Skip to content 
You specified a decorative screen, the fabricator came back saying the pattern will not hold at full panel height, and nobody warned you about minimum web widths. This is a common gap, and it usually surfaces late, after the pattern is already in the drawing set and the schedule has no room for a redesign.
The short answer is that there is no single published table of CNC screen tolerances you can cite, because the limits depend on the material, the panel height, how the panel is supported, and how open the pattern is. What follows is how to reason about those limits before the pattern is committed, and what to ask a fabricator for while the design still has room to move.
Why There Is No Universal CNC Screen Tolerance Table
A decorative screen is a structural element pretending to be a graphic. The moment you cut a pattern into a panel, the remaining material, the webs, have to carry the panel’s own weight, resist handling during installation, and stay flat over the full height without visible bow or racking.
The North American Architectural Woodwork Standards, administered in Canada by AWMAC, cover grade, joinery and finish for architectural woodwork. They do not set out tolerances for a decorative CNC screen that tell you the minimum web width for a given pattern at a given height. That absence is not an oversight so much as a reflection of how many variables are in play. A web width that is fine in 19mm plywood at 1200mm tall is a problem in 12mm MDF at 3000mm tall.
So the reference you are looking for is really a method, not a number.
What Actually Drives the Limit
Several things move the structural limit at the same time, and they interact.
The core material matters first. MDF, plywood, particleboard, solid wood and aluminum each behave differently under their own weight and under routing. MDF cuts cleanly and takes a painted finish well, but it sags over long unsupported spans and its cut edges are weaker than its faces. Plywood holds a narrower web because of its cross-laminated structure, though voids in cheaper cores will show up as broken webs at the router. Aluminum lets you go far thinner than any wood-based sheet, which is why deep, fine patterns at full height are often not a millwork panel at all.
Panel height and unsupported span are the next factor. A screen fixed only at top and bottom behaves very differently from one captured in a frame on all four sides or split by an intermediate rail. Adding a horizontal support at mid-height can rescue a pattern that will not stand on its own, and it is usually cheaper than reworking the geometry.
Open area is the one designers underestimate most. A pattern that removes seventy percent of the panel has far less material left to do the structural work than one that removes forty percent, even if the individual web widths look identical on the drawing. The connectivity of the pattern matters too: continuous webs that run the full height carry load, while a field of isolated shapes joined by short bridges concentrates stress at those bridges and cracks there first.
Finally, the cut geometry itself sets a floor. A router bit has a diameter and cannot cut an inside corner sharper than its radius, so very fine patterns with tight internal corners either will not machine cleanly or will leave a fillet you did not draw.
The Honest Version: Sometimes the Answer Is a Different Product
Here is the part most suppliers will not put in writing. If your pattern needs very narrow webs at full room height with a high open area, a wood-based panel may be the wrong material regardless of how it is detailed. A laser-cut or waterjet-cut aluminum or steel screen will hold geometry that MDF or plywood simply cannot, and pretending otherwise on the millwork side only produces a deficiency list at handover.
We make cabinetry and architectural millwork locally, so when a decorative screen is a metal fabrication problem rather than a millwork one, the useful thing we can do is say so early, while there is still time to reassign the scope, rather than accept the package and fail at installation. That is worth more to your schedule than a fabricator who says yes to everything.
How to Specify a Screen That Will Actually Build
Before the pattern lands in the construction documents, get the geometry checked against the real constraints. The workflow that avoids the late RFI looks like this:
- Send the pattern, intended material, panel dimensions and support conditions to the fabricator as a design-assist request, not as a finished submittal.
- Ask for the minimum machinable web width in that specific material and thickness, and the maximum unsupported span before intermediate support is required.
- Confirm whether the internal corner radius the router leaves is acceptable, or whether the pattern needs to be redrawn to suit the bit.
- Decide support conditions early, because a mid-height rail or a full perimeter frame changes what web width is achievable.
Doing this during design development costs a review cycle. Doing it after the pattern is issued for construction costs an addendum, a resubmittal, and often a redesign of the visible feature everyone has already approved. What moves the number here is not the screen itself so much as when you ask the question.
The One Thing Worth Doing Next
Treat any decorative CNC screen as a design-assist item and run the geometry past a fabricator before it is issued for construction. Send the pattern, material and height, and ask for minimum web width and maximum unsupported span in writing. If you want that check on a millwork screen for a project in the Edmonton area, start a conversation with us while the pattern is still editable.
