-
Who this checklist is for
-
Step 1: Lock in tolerances and surface finish before you open a catalogue
-
Step 2: Match the carbide grade to the workpiece material
-
Step 3: Specify the tool holder interface and spindle sync requirements
-
Step 4: Demand a live sync-cutting demo, not a catalogue page
-
Step 5: Write down the first-article protocol and sampling rules
-
Step 6: Validate tool life on a fixed workpiece
-
Step 7: Decide whether a hybrid process route makes more sense
-
Common mistakes (the ones we actually made)
If your shop is bringing a multi-axis lathe online this year—or moving a part from single-axis turning onto a multi-axis process—this checklist is for you. I'm a quality and brand compliance manager at a manufacturing company. Over four years, I've reviewed 200+ tooling spec sheets and rejected about 12% of first-article submissions last year. Not because the vendors were bad, but because the specs were loose enough that the supplier had to guess. Here are the 7 steps we now use, in order.
You already know CNC basics. I'm not gonna start with what a turning insert is.
Who this checklist is for
Three types of readers:
- Quality engineers signing off on the first multi-axis part.
- Purchasing writing the first tooling PO for a new multi-axis program.
- Process engineers deciding whether a part should go on a multi-axis lathe at all.
If you're strictly in R&D or prototyping, this checklist is overkill—too fine-grained. But if you're specifying production tooling for parts headed into molds, aerospace fixtures, or automotive powertrain components, read on.
Seven steps total. Steps 3, 4, and 7 are the ones I see skipped most—and most of our rework traces back to those.
Step 1: Lock in tolerances and surface finish before you open a catalogue
A huge number of spec sheets start with "we want to use insert grade XYZ." That's the wrong order.
Pull the part drawing first. Circle three things:
- The tightest diameter tolerance (not typical—the tightest one on the print)
- All mating-surface roughness requirements
- Any interrupted cuts (interrupted, semi-interrupted, keyways, cross-holes)
These aren't 'reference data.' They drive insert grade, edge prep (T-land width, for instance), and cutting parameters directly. Skip this and every step downstream is patching a wrong assumption.
Checkpoint: your spec sheet should carry one line stating both the tightest tolerance and the coarsest allowed finish. If your wording is "typical ±0.05 mm," the supplier is going to quote the cheapest configuration that technically meets it.
"The 'just get the part out' thinking comes from the single-axis era. Today, most multi-axis jobs are won or lost on finish and cycle time together—you can't trade one for the other."
Ten years ago that logic held. Today's multi-axis machines won't return a good part just because you asked them to.
Step 2: Match the carbide grade to the workpiece material
Everyone knows this in theory. In practice, most spec sheets just say "carbide" and leave ISO classification off entirely.
Workpiece material sets the direction. The ISO 513 categories are industry common knowledge at this point: P steel, M stainless, K cast iron, N non-ferrous, S heat-resistant alloys, H hardened steel. If you've spec'd an S-class superalloy (say Inconel 718) but the quote came back with a P-grade insert, the supplier probably isn't going to flag it—they'll quote to your sheet, then the first article goes sideways.
I've had this bite me. In 2022, an aerospace part. We spec'd a general P-grade because that's where the quote landed. First articles kept missing surface roughness. I spent two weeks chasing cutting parameters before our tooling vendor pointed out the obvious: this was Inconel 718, and it wanted an S-grade fine-grain insert. Batch scrapped. Three-week delay.
Checkpoint: your spec sheet needs a line saying "ISO 513 category: X." Add a note that cross-category substitutions require written justification from the supplier.
Step 3: Specify the tool holder interface and spindle sync requirements
This is the step most people quietly hand off to the supplier. Don't.
The difference between single-axis and multi-axis turning is synchronization. Your main spindle, sub-spindle, and live tool turret have to move in sync, and the tool holder and turret interface accuracy directly determines whether you can hold consistency during cross-machining. You'll run into two kinds of suppliers: the ones who hand you a complete interface spec with runout data, and the ones who send a catalogue page and a price.
Find the first kind. Specifically, ask for:
- Spindle-end interface standard (HSK-T, Capto, VDI—which one, exactly)
- Measured tool holder radial runout (TIR), not nominal
- A written note on multi-turret sync consistency
Checkpoint: if the quote has no interface spec and no runout data, don't sign yet. You can just ask for it—established tooling suppliers hand it over. Sandvik Coromant's official website is actually the cleanest place to pull master catalogue PDFs with that data; going through third-party distributor pages usually ends in an outdated copy.
Step 4: Demand a live sync-cutting demo, not a catalogue page
Real sync performance only shows up on your workpiece.
The method is simple: take a real workpiece (or a coupon with matching material and geometry), and require the supplier to run a sync-cutting demo on your machine. That means they come in—not you buying inserts and trying it yourself. The demo should cover, at minimum:
- Main and sub-spindle turning a known diameter in sync
- Live tool cross-milling a face or slot
- Surface roughness measured when both turrets are cutting at once
Checkpoint: at the end of the demo, have them measure on the machine, with your person watching. Put the numbers in the meeting minutes. Don't accept "it's in our catalogue" as a response.
This is the one most people skip. I skipped it at first—figured 'major brands don't need to be tested.' Then on a multi-axis project, first articles showed a step at the main/sub-spindle joint. Two weeks of chasing before we found it was a turret sync parameter mismatch. A single live demo would've caught it in two days.
Step 5: Write down the first-article protocol and sampling rules
"First article passes, so go to production" is single-axis thinking. Multi-axis turning has far more variables—sync, live tools, sub-spindle—and first-article passing doesn't mean batch stability.
What we do:
- First article: full dimensional inspection + surface roughness + CMM on key mating features
- Part 5, part 25, part 100: spot-check key dimensions
- Every 200 parts: one spot check until the batch closes
- First part after insert change: mandatory check
Checkpoint: this protocol goes into the contract attachment. Verbal agreement doesn't survive a shift change.
Step 6: Validate tool life on a fixed workpiece
When a catalogue says "X parts per insert," that's from their test coupon. Your geometry, material lot, coolant setup, and operators are all different.
The method: take one of your real workpieces, run it continuously with the same insert and same parameters. Record:
- Surface roughness at part 1
- Surface roughness at part 50
- Part count when roughness goes out (or at 100, whichever hits first)
- Flank wear land width (VB) on the insert
Then calculate two things: insert cost per part, and machine time consumed by insert changes. The second number gets ignored a lot—if a change is 4 minutes and you change 5 times a day, that's 80+ hours a year.
Checkpoint: the life-validation data gets signed off. Don't use "the supplier said X parts" as justification.
Step 7: Decide whether a hybrid process route makes more sense
Not every part is a good fit for pure multi-axis turning.
If the part has deep cavities, internal channels, or topology too complex for a multi-axis lathe to reach—don't force it onto the machine. That's when it's worth evaluating metal 3D printing (SLM, for instance) for the blank or near-net shape, then finishing with multi-axis turning. Sandvik Group has a foot in additive manufacturing as well, so their official site actually has both subtractive and additive content side by side—which is convenient.
One warning though: metal printing safety management runs on a totally different rulebook from conventional machining. Powder handling, inert gas, explosion protection, post-processing—each has its own standard. If you search "are 3D printers harmful" online, most answers are about desktop plastic printers. Industrial metal printing has a different risk profile entirely—don't borrow those answers.
As for "awesome 3D printers"—that consumer-grade stuff is fine for prototyping. Don't put it on a production evaluation. Precision, material systems, and cost structure aren't comparable.
Checkpoint: when you're deciding the process route, list at least two options: pure multi-axis turning, and additive + multi-axis finishing. If you can't list two, you haven't understood where the part's real difficulty lives.
Common mistakes (the ones we actually made)
Mistake 1: writing "per industry standard" on the spec sheet. Industry standards have several interpretations. The supplier will pick the one most favorable to them.
Mistake 2: skipping the sync demo. I mentioned earlier—we've skipped this twice. Both times, the first article had issues and we burned two weeks diagnosing.
Mistake 3: going to production after the first article passes. Multi-axis turning has bigger batch-to-batch variation than single-axis. One part doesn't represent the batch.
Mistake 4: treating insert life as a single number. Change time, setup time, insert consistency—all of it is cost. Calculate total cost, not unit price.
Mistake 5: treating 3D printing and CNC as opposites. They're not a binary choice. Hybrid process routes are increasingly common in multi-axis work—provided you evaluate both sides for risk and cost separately.
This checklist won't guarantee you pick "the best insert." There's no such thing—there's only the insert that fits this batch of parts, this machine, and this shift crew. If after all seven steps you still can't tell whether a part belongs on a multi-axis machine, run the sync demo from Step 4 first. That'll tell you more than any evaluation spreadsheet.