How to Repair Wires on CNC Towers
A broken insert or a re-chucked part does not have to scrap the thread. This guide is for turners and setup engineers who need to pick up an existing spiral groove on a CNC lathe, not start a new one. Read it and you can decide whether to chase the thread or cut it off and start over.
Key takeaways
Why a thread breaks mid-cut
A thread breaks in the middle of the cut for three common reasons. The insert edge chips on a hard spot or an interrupted surface. The part shifts in the jaws after a heavy roughing pass. Or the operator changes a finishing insert and the new tip sits a few hundredths away from the old one. All three leave you with a half-finished helix and a decision to make.
There is no mechanical link between spindle rotation and Z travel on a CNC lathe. The spindle encoder sends pulses, the control counts them, and the Z servo moves the turret to match the programmed lead. That relationship is electronic, not geared. Once the tool leaves the cut, or the part moves, the control has no memory of where the groove actually is.
This is why a simple re-home and re-start produces a crossed thread. The control begins the cycle at the programmed Z start point, but the groove on the part sits somewhere else. The insert cuts a second helix next to the first one. On a 1.5 mm pitch that offset can be visible within one revolution.
A useful mental model: the thread is a rigid stamp already pressed into the part. Repair work means finding that stamp and dropping the tool back into it. Every step below serves that one goal.
What to check before a repair pass
Measure the part before touching the control. Use a thread micrometer or a three-wire set on an external thread, and a pitch gauge plus a bore gauge on an internal one. Write down the actual pitch diameter. If the finished size is already within 0.1 mm of the limit, a repair pass will likely push it out of tolerance.
Inspect the groove under a loupe. Torn crests, galling and a crossed start are not repairable by re-cutting. A clean, partially cut groove with a sharp crest is a good candidate. If the first two or three revolutions look crossed, stop and cut the thread off.
Confirm the pitch and the start count. A multi-start thread has a start angle offset between each helix, so picking up the wrong start ruins the part. Count the starts on the print or on a sample before you index the spindle.
Check the insert and the holder. A repair pass removes very little material, so the tip radius and the flank angle must match the original insert exactly. A different nose radius changes the contact pattern and can rub instead of cut.
What breaks the repair pass
The most common error is re-homing the machine between the break and the repair. Homing resets the Z reference to the machine zero, and the encoder-to-servo phase is rebuilt from that point. The groove on the part has not moved, so the relationship is lost. Avoid any reference return once the thread is partly cut.
Changing spindle speed mid-repair is the second trap. On many controls the Z start point is tied to the encoder index pulse. Run the proving pass and the finish pass at the same rpm. If you must slow down for a deep internal thread, slow down for every pass, including the first one.
A worn or wrong-radius insert rubs instead of cutting. Repair passes remove 0.02–0.05 mm per side, which is below the threshold where a dull edge will cut cleanly. Fit a fresh insert with the same ISO code as the original, and check the seat for chips before clamping it.
Finally, do not skip the measurement. Threads that are repaired by feel alone tend to come out at the top of the tolerance band, and then fail a go/no-go gauge at inspection. Measure between every pass on the first part of a batch.
Limits of thread chasing
Thread chasing works when the original groove is largely intact and the part has not moved. It fails when the helix start is crossed, when the material has galled onto the flanks, or when the part has been re-chucked with runout above 0.05 mm. In those cases a full re-cut removes 0.3–0.5 mm of stock on the diameter, which may not be available on a finished part.
Hardened material above roughly 45 HRC is a gray zone. A repair pass with a fresh CBN or carbide edge can work, but the risk of chipping rises with every extra pass. If the thread is a structural feature, cutting it off and starting from a soft state is usually cheaper than a scrapped part.
For production parts, the better answer is prevention. Use a roughing and finishing insert pair on the same cycle. Keep the part clamped through the finish pass. Log the Z wear offset after each tool change so the next setup starts from a known point.
Step by step: re-syncing to the existing groove
Work cold, with the part still clamped if possible
- 1Keep the part in the chuckDo not unclamp after a break unless the part moved. The jaws hold the original angular position, which is half the battle. If the part must come out, re-true it and check runout at 0.02 mm or better before any repair pass.
