The honest answer starts with a definition, because the fix depends on which problem you actually have. Cross-threading, strictly, means a second helix has been cut or forced across the first: two interleaved spiral paths in the same diameter. That part is usually scrap, and no code changes it, because a threading tool can follow one helix or the other but cannot merge two paths back into one thread. What G-code genuinely fixes is the family next door, threads that are damaged, shallow, or incomplete on the correct helix, and the technique is re-chasing: cutting the existing thread again, in sync with it.
Why synchronization is the whole problem
A threaded pass is not just a tool moving at a pitch; it is a tool moving at a pitch starting from a repeatable spindle position. CNC lathes synchronize threading moves to the spindle encoder, every pass of a G76 cycle triggers from the same spindle angle, which is why multi-pass threading works at all: each pass traces the same helix as the last. That same property is the entire repair theory. As long as the synchronization between machine and part is intact, running the cycle again continues the original helix; the moment it is lost, a new cycle cuts a new helix wherever it lands, which is how well-meaning repairs manufacture genuine cross-threading.
The decision table
| The situation | Sync status | The honest move |
|---|---|---|
| Thread undersized or incomplete, part never unchucked | Intact | Re-run the cycle deeper: it continues the same helix |
| Burrs or assembly damage on first turns, part unchucked | Lost | Pickup procedure, then re-chase, or hand-chase with a die |
| True second helix cut across the first | Irrelevant | Scrap, or single-point a new thread at the next size where the design allows |
| Torn material below the root, safety-critical joint | Irrelevant | Scrap: a cosmetic re-chase cannot restore integrity |
The first row is the everyday save and barely counts as repair: measured undersize after the cycle, add depth, run again. Untouched setup is the condition doing the work, the same reason measuring before unchucking is lathe discipline everywhere.
The pickup procedure, for the re-chucked part
When the part has been out of the chuck, the spindle-to-part relationship is new, and the cycle’s default helix will land where it lands. The pickup re-establishes the relationship manually:
Stop the spindle. Bring the threading insert slowly into an existing groove near the thread start, by hand wheel, at a depth where the flanks locate the tip, and note the machine’s Z and X. The offset between where the cycle would start and where the groove actually is becomes a Z shift of the threading cycle’s start position, so the commanded helix drops into the existing groove instead of across it. Some controls support a phase or start-angle word in the threading cycle that does the same job arithmetically; the thread mechanics are identical either way, and dialect details for spindle-synchronized moves live with G33 and G76 documentation per control.
Then the universal verification: a scratch pass with effectively zero depth, inspected by eye or marker ink, before any pass that cuts. The scratch tells you in ten seconds whether the helix landed in the groove, and it is the difference between a pickup and a gamble. Thread geometry questions that surface during this work, why the pass count and depths are what they are, why RPM stays locked during threading, have their own treatments.
The economics nobody skips twice
A pickup done carefully costs real setup minutes, which is the honest comparison against the blank’s cost: re-chasing a hardware-store bolt loses money before the spindle turns, while saving a finished part with an hour of machining behind it pays for the care. The third column of the decision table is the discipline: safety-critical threads with unverifiable damage are scrap at any economics, and a left-hand or unusual thread raises the pickup care another notch. Like most rescue work, the skill underneath is unglamorous: knowing the threading cycle’s anatomy cold, so the shift you command does what you intend, fluency the free 60-second drills on the G-code practice page keep current for the day the expensive part needs saving.
Sources
- Helman CNC: Fanuc G76 threading cycle
- Wikipedia: Threading (manufacturing)
- LinuxCNC: G-code reference
Frequently asked questions
Can you fix a cross-threaded part with G-code?
Usually no: true cross-threading means a second helix was cut across the first, and two interleaved paths cannot merge into one thread. What G-code does fix is the neighboring family, threads damaged, shallow, or incomplete on the correct helix, by re-chasing in synchronization with the existing groove.
How does thread re-chasing work on a CNC lathe?
Threading cycles trigger from the same spindle position each pass, so on an untouched setup, re-running the cycle deeper continues the original helix. If the part was re-chucked, a pickup procedure must re-align the tool with the existing groove first.
What is the thread pickup procedure for a re-chucked part?
Spindle stopped, locate the insert in an existing groove, note Z and X, and shift the cycle’s start position so its helix lands in the groove. Then verify with a zero-depth scratch pass before cutting.
When is a damaged thread worth saving versus scrapping?
Re-chase when the helix is correct and damage is shallow. Scrap when a second helix exists, material is torn below the root, or the joint is safety-critical and integrity cannot be verified.