Mazak Sub-Spindle Transfer: Why Parts Drop, Crash, or Come Out Wrong

Twin-spindle work is where Mazak lathes earn their money — and the transfer is where they lose it. In thirty seconds the sub-spindle has to come up, match speed with the main, close on a spinning part, and take it away cleanly. When any part of that choreography is off, the failure is not subtle: a dropped part, a chuck collision, or a second-side part that's wrong because it was never held where the program thought it was.

I spent 19 years at Mazak Corporation as an Applications Engineer setting up twin-spindle machines and the years since doing the same work independently. Transfer problems follow a pattern. They are rarely the program's fault, and they almost always trace back to one of a handful of setup items nobody re-verified. I covered the short version of this in my Mazatrol tips and tricks article — this is the longer version.

The three checks before you ever press cycle start

1. Synchronization speed has to match. The sub-spindle grips the part while both spindles are rotating. If the speeds don't match at the moment of grip, the part gets twisted in the chuck, marked, or torn loose. Verify the programmed sync speed is the same on both sides of the transfer — it's a setting people change for one job and forget for the next.

2. The sub-spindle WPC is set independently — because it is independent. This is the one that bites experienced people. The main spindle's work coordinate does not carry over to the sub. The sub-spindle has its own WPC, and second-side machining dimensions are only as good as that setting. A part can transfer perfectly and still machine wrong on the second side because the sub WPC was assumed instead of measured.

3. Chuck pressure has to match the part, not the last job. Too little pressure on the sub and the part slips or drops during the transfer or the cutoff. Too much and you crush thin-walled parts or distort the finished first side you're now holding on. Pressure is a setup-sheet item for every twin-spindle job, not a number that lives in the operator's memory.

The transfer position and Z offset, in plain terms

Conceptually, the control needs to know where the sub-spindle's chuck face ends up relative to the part when it comes in to grip — how far the sub travels, where it stops, and how much of the part it's taking hold of. That stopping relationship is what the transfer position / sub-spindle Z offset data establishes: the grip point on the part, measured against the machine's coordinate system.

Get it wrong in one direction and the sub stops short — it grips less of the part than you think, or misses its intended land and grabs a diameter that wasn't meant to be held. Get it wrong in the other direction and the chuck runs into the part, the main chuck, or the tool that just cut off. The exact screens and data names vary by control generation — T-Plus and T-32 handle it differently from Matrix and Smooth — but the logic is the same everywhere: the offset is a measurement, not a guess, and it has to be re-established when the setup changes — different chuck, different jaws, different grip length on the part.

One related trap: gripping on a finished diameter with hard jaws because the setup was rushed. The transfer can be geometrically perfect and still scrap the part cosmetically. Bored soft jaws or a collet on finished surfaces is not fussiness, it's the job.

C-axis orientation on Matrix Nexus class machines

On machines with C-axis work — milling on the lathe, cross holes, flats — the transfer has a second alignment problem: rotational position. The sub has to take the part in a known orientation or the second-side milling features land in the wrong place. On the Matrix Nexus generation this is handled with C-axis orientation functions (you'll see references to orientation codes like M316/M319 in the forum threads on this), so the spindles align angularly before the grip. If your second-side milled features are clocked wrong but dimensionally right, the transfer itself is fine — the orientation step is what's missing or mismatched. Verify the orientation is actually being commanded on both sides and that the part isn't slipping rotationally in worn jaws at the grip.

After a transfer crash: verify before you re-run

The instinct after a transfer crash is to fix the one number that looks wrong and run it again. Resist it. A crash moves things.

1. Check both chucks and jaws for damage and for a jaw that shifted on its serrations.
2. Re-verify the sub-spindle WPC — a collision can make yesterday's setting meaningless.
3. Re-check the transfer position data against the actual grip point, in single block, at reduced rapid, with your hand on the hold button.
4. Confirm spindle sync behavior at low speed before trusting it at production speed.
5. Inspect the first transferred part completely — grip marks, second-side dimensions, feature clocking — before the run continues.

A transfer crash is the machine telling you the setup data and the physical setup disagree. Find the disagreement on the ground, not in the next crash.

When something on a Mazak lathe isn't working

I do this for a living. If you have a twin-spindle Mazak dropping parts, crashing transfers, or producing second-side work that doesn't match the first, send me the part, the machine, and the control version. I'll tell you what's possible.

Thank you, Tom Herzog