CNC Lathe Machining Accuracy: Where the Error Actually Comes From
This page explains what moves a turned diameter off nominal: thermal growth, tool wear, spindle error, workholding deflection and servo lag. Written for engineers and buyers who need to judge whether a turned part can hold ±0.005 mm, and when it cannot.

What moves a turned diameter off nominal
A lathe cuts a surface of revolution. The tool sits at a fixed radial position while the part rotates, so the finished diameter is set by one distance: from the spindle axis to the tool tip. Every error source that matters acts on that distance. If the tool edge wears 0.010 mm, the part grows 0.020 mm on diameter. If the spindle axis shifts 0.003 mm, the part goes oval by roughly the same amount.
That is why turning accuracy is never a single figure. It is a static offset you can dial out with a test cut, plus a group of dynamic errors that keep changing during the cut. The static part is easy. The dynamic part decides whether a 0.005 mm tolerance band is realistic.
Take a 40 mm diameter shaft in 1045 steel. The diameter callout is ±0.005 mm, so the total radial budget is 0.005 mm. Tool nose wear alone can eat 0.002 mm in sixty seconds of continuous cutting. Now the spindle and the servo system have 0.003 mm left to fight over.
Errors add up in ways that surprise people. Radial error shows up doubled on the diameter. Axial error does not touch the diameter at all but shifts shoulder positions and lengths. Separate the two before you start chasing numbers on a gauge.
Thermal growth: the error that arrives after thirty minutes
A lathe is not in thermal equilibrium at 8 a.m. The spindle bearings, ballscrews, hydraulic chuck and coolant pump all add heat. A spindle that grows 0.020 mm axially between cold start and steady state will move every Z position with it. Diameters stay close. Lengths drift.
The classic pattern is a good first article and a drifting tenth part. Nothing changed in the program. The machine warmed up. On a lathe running at 4,000 rpm, spindle growth typically settles within 30 to 60 minutes, depending on the spindle design and the coolant routing.
The fix is boring but effective. Run a warm-up cycle at the cutting speed before touching a tight part. Keep the coolant temperature stable. On long runs, measure the first few parts at intervals and watch the direction of drift, not just the size of it. If the diameter walks in one direction, you have a thermal or wear problem. If it wanders both ways, look at the servo or the workholding.
For parts held to ±0.005 mm, we schedule the warm-up into the process rather than hoping the operator remembers it. It costs fifteen minutes and saves a scrapped batch.
Tool wear, spindle error and servo lag
Flank wear is the slow one. As the insert wears, the effective cutting edge moves, and the part grows. On a finishing pass with a 0.4 mm nose radius, wear of 0.005 mm on the flank translates to about 0.010 mm on diameter. That is the whole tolerance gone on a tight shaft.
Spindle error splits into two parts. Radial runout of the spindle taper makes the part oval, and axial float makes face positions inconsistent. Thermal growth is the third component and the largest on most machines. Check radial runout with a test bar and an indicator before blaming the program.
Servo lag shows up on contoured profiles, not on straight turning. When the Z axis accelerates into a radius while the X axis is still catching up, the tool cuts a shape that is not in the program. Feed forward, servo tuning and corner rounding parameters decide how big that error gets.
The practical test is simple. Turn a straight cylinder with a slow feed and measure. Then turn the same shape at production feed. If the diameter moves, the servo system is contributing. If it does not, look at thermal and wear instead.
How to tell which error you have
Measure a part, not a feeling. Start with a micrometer at three positions along the length and at four clock positions around the diameter. Taper shows up as a diameter change along the length. Ovality shows up as a difference around the circumference. Each one points at a different cause.
A diameter that grows steadily across a batch is wear or thermal drift. A diameter that is correct at the chuck and small at the free end is deflection, from too much tool pressure or a part that needs a tailstock. A diameter that changes with spindle speed is balance or bearing preload.
Surface finish tells you a second story. A finish that starts at Ra 0.8 μm and decays to Ra 2.0 μm during one pass is chatter or wear. A finish that is fine but the size is wrong is a geometry or offset problem. Do not change offsets to fix a finish.
