PEEK machining tolerances: what a saw can and can't hold
Tighter tolerance only holds over a shorter span. That's physics, not a policy.
Ask a distributor for a tight tolerance on a large blank and you'll often get a quote back anyway - and a part that doesn't actually hold that tolerance once it's off the saw. The honest answer is that tolerance and span are linked, and a straight cut on a sliding table saw simply cannot hold a tight number over a long distance the way it can over a short one.
Three tolerance tiers are offered here, and each one comes with a real span limit, not a suggested one. Standard tolerance, at plus or minus 0.030 in, holds up to a 47 in span. Precision, at plus or minus 0.015 in, holds up to 36 in. Tight, at plus or minus 0.010 in, holds up to 24 in. Ask for tight tolerance on a 40 in part and the quote tool will refuse it and tell you why, rather than quietly quoting something it can't deliver.
Why does the limit shrink as the tolerance tightens? Two compounding reasons. First, blade deflection and feed-rate effects accumulate over distance - a saw blade that tracks within a hair over six inches can wander further than that hair over four feet, especially in denser or more abrasive materials. Second, thermal expansion is a real, physical number: a longer part expands and contracts more in absolute terms for the same temperature swing, and a tight tolerance has less room to absorb that movement before the part reads out of spec.
There's a third factor that's specific to a few materials in this catalog: residual stress. PEEK, Ultem, and PPS plate can move a few thousandths in the two days after cutting as internal stress releases - see the companion article on why these blanks warp. A tight tolerance on a stress-flagged material is exactly where that matters most, which is part of why the pricing engine treats residual-stress-flagged material differently on precision and tight tiers, and why annealing exists as an option at that point.
Squareness is the other half of this conversation, and it's just as span-dependent. Standard tolerance holds squareness to 0.010 in per 12 in of edge. Precision holds 0.006 in per 12 in. Tight holds 0.004 in per 12 in. If a design needs a part to sit flush against a perpendicular reference over a long edge, squareness - not just the linear dimension - is usually the number that actually matters, and it's worth checking against the same span limits.
None of this is a reason to avoid tight tolerance where it's genuinely needed - it exists because some parts really do need it, and the quote tool prices the extra passes and inspection time that holding it requires. It's a reason to be honest about span at the design stage: if a 30 in part needs plus or minus 0.010 in, the real answer is to either split it into two shorter pieces that each qualify, or accept the precision tier's slightly looser number over the full length.
The quote tool enforces all of this automatically - pick a tolerance tier, enter a dimension past its limit, and it tells you plainly rather than letting an order through that can't be held. That's a deliberate choice: a rejected quote with a clear reason is a five-second fix. A part that ships out of tolerance because a limit was never enforced is a much more expensive conversation for everyone.