling a zone in KiCad after other changes is normal and shows up as a changed pour.
- If you edited your own copy after ordering, compare against the file you uploaded — the order page always keeps that exact file.
You can try the Git recipe on our open demo board: the placed, unrouted file and the routed one differ only in tracks, vias and pours. Open either in the 3D Studio to inspect it.