The optimal speed for turning external threads on a lathe

• 1
  Why the optimal speed for turning external threads is a range
• 2
  Convert surface speed to RPM before you touch the offset
• 3
  Pitch, thread height, and infeed method set the real limit
• 4
  Material decides the window before anything else
• 5
  Setup rigidity caps the speed you can actually use
• 6
  Match the symptom to the cause before changing the speed
• 7
  Surface speed and pass count by material and pitch

On a CNC lathe the spindle and the Z axis are electronically geared. The controller keeps the tool moving one lead per spindle revolution, so in principle the spindle could run at any speed and the thread would still come out at the right pitch. That is where the idea comes from that threading speed does not matter.

In practice it matters a lot. The insert has to survive a cut where the chip is thin at the flanks and wide at the crest, and where the tool is fully engaged around the part. Surface speed drives heat, and heat drives flank wear. Too fast and the tip rounds off within a few parts. Too slow and the chip tears instead of shearing, leaving a rough flank and a thread that will not gauge cleanly.

The working answer is a range, not a value. You pick a starting surface speed from the material, then move inside that range based on pitch, thread height, and how much the part hangs out of the chuck. A short, thick, well-supported thread tolerates the top of the range. A long thin stud on a slender shaft does not.

The range also depends on what you are willing to do after turning. If the thread only needs to assemble, a wider range works. If it has to pass a go/no-go gauge at class 2A or better, the upper end of the range is usually closed off.

Every threading insert comes with a surface speed window in m/min or sfm. That number is for the cutting edge, not the spindle. Convert it with RPM = (surface speed × 1000) ÷ (π × major diameter) when you work in metric, or RPM = (sfm × 3.82) ÷ diameter in imperial. For a 20 mm coarse thread in 1045 steel at 120 m/min, that is about 1,900 rpm. For a 60 mm thread in the same steel, the same surface speed is about 640 rpm.

This is why a single rpm number copied from a forum post is worthless. Diameter dominates. A speed that burns a 12 mm insert will barely warm a 60 mm one.

On our turning centers we treat the surface speed window as the contract with the insert and the rpm as a derived value. If the derived rpm is higher than the machine can synchronize cleanly, we lower the surface speed instead of fighting the spindle.

Spindle acceleration matters on short threads. The controller has to reach the commanded rpm before the tool enters the thread. On a 6 mm long thread with a 2,000 rpm command, the ramp eats a real part of the cut. Set the spindle start point far enough ahead of the thread that rpm is stable at entry.

• 1
  Start from the insert windowUse the manufacturer's surface speed for the material group, not a guess.
• 2
  Convert per diameterA 5× diameter change is a 5× rpm change at the same surface speed.
• 3
  Check the rampShort threads need a longer lead-in so the spindle is stable at entry.

Pitch is the distance between crests, and it sets how much material one revolution removes. A 1.5 mm pitch removes roughly 1.5 mm of axial material per turn at the tool nose. A 3 mm pitch removes twice that. Higher pitch means higher cutting load per revolution, which pushes you toward the lower half of the surface speed range.

Thread height is the other half. The radial depth from major to minor diameter on a 20 × 2.5 mm thread is about 1.5 mm. That is not one cut. We normally take 6 to 8 passes for a coarse thread in steel: a few roughing passes that remove most of the depth, then two or three finishing passes at constant radial depth to clean the flanks.

Infeed method changes where the load sits. Radial infeed cuts with both flanks at once and produces a V-shaped chip that is hard to evacuate. Flank infeed cuts with one flank, which lowers the load per pass and improves chip control but needs more passes. Alternating infeed splits the wear across both flanks and is a good default on stainless and titanium.

The rule we use: as pitch and thread height go up, surface speed goes down and pass count goes up. On a fine 0.8 mm pitch on a 10 mm diameter, we can run near the top of the insert window. On a 3 mm pitch on the same diameter, we drop 20 to 30 percent and add passes.

Aluminum threads fast. On 6061 and 7075 we run 200 to 300 m/min with sharp uncoated or polished inserts, and the main risk is built-up edge, not wear. A slightly higher speed and a positive rake help. On 2024 the chip is gummier, so we keep the speed up and the depth per pass small.

Carbon steel 1018 and 1045 sit in the middle. A coated carbide insert at 100 to 150 m/min with 6 to 8 passes is a normal starting point. On 4140 and 4340 we drop to 80 to 120 m/min and expect to index the insert more often because these grades work-harden at the flanks.

Stainless 303, 304, 316, and 17-4PH are where threading speed gets people in trouble. These grades work-harden, so a slow rubbing pass is worse than a fast clean one. We keep surface speed at 60 to 90 m/min, take a real depth on every pass, and never let the tool dwell. 17-4PH in the H900 condition is harder again and often runs at the bottom of that band.

Titanium Ti-6Al-4V and Inconel need the slowest speeds and the most passes. Ti-6Al-4V runs around 30 to 50 m/min with generous coolant and a rigid setup. Inconel runs lower still and is a case for a thread mill or a ground thread rather than single-point turning on small diameters.

A threading tool is a spring. It deflects under radial load, and the deflection shows up as a pitch error or a torn flank. The longer the tool overhang, the lower the speed you can use before chatter starts. We keep threading tools as short as the holder allows and prefer a boring-bar-style holder over a long straight shank.

Part support is the second half. A thread on a part held in a collet close to the chuck is stiff. The same thread on a 150 mm overhang is not. For long slender threads, a tailstock or a steady rest lets you keep the surface speed up. Without one, drop 30 to 40 percent and accept more passes.

Coolant delivery changes the limit too. Threading produces a chip that wants to stay in the groove. High-pressure coolant aimed at the leading flank clears it and lets you run faster on stainless and titanium. Flood coolant with a weak stream is the main reason people think these materials must be threaded slowly.

The last variable is the machine itself. A lathe with a worn Z-axis ballscrew or a loose turret will not hold pitch at high rpm no matter what the insert can take. If the thread gauges tight at the start and loose at the end, the problem is the machine, not the speed.

Torn or rough flanks usually mean the surface speed is too low for the material, or the depth per pass is too light. Stainless and titanium tear when the tool rubs. Raise the surface speed 15 to 20 percent, or take a deeper finishing pass, and the flank cleans up. If the roughness is only on one flank, the infeed method is the problem, not the speed.

Chatter that shows as a rippled flank or a repeating pitch error points at rigidity. Lower the surface speed 20 percent and shorten the tool overhang before you touch anything else. Adding a steady rest or a tailstock is usually more effective than any speed change.

Fast insert wear on the nose but clean flanks means the speed is too high for the thread height. Drop to the lower half of the band and add a roughing pass so the finishing passes only clean. If the wear is on one flank only, check the infeed method.

A thread that gauges oversize at the start and on size at the end is a machine or thermal issue, not a speed issue. Let the spindle warm up, check the Z-axis backlash, and confirm the lead compensation is correct before you chase the surface speed.