Docs · post processors

Generic ISO G-code

The generic ISO / RS-274 post for the routers most small shops actually own — Mach3 and Mach4, GRBL 1.1, LinuxCNC and Masso.

Awaiting a first cut

Built, simulated and verified against its own re-read — but no machine has cut one of these yet. Run it on scrap first.

File
ISO / RS-274
Controls
Mach3 · Mach4 · GRBL 1.1 · LinuxCNC · Masso
Machines
Avid CNC · Onefinity Elite (Masso series) · Shapeoko · LongMill · most imported 4×8 routers

What the post writes

  • G0 / G1 / G2 / G3 with X / Y / Z, a modal feed, and a spindle and tool word on the positioning move.
  • Arc centers as incremental I / J — with an explicit G91.1 for Mach and LinuxCNC, and never one for GRBL, which does not accept it.
  • Units per machine card: G21 millimeters or G20 inches, with every word and every feed converted rather than relabeled.
  • Tool changes as T# M6 (plus G43 H# where the control wants it), or an M0 pause that names the bit by hand — the machine card chooses.
  • Bores by capability: a bit of the same diameter that is flagged to drill runs G83 pecks (expanded to G0/G1 pecks where the control has no canned cycles); a wider hole — a 35 mm hinge cup, an 8 or 10 mm system hole — is helix-bored with the largest suitable smaller bit.
  • Lines kept to 78 characters, because GRBL’s receive buffer is 80.

Conventions chosen, and why

Where the public description of this format left a token ambiguous, the conservative reading was taken — the one a mistake cannot bury a bit under. Each is written into the program's own header too, so it can be read on the machine before a scrap cut. Check these against your own setup sheet.

  • Z zero sits at the TABLE by default — spoilboard top — so the stock top is at +DZ and a through cut ends 0.127 mm into the spoilboard. That is the deliberately safer of the two conventions: with Z0 at the table, a mistyped stock thickness cannot bury a bit. Z0 at the stock top is a machine-card option if that is what your shop runs.
  • When nothing in the tool library can make a hole, the sheet refuses and names the bit to add. It does not substitute a bit and hope.

Where this program comes from

WoodWright has one toolpath planner. The Xilog generator plans the program, the simulator clears it, and every post below translates that cleared program op for op — never a second planner with a second set of rules. Whatever your machine reads is the same toolpath every other machine reads.

Every file this post writes is read back by the post’s own parser and compared to the source program move for move — same XY, same depth, same order, every bore accounted for. A translation that does not survive its own re-read is not a program, and is not offered.

Written from public format documentation only. No vendor post-processor was copied, decompiled or referenced.

Questions

Straight answers

Which controls does the ISO post cover?
Mach3 and Mach4, GRBL 1.1, LinuxCNC and Masso — which between them cover most of the routers a small shop actually owns, including Avid (Mach4), Shapeoko and LongMill (GRBL) and the Onefinity Elite Masso series. Pick the card for the control software your machine runs, not the brand. They all read the ISO dialect; the differences that matter are arc-center form, unit words, tool-change form and buffer length, and the machine card carries each one.
Where is Z zero, and why does it matter?
At the table — the spoilboard top — by default, with the stock top at plus the measured thickness and a through cut ending 0.127 mm into the spoilboard. That is the safer of the two conventions: if the stock thickness is typed wrong with Z0 at the table, the cut is shallow or the spoilboard takes it, whereas with Z0 at the stock top the same mistake drives the bit through the table. Z0 at the stock top is available on the machine card if that is what your shop runs.
What happens if my tool library cannot make a hole?
The sheet refuses and tells you which bit to add. It does not quietly substitute a nearby diameter. A hinge cup drilled 3 mm undersize is a door that does not hang, discovered after the panel is cut.
Has this post cut on a real machine?
Not yet. It is planned, simulated, emitted, and then parsed back and compared to the source program move for move — but no shop has run one on iron. Cut on scrap first and check the first program against the drawing. When a shop does run one and reports back, this page changes.

Other controllers: Layered DXF · ShopBot OpenSBP · Homag / Weeke woodWOP · Biesse CIX · SCM Xilog