Before You Start

If you're a workshop buying a CNC lathe for a specific job function or a job-shop contract, this checklist is for you. I'm a quality compliance manager at an industrial equipment company. Every machine gets reviewed before it ships or gets installed—roughly 40 to 60 units a year. In 2024, I rejected about 15% of first deliveries due to spec mismatches.

No sales pitch here, just the checklist I actually use. Ten items, each one maps to a verifiable checkpoint. Work through it across two or three supplier conversations, and it's about 30 to 45 minutes per item if you do it properly.

Quick disclosure: I'm writing this from the quality side, not procurement. My job is to catch problems before we sign the acceptance line. Buyers, in my experience, care more about lead time and commercial terms—fair enough. But by the time we get to acceptance, the call is mine, and the specification is what it is regardless of what the sales rep promised. That's what items 1 through 10 are for.

The 10-Item Checklist

1. Start with your parts, not the machine catalog.

Before you send any inquiry, pull up the parts that will actually go onto the chuck—material, geometry, tolerances, surface finish, run sizes, and the highest throughput you'll realistically need. A supplier quoting without seeing your parts should be an immediate stop. If they haven't asked about material, tolerance, or run size, their quote isn't a baseline—it's a placeholder.

Here's the oversimplification I see most often: it's tempting to think you can just compare unit prices against the same spec sheet. But identical paper specs from different suppliers can result in wildly different outcomes. A "20 HP" spindle from one vendor with used castings and another with a new casting are not the same machine. Nothing replaces sending your parts list first.

2. Pin down spindle bore diameter and chuck capacity.

Spindle bore diameter (bar capacity) is the spec most buyers underweight. It determines the largest bar stock you can feed. On a fixed-headstock CNC lathe, this is often a hard bottleneck—and no, a bar feeder doesn't magically increase it. Confirm the chuck size and jaw type at the same time. An 8-inch chuck is common on mid-range machines, but if your parts sit anywhere near that diameter, confirm the actual clamping diameter and the swing over the bed—not just the chuck housing.

3. Verify positioning and repeatability against ISO 230-2.

This is the heavyweight item. Positioning accuracy and repeatability are two different numbers, and repeatability is usually what most job shops actually need. Under ISO 230-2:2014 ("Test code for determination of accuracy and repeatability of positioning of numerically controlled axes"), you should be seeing both figures stated in writing. A common target for a two-axis lathe is around ±0.005 mm over a 300 mm travel, but tie that back to your specific parts before locking it in.

Hindsight here—I once signed off on a supplier whose sample sheet showed a beautiful static positioning number, but I didn't ask to see the repeatability records. The machine arrived, got installed, and started drifting on consecutive trial cuts in a way that no amount of compensation could hide. If I could redo that decision, I wouldn't have signed the PO without a repeatability report carrying VDI/DGQ 3441 results, or at minimum a multi-directional ISO 230-2 cycle log.

4. Check chuck repeatability, not just chuck bore.

For production work, chuck repeatability matters far more than chuck brand. Ask for radial runout measured with a certified test bar and face runout checked against a ring gauge if the supplier has one. If they can't produce a recent runout figure, assume the sample value is worse in practice. A reasonable target on a new machine is 0.005–0.01 mm radial runout at the spindle nose—treat that as a baseline rather than a warranty.

Trust me on this one: a chuck that looks fine in a demo can lose its repeatability after 6–12 months of shift work. The check now saves you a fight later.

5. Confirm turret station count, index time, and tool-holder compatibility.

Turret station count is the easiest spec to overpay for, and honestly most standard lathes have enough. The real issue is usually tool-holder interface and index time. Is the holder interface standard VDI, or a proprietary quick-change system? How many holders are included vs. extra? Will you need a calibration pass every six months? If the supplier can't answer these clearly, that's a red flag—not a deal-breaker, but a flag.

6. Lock CNC system compatibility and insist on a post-processor.

This one is the most operationally consequential item on the list and it's rarely in an acceptance checklist. The choice of FANUC vs. SIEMENS (or HEIDENHAIN) decides how easy it will be to hire operators and whether your CAM shop has a post-processor that matches this specific machine and control configuration. Ask for a post-processor that's been tested against your machine and controller—not a generic one.

Better yet: if you already have a programmer, have them sit at the control with your actual part program during a visit. That's worth more than a spec sheet with a lot of bullet points.

7. Insist on a real cutting test.

Demonstration cuts aren't enough. Ask for a cutting test using one of your typical parts, at the material and depth of cut your contract requires. ASME B5.54 and ISO 10791 are the reference standards for performance evaluation of machining centers—running a cutting test and measuring surface roughness, dimensional holding, and thermal drift on the spot is completely normal. If a supplier refuses a cutting test, walk. Simple cutting tests expose vibration, unstable surface finish, and thermal compensation drift that a two-minute demo will hide.

8. Audit calibration records and maintenance coverage.

Ask for the most recent laser interferometer calibration log or geometric accuracy sheet—both for the machine you'd actually receive and for their own service capability. While you're at it, ask what's covered under their standard maintenance package (lubrication, air supply, alignment) versus what's billed separately. Records don't guarantee honesty, but their presence is a decent proxy for whether the supplier actually verifies anything.

9. Put acceptance criteria in the contract, not just price terms.

This is the one that needs legal before anything gets signed—not after the machine hits your floor. Acceptance criteria should cover positioning accuracy, repeatability, surface roughness, and what the remedy is if a machine fails a check. Suppliers love vague tolerances followed by "within industry standard," which usually means they haven't actually measured. Get the measurement method, the instrument, and the acceptance number written down.

10. Get the acceptance test results signed and archived.

Last item, and the least glamorous—but this one gets skipped all the time. Ask for a signed acceptance test report with measured values, date, and the person who ran it, filed against your equipment record