- 2Index the spindle to a known angleUse M19 with an S value or a spindle orientation command to lock the spindle at a fixed encoder count. Note the value. Every later pass must index to the same count, or the Z reference shifts by one encoder pulse.
- 3Touch off X on the major diameterBring the insert in slowly until it just kisses the crest. Set this as the X reference for the thread. On a 60° insert, expect the tip to sit 0.05–0.1 mm below the theoretical major diameter.
- 4Find Z with the insert in the grooveJog Z until the tip drops into the groove by hand feel, or use a dial indicator against the turret face. Move in 0.01 mm increments near the target. The aim is to land within 0.05 mm of the original start point.
- 5Enter the Z shift as a wear offsetDo not rewrite the program. Put the measured difference into the Z wear offset for that tool. This keeps the original cycle intact and makes the change reversible if the pass needs repeating.
- 6Run one proving pass at reduced infeedProgram a single G92 or G76 pass with 0.02–0.05 mm radial infeed and the spindle speed cut to 60–70% of normal. Watch both flanks. A clean chip from both sides means the sync is correct.
- 7Check the pitch diameter againMeasure after the proving pass. If the size moved less than 0.02 mm, add infeed in 0.02 mm steps until the part reaches the mid-limit of the tolerance band.
- 8Finish with a spring passOne pass at the final X with no additional infeed cleans the flanks. Keep the spindle speed steady. Changing speed between the proving pass and the finish pass shifts the Z phase and can cross the thread.
Repair or re-cut: pick by condition
Use the left column to find your situation
| Condition | Repair pass | Re-cut from scratch |
|---|---|---|
| Groove cut less than half depth | Yes, low risk | Not needed |
| Complete thread, size still in tolerance | Yes, one spring pass | Wastes 0.3–0.5 mm of stock |
| Crests torn or galled | No | Yes, first choice |
| Start of helix crossed | No | Yes, thread is lost |
| Part re-chucked, runout over 0.05 mm | No | Yes, after re-truing |
| Hardened steel above 45 HRC | Risky, use a fresh edge | Safer with a full cycle |
| Internal thread, deep bore | Yes, watch chip evacuation | Requires a boring bar change |
Repair the groove, do not re-cut it
If the helix start is intact and the part has not moved, a single proving pass at 0.02–0.05 mm will tell you in one minute whether the sync is right. If the crests are torn or the runout is over 0.05 mm, cut the thread off and start clean.
Repair wires on CNC towers: common questions
Can I pick up a thread without a spindle orientation command?
On most controls you can, but it is slower and less repeatable. Touch the insert into the groove by hand, then lock the spindle with M19 before you record the Z offset. If the control has no orientation option, use a low-speed jog and mark the chuck jaw position so every restart uses the same angular reference.
How much Z error is acceptable before the thread crosses?
Keep the error under 0.05 mm for a standard 60° thread. Beyond about 0.1 mm the insert cuts a visible step on the flank, and the go gauge will not enter. On fine pitches below 1.0 mm, tighten the target to 0.03 mm.
Can I repair an internal thread on a deep bore?
Yes, but chip evacuation decides the result. Retract fully between passes and use through-tool coolant or an air blast. A boring bar with a small shank deflects under load, so reduce the infeed to 0.02 mm and check the pitch diameter after every pass.
What if the part came out of the chuck?
Re-true the part first. Indicate the major diameter and the face, and get runout under 0.02 mm. Then follow the same Z picking procedure. If runout cannot be brought under 0.05 mm, the angular position is no longer reliable and the thread should be cut off.
Does the repair pass work on multi-start threads?
It works if you find the correct start. Index the spindle, then feed the insert slowly along Z until it seats in one helix. Mark the start number on the part and the control. Picking the wrong start produces a thread that gauges on one flank and fails on the other.
How do I keep this from happening on the next run?
Use separate roughing and finishing inserts, keep the part clamped until the finish pass ends, and record the Z wear offset after every tool change. Parts that run in one continuous cycle rarely need chasing. A 100% inspection pass before shipment catches the rest.
Send us the thread drawing
Upload the part and the thread callout. We return a quotation and a free DFM analysis within 12 hours, with no minimum order quantity.
12-hour quoteNo minimum order100% inspection
More machining notes
We publish setup notes, tooling trials and inspection data from the factory floor.