We inspect 100% of parts before shipment, with raw material checks, in-process monitoring and a final inspection. Reports are available on request. If you want to know what a specific feature needs, send the drawing and we will tell you which error source it is sensitive to.
Error source vs symptom vs what to change
Read the symptom column first, then the fix.
| Error source | Symptom | Typical size | What to change |
|---|---|---|---|
| Thermal growth | Size drifts one way over hours | 0.010–0.030 mm | Warm-up cycle, stable coolant |
| Flank wear | Diameter grows gradually in one pass | 0.005–0.015 mm | Insert grade, tool life limit |
| Spindle radial runout | Part is oval, size is stable | 0.002–0.008 mm | Test bar check, bearing preload |
| Workholding deflection | Taper or size change toward free end | 0.010–0.050 mm | Tailstock, softer jaws, lighter pass |
| Servo lag | Size right on straight cuts, wrong on radii | 0.005–0.020 mm | Feed forward, servo tuning |
| Chip recutting | Random size scatter, poor finish | 0.005–0.030 mm | Coolant pressure, chip breaker |
When ±0.005 mm makes sense, and when it does not
If the function is a bearing seat, a hydraulic spool or a press-fit bore, hold the ±0.005 mm band and plan for thermal control and tool life limits. If the part is a bracket, a cover or a spacer, tightening the tolerance adds cost without adding function. Tell us what the feature does and we will hold the band that the function needs.
Questions we get on turned tolerances
Can a lathe hold ±0.005 mm on a long shaft?
Yes, within limits. The limit is the length-to-diameter ratio, not the tolerance. A shaft with an L/D above 4 usually needs a tailstock or a steady rest, otherwise cutting force pushes the free end away from the tool and you get taper.
With proper support, we hold ±0.005 mm as a production tolerance, not a one-off claim. The inspection report shows where the measurement was taken, because a shaft that is round at the chuck and small at the end is not the same as one that is round everywhere.
Does spindle speed change the final diameter?
It can. Higher speed raises spindle and bearing temperature, so the spindle grows and the tool position shifts. Centrifugal effects at very high speed also change bearing preload.
For tight work we pick one speed for the finishing pass and keep it constant across the batch. Changing speed mid-batch is one of the most common causes of a slowly drifting diameter.
How much does tool wear move the diameter?
On a typical finishing insert, flank wear of 0.005 mm moves the diameter by about 0.010 mm. That is twice the wear number, because the error acts on the radius and shows up doubled on the diameter.
Set a tool life limit based on measured parts, not on a clock. When the measured diameter reaches the middle of the tolerance band, change the insert.
Why is the diameter right at the start of the pass and wrong at the end?
That is almost always deflection, not wear. Cutting force bends the tool or the workpiece, so the depth of cut drops as the tool moves away from the support. A tailstock, a follow rest or a lighter finishing pass usually fixes it.
Wear acts over the whole batch, not over one pass. If the error grows within a single pass, look at stiffness first.
Does coolant affect accuracy?
Yes, through temperature. Flood coolant carries heat away from the cutting zone, but it also absorbs heat from the pump and the machine. A coolant temperature that swings 5 °C during a shift moves the machine structure.
Keep the coolant volume large and the temperature stable. On tight work, run the coolant for a while before the first finishing cut so the machine reaches a steady state.
What tolerance can you hold on plastics and titanium?
Both behave differently from steel. Plastics like POM and PEEK move with temperature and spring back after cutting, so the measurement taken immediately after the cut can differ from the one taken an hour later.
Titanium alloys such as TC4 (Ti-6Al-4V) cut hot and wear tools quickly, so tool life limits matter more than machine capability. Send the material and the feature, and we will tell you which band is realistic.
Send a drawing, get a tolerance judgement
We review the features that drive your tolerance and tell you which ones are worth holding tight. Quotation and free DFM analysis within 12 hours.
12-hour quote100% inspectionNo minimum order quantity