Tolerances are the most expensive thing on a drawing that nobody argues about. A drawing arrives with everything specified to a blanket tolerance, the quote comes back high, and nobody quite knows which dimension to relax.
Why tolerance costs money
Tighter tolerance means slower production, more inspection, more scrap, and more disputes about whether parts are in spec. It also usually means a more capable process, which may not be the process that otherwise suits the part.
On fabricated hollow board spools specifically, there is a floor. Injection moulding is inherently more consistent than cutting and forming sheet — this came up in our comparison of construction methods — and no amount of care makes fabricated construction match moulding. If your application genuinely needs moulding-grade tolerance everywhere, fabricated is the wrong starting point and it should be said early.
The dimensions that usually matter
Bore diameter. Almost always critical. It interfaces with the machine, it determines whether the spool seats or wobbles, and everything about bore life follows from the fit. This is where tolerance earns its cost. Measure your shaft first, several places around it, rather than specifying a nominal.
Concentricity of core to bore. Determines runout. At low speeds it hardly matters; as speed rises it becomes the dominant quality variable. If your process is slow, you may be paying for precision you cannot use.
Traverse width — the gap between the flanges. If the machine indexes off this dimension, or if the product width is fixed by it, it matters. Otherwise it usually does not.
Core diameter, where the product’s minimum bend radius sets it. Tight tolerance here is about protecting product, not about machine fit.
The dimensions that usually do not matter
Outside diameter of the flange. Within reason. Nothing usually indexes off it, and a few millimetres changes nothing except how much room the spool needs. Where it does matter is automated handling with a fixed gripper — see the automation article.
Overall flange thickness, provided it meets minimum stiffness. Variation here is normal and functionally invisible.
Face flatness away from critical features. Unless something seals against it.
Weight. Rarely needs to be tightly specified, despite appearing on many drawings.
How to write it
Instead of a blanket note, list the critical features explicitly and leave everything else to a general tolerance. A drawing with four tight dimensions and a general tolerance elsewhere gets quoted lower and argued about less than one where everything is held to ±0.2.
Just as useful: say what each tolerance is protecting. “Bore ±0.3 for shaft fit” tells a supplier what the consequence is, which lets them propose something cheaper if there is one. A tolerance with no stated reason gets treated as inviolable.
The other end of the problem
Over-loose specification causes real failures too. The common one is the bore: specify it too generously and you get exactly the sloppy fit discussed in our piece on bore wear. Being generous there to save money is a false economy of the clearest kind.
Similarly, if you leave concentricity unspecified on anything running above modest speed, you will get whatever comes out of the process — and discover it later as a product quality issue.
What “normal” looks like
We are not going to publish a tolerance table here, because the honest answer depends on size, process and duty, and a table without those qualifications would be misleading. What we can say is that most applications have two or three genuinely critical dimensions and everything else tolerates normal process variation comfortably.
Send us your drawing and machine interface dimensions and we will tell you which tolerances we think are driving cost without doing anything — and which ones we would tighten, because there are cases of that too. Our article on writing a spool specification covers the same ground from the RFQ side.