
When tight tolerances save money – and when they just burn it.
An IT7 tolerance instead of IT11 can double the unit price – with no benefit to function. Where tight tolerances make sense and where they only cost money.
Understanding Tolerances: What Tight Tolerances Really Cost
Tolerances are one of the few design features that translate directly, one-to-one, into unit price. If you tolerance tightly across the board when drafting, you pay a premium that often serves no technical function. This article breaks the topic down from the perspective of a CNC contract turning shop with 35 machines and 77 years of practice.
On a typical rotationally symmetric turned part, the required tolerance class determines how many machining steps, which tools and how much inspection effort are needed. An ISO general tolerance medium (DIN ISO 2768-m) can be achieved with a standard tool, a standard spindle speed and a sample-based inspection scope. An IT7 fit on a functional diameter can double the same machining step – roughing plus a finishing cut with a defined tool, plus measuring equipment with finer resolution and a tighter inspection interval.
The price difference between DIN ISO 2768 medium and an IT7 fit is, in our experience, between 30 and 80 percent on the diameter in question. If the tight tolerance is functionally justified (fit, sealing function, bearing seat), the money is well invested. If it is there out of habit or as a blanket entry in the drawing template, it is money burned.
A second hidden cost source is geometric tolerances – roundness, cylindricity, concentricity. On a standard Swiss-type turning center, a roundness of 3 µm is state of the art. If you demand a roundness of 1 µm, you are asking for a significantly more elaborate clamping setup, possibly a downstream grinding operation and a different inspection station. Unit price: often more than doubled.
Our recommendation to design engineers: distinguish very deliberately between functional diameters and purely geometric diameters. Functional diameters need the tight tolerance – usually 2 to 4 dimensions per part. All other dimensions can live with general tolerances. This is exactly the differentiation we regularly work out together with our regular customers in the first article review – without any loss of function.
A third lever is surface finish. Ra 0.8 is sufficient for most applications and reproducibly achievable with standard turning tools. Ra 0.4 requires special tools, slower feed rates and tighter inspection intervals. If you demand Ra 0.2, you usually end up grinding – that is a different process and costs accordingly.
In practice, we see optimization potential in the tolerancing alone in more than half of our first-time inquiries. A 30-minute meeting between design and manufacturing frequently saves 15 to 25 percent in unit costs on a series of 10,000. That is money that often has the biggest leverage in the early phase of a project.
In assemblies, individual tolerances add up along the dimension chain: with five dimensions of ±0.05 mm each in series, the closing dimension can vary by up to ±0.25 mm in the worst case. To secure the functional dimension, engineers therefore often tighten every individual dimension to IT7 or IT6, even though only the overall dimension matters. This effort frequently arises purely from sloppy datum dimensioning: if dimensions are chained instead of referenced from a common datum, the tolerance multiplies unnecessarily. A well-thought-out datum structure defuses the chain and noticeably reduces manufacturing effort.
How tightly a dimension can be manufactured depends on process capability. Indicators such as Cpk and Cmk describe how reliably the scatter stays within the tolerance; a Cpk of 1.33 is widely regarded as the minimum target. Tightly defined dimensions usually require tighter inspection, and this is exactly where costs arise. A simple sample check is inexpensive, ongoing statistical process control (SPC) with documented records makes the part more expensive, and a required 100 percent inspection of tight dimensions hits hardest. Specify tight limits only where the function truly demands them.
With µm tolerances, the measurement itself becomes a factor. The standard-compliant reference temperature is 20 °C per DIN EN ISO 1. Stainless steel 1.4404 expands by around 16 µm per meter and kelvin, aluminium by about 23 µm; even a few degrees of deviation shift the measurement result by micrometers. A dimension of ±5 µm cannot be reliably inspected in a warm shop because workpiece and measuring equipment react differently. Such tolerances require a temperature-controlled measuring room and temperature equalization of the parts. Take this effort into account from the outset when specifying tolerances.
The key takeaways.
- 01Separate functional diameters (tight tolerance needed) from geometric diameters (general tolerances sufficient).
- 02IT7 vs. DIN ISO 2768 medium typically increases the unit price by 30–80 % on the diameter in question.
- 03Geometric tolerances (roundness, concentricity) are often the most expensive cost drivers – demand them only where functionally necessary.
- 04Surface finishes below Ra 0.4 require special machining or grinding – and cost accordingly.
- 05A 30-minute DFM review with your supplier often pays off with 15–25 % savings in unit costs.
FAQ on this topic.
Which ISO general tolerance do you use by default?+
Up to which tolerance class do you manufacture without extra effort?+
Can you carry out a DFM analysis before series start?+
How do tight tolerances affect delivery time?+
Can you inspect and document geometric tolerances?+
What does a first article inspection report cost?+
Your drawing on our desk.
We check feasibility, suggest optimizations and get back to you within 48 hours with a first assessment.