CNC router soundproof enclosure: noise and dust together

Short answer: build the shell from two decoupled 18 mm [0.7 in] plywood skins with a 40 mm [1.6 in] air gap, feed your vacuum's make-up air through a 4 in [100 mm] lined duct with two 90° bends, and capture chips at the spindle with a dust shoe drawing 50-80 CFM [85-135 m³/h]. One skin with a hose pushed through a bare hole buys you a few dB at best. The decoupled pair adds about 10 dB above 300 Hz, and the baffled intake protects that gain from the openings that would erase it. Keep the VFD, power supply and controller in their own sealed cabinet outside the shell, because their cooling fans are the highway dust takes into your electronics. The sizes, the airflow arithmetic and the checks that prove the box are below.

Where the sound gets out

Sound transmission loss measures how much sound energy a barrier blocks, and it climbs with surface mass and frequency. The mass law, the accepted single-panel approximation, adds about 6 dB of loss per doubling of mass or frequency. An 18 mm plywood skin at roughly 9.5 kg/m² [2 lb/ft²] works out to 20 × log10(9.5 × 500) − 47 ≈ 27 dB at 500 Hz. Real panels land a few dB below that around their coincidence dips, so carry 25 dB in your plans.

Openings undo walls. A bare hole with just 1% of the wall's area drives the number through the composite-wall formula: 0.99 × 10^-2.7 + 0.01 × 1 = 0.012, so the wall's loss falls to −10 × log10(0.012) ≈ 19 dB. The hole sets the number, not the plywood. Every penetration therefore gets a collar, a gland or a lined duct.

Decoupling is the other half of the design. Two 9.5 kg/m² leaves separated by a 40 mm [1.6 in] cavity have a mass-air-mass resonance near 140 Hz by the standard double-wall formula. Above roughly 300 Hz the pair beats a single skin by about 10 dB, and wool in the cavity adds a few dB more. Screw the skins together at even a few points and the bridge hands most of that back.

These are standard acoustic-enclosure design values in the 500-2,000 Hz band:

What you need

Build the shell step by step

1. Set the envelope. Start from the machine's footprint plus full travel, then add the motor overhangs before you buy plywood. Stepper bodies stick out past the frame on most routers, and they set the outer width more often than the table does. Add 4 in [100 mm] of clearance all round and 6 in [150 mm] behind for the hose sweep.
2. Cut both panel sets. Make the inner set smaller on every face so it stands 40 mm [1.6 in] clear of the outer shell all round.
3. Build the inner box first. Glue and screw it square, then bond 50 mm [2 in] faced mineral wool to its inside faces, facing toward the chamber so fibers and chips stay out of the wool.
4. Stack the boxes. Pack wool loosely into the 40 mm cavity, then set the inner box on rubber pads inside the outer. Keep every other contact between the skins resilient; the air gap only works if nothing solid crosses it.
5. Hang the door on the outer skin. Run 10 mm [0.4 in] EPDM tape around the door stop and place latches every 12 in [300 mm] so the gasket compresses 30-50% along its whole length. Your hand should feel the same resistance at every latch.
6. Glaze the window. Fit the 4 mm and 6 mm panes on separate EPDM strips, 50 mm [2 in] apart, and seal both perimeters. The thickness difference keeps the two panes from dipping at the same frequency.
7. Collar every penetration. Motor, limit-switch and probe cables pass through glands on a small plate bolted to the outer skin; the cutouts and glands guide covers matching a gland to cable OD.
8. Fit the intake and the hose collar. Put the intake low on the wall opposite the hose: two 90° bends and about 16 in [400 mm] of lined 4 in duct. The hose collar sits high on the roof or back wall.
9. Leave the support gear outside. The vacuum, its hose and the coolant reservoir all sit outside the shell. The vacuum is often as loud as the router, and parked outside it costs you nothing.

The airflow arithmetic

Size the intake duct at the vacuum's full free-air rating so it can never starve the shoe. The worked example uses 100 CFM [170 m³/h], a common mid-size wet/dry figure — replace it with the rating printed on your machine. Convert: 100 CFM = 0.0472 m³/s. A 4 in [100 mm] duct presents an area of π × (0.05 m)² = 0.0079 m², so the air moves at 0.0472 / 0.0079 ≈ 6.0 m/s [20 ft/s]. Ducts start to whistle audibly above about 10 m/s [33 ft/s] (standard duct-velocity practice), so you hold margin, and the two bends plus 16 in of lining strip 10-15 dB off whatever the duct carries (typical lined-duct attenuation tables).

The same airflow carries the heat out. Assume the spindle wastes 200 W to the air — a quarter of an 800 W air-cooled spindle; use your spindle's figure, and note that a water-cooled spindle sends most of its waste heat to the water instead — and let the four axis steppers add about 12 W each. That is 250 W inside the shell. The rise comes out as 250 W / (1.2 kg/m³ × 1005 J/(kg·K) × 0.0472 m³/s) ≈ 4.4 K [8 °F], so air leaves the box about 4 K warmer than it entered. With the vacuum off there is no purge at all. Run extraction whenever the spindle runs; the dust section shows why that habit pays twice.

The same lined-bend pattern keeps cooling air quiet in baffled enclosure airflow for noise reduction and cooling; borrow its baffle proportions if you want a shorter duct run.

Dust: capture at the shoe

Dust is a maintenance bill, not a housekeeping gripe. Chips that get past the bearing wipers mix with grease into a lapping paste, rails bind, and every fan inside the shell inhales whatever is airborne. Capture at the source: a dust shoe on the spindle drawing 50-80 CFM [85-135 m³/h] at the shoe, which a 100 CFM-rated vacuum reaches through 6.5 ft [2 m] of 1.5 in hose (common hobby dust-shoe practice). Judge your capture on MDF, not softwood; its fine dust escapes a weak shoe first.

With the shoe pulling and the intake feeding, the chamber runs slightly below room pressure, so leaks draw clean air in rather than push dust out. Keep the vacuum on for the whole program, including the passes you would rather skip. If the controller has a spare relay output, drive the vacuum from the job itself: running a dust-shoe relay from tool-change subprograms shows one wiring and code approach.

For the wider design space, CNC dust containment enclosure design covers extraction-first builds, and dust foot versus full enclosure on a desktop CNC helps when a full box is more machine than the job needs.

Keep the electronics out of the shell

Put the VFD, power supply and stepper controller in their own sealed cabinet on the table frame; the acoustic box is for the machine. Their cooling fans are the dust path into your drives, and their heat stays out of the plywood too: a 2.2 kW VFD losing 3% of rating sheds about 66 W you would otherwise have to move through the wall. Lay that cabinet out around duct and rail as in wiring duct and DIN rail in a CNC control cabinet.

If you would rather buy that cabinet than build it, our stainless wall-mount boxes are manufacturer-rated NEMA 4X-type / IP66; not UL listed, with a hinged gasketed door, removable mounting plate and wall-mount feet. The stainless enclosure size finder filters them by outside size from 6 x 4 x 4 in up to 60 x 36 x 16 in (H x W x D), and you place every cutout.

Hold the line on holes: no cooling fan punched through the acoustic wall to help the steppers, which are built to run hot. A fan opening behaves like the 1% hole above. If a sealed electronics cabinet ever truly needs cooling, see putting a fan on a NEMA 4X enclosure: what it costs you for the arithmetic before you cut.

Checks before you cut a part

Spindle stopped and at rest before any of these. Then:

• Airflow. With only the vacuum running, hold a strip of paper against the intake grille; it should stick, proving the make-up path feeds the shoe. Listen for a whistle: whistles come from leak edges, not from a clean 6 m/s [20 ft/s] duct.
• Leaks. Trace a smoke pencil along the door gasket, window edges and gland plate with extraction running. Smoke should curl inward; any outward stream is a dust exit waiting for your next job.
• Sound. A phone SPL app (C-weighted) at 1 m [3 ft], same program and stock before and after. A sealed double-skin shell typically delivers 10-20 dB at mid and high frequencies (standard acoustic-enclosure practice); under 200 Hz you will see far less, and that is the physics, not a fault.
• Dust. Run 30 min of MDF, then wipe a white card along each rail and check the wipers. A clean card means the shoe is holding.
• Heat. Air beside the spindle should read about 4-5 K [7-9 °F] above room with extraction running, matching the arithmetic above. A bigger rise means a dead corner: aim the intake at the spindle end.

Mistakes that cost you the quiet

• Bridging the skins. A screw through both leaves couples them; isolate everything crossing the gap, including the window frame and the door stop, which lives on the outer skin only.
• The bare hole. An open 4 in hole for the hose is the 1%-hole rule with worse manners. Keep the lined, bent intake and a collared roof entry.
• Foam-only walls. 2 in of foam inside one thin skin cleans up reverberation and adds almost no transmission loss. Mass and decoupling block sound; absorption only tidies the inside.
• A starved vacuum. Shut the intake and shoe capture collapses. The intake is part of the extraction system, not a convenience.
• Cutting with the vacuum off. One switch loses both the dust capture and the 4 K [7 °F] purge in the same instant.
• Partial gasket compression. Where a latch does not compress, dust puffs out on the next door opening. Add a latch rather than stretching the tape.

Catalog dimensions to compare

Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.

Questions