Feeds, speeds and the post
One number decides most cuts. Most machine failures are decided at the computer.
Green band: ShopBot's plywood range for a ¼ in cutter (0.13–0.5 mm/tooth), scaled to the cutter Ø. Too thin: rubbing, heat, burn. Too thick: deflection, chatter, snapped bit. Plunge at ≤ half the feed.
Used in Icelandic Fab Labs
6 mm two-flute unless stated. Record the material, cutter, feed, rpm and any fault in your job notes and report problems to staff.
| Material | Cutter | rpm | Feed mm/min | mm/tooth | Notes |
|---|---|---|---|---|---|
| Birch / spruce plywood | 6 mm 2-flute | 14 000 | 4 200 | 0.15 | Labs report 2000–6500 mm/min (0.07–0.23 mm/tooth); below ~3600 mm/min at 14 000 rpm you are under ShopBot's range. 3 mm pass, 40 % stepover; down-cut to 3 mm then up-cut to 0.2 mm past the thickness; tabs 12 × 3 mm. To go gentler, drop rpm, not feed.[11,56,57,58] |
| MDF | ¼ in | — | similar to plywood | 0.33–0.40 | Fine dust: extraction and a mask. MIT's table; plunge ≤ 50 % of the XY feed.[59] |
| EB700 epoxy tooling board | 12 mm 3-flute | 8 000 | 8 400 | 0.35 | Plunge 300 mm/min; adaptive 20 mm max / 5 mm fine stepdown. One worked job, not a validated recipe; respirator.[61] |
| PU foam / tooling foam | — | — | — | 0.28–0.33 | Cuts easily; the dust is the hazard.[59] |
| Plastics (HDPE, acrylic) | 1-flute (O-flute) | low | high | — | The test is a clean curl, not a weld.[11,35] |
| Aluminium | 1-flute | — | high | — | Marginal on a router: shallow depth, lubricant; if it welds, less depth and more feed.[35] |
Why chipload, with a worked example
The cutter removes a chip every time a flute passes. The thickness of that chip, the chipload fz, decides almost everything: too thin and the edge rubs, heats, dulls and burns; too thick and the cutter deflects, chatters and snaps.
Worked example (the calculator's default): 4200 mm/min ÷ (14 000 rpm × 2 flutes) = 0.15 mm per tooth, a reasonable chip for a 6 mm cutter in plywood. CAM works in mm/min; SB3 and the .sbp file use mm/s — divide by 60 (4200 mm/min = 70 mm/s).[11]
Ranges, sources and rules of thumb
Starting points used in Icelandic Fab Labs (6 mm two-flute unless stated)
- birch / spruce plywood
- Start at 14 000 rpm, 4200 mm/min (0.15 mm/tooth), 3 mm pass, 40 % stepover. Icelandic labs report 2000–6500 mm/min working (0.07–0.23 mm/tooth), but below ~3600 mm/min at 14 000 rpm you are under ShopBot's plywood chipload range (0.13–0.5 mm/tooth for a ¼ in cutter): if you need to go gentler, drop rpm, not feed.[11] down-cut to 3 mm then up-cut to 0.2 mm past the thickness, tabs 12 × 3 mm.[56,57,58]
- MDF
- similar feeds, expect fine dust: extraction and a mask. MIT's table: ¼ in cutter 0.33–0.40 mm/tooth, plunge ≤ 50 % of the XY feed.[59]
- EB700 epoxy tooling board, 12 mm 3-flute
- 8000 rpm, 8400 mm/min (0.35 mm/tooth), plunge 300 mm/min, adaptive 20 mm max / 5 mm fine stepdown — one worked job, not a validated recipe; respirator.[61]
- PU foam / tooling foam
- 0.28–0.33 mm/tooth; it cuts easily but the dust is the hazard.[59]
- plastics (HDPE, acrylic)
- single-flute O-flute, high feed, low rpm; the test is a clean curl, not a weld.[11,35]
- aluminium
- possible but marginal on a router: shallow depth, high feed, lubricant, single flute; lower depth and raise feed if it welds.[35]
Rules of thumb. Plunge at half the XY feed or less; ramp into pockets. Keep the bit's stick-out as short as the job allows. If depth of cut is 2× the diameter, reduce chipload 25 %; at 3× diameter, 50 %. A spindle loses torque at low rpm: if it bogs, raise rpm and feed together to keep the chip the same. Record the material, cutter, feed, rpm and any fault in your job notes and report problems to staff.[11,59]
From CAM to the machine
Most "machine" failures are decided at the computer. Go through this list every time you post a file. Post with Autodesk Fusion's official ShopBot (OpenSBP) post processor. The correct machine settings are already loaded on the control PC; setup is then checked by hand at the machine.[54,63]
- Units. File and software agree (mm). If SB3 stops with UNIT_ERROR, the file and SB3 disagree on mm/inch: re-post the file in mm. Do not change or reload the machine settings — they are already correct on the control PC; ask staff if the units still disagree.[10,54,63]
- Origin corner and Z reference. Which corner is X0 Y0? Is Z0 the material top or the table surface (
&PWZoriginin the posted file)? It must match what you touch with C2. Bed zero with a top-zero file puts every rapid and retract inside the stock and every cut a sheet thickness into the spoilboard; the reverse cuts air.[4,6,54] - Stock thickness measured, not assumed. Plywood sold as 18 mm is 17.4–18.6 mm; through-cuts and onion skins depend on it.[56]
- Clearance / retract height above the tallest clamp or screw head, with margin — and below the Z stroke.[10,12]
- Hold-down planned in the CAM: screw positions in the waste, screws and clamps clear of every toolpath including the tool's radius; tabs or a 0.25–0.75 mm onion skin for every part that will be freed. Plan the full cutter envelope in CAM and check it in the simulation.[12,63]
- Feeds, rpm, flutes, depth per pass computed, not copied; check the first-pass depth in the simulation and posted file before running.[11,62]
- Simulate and look for: straight plunges into solid stock, rapids posted as feed moves above the stock (slow and dangerous at clamp height), curves turned into thousands of tiny steps (the machine stutters) — ask staff about the post settings.[62]
- One file per tool unless you are sure of the tool-change prompts; name files with the tool and the date.[62]
- Open the .sbp in a text editor and read the header: units,
MSfeed,TRrpm,&PWZorigin, the firstJZandM3. If a number surprises you, it will surprise the machine.[10,62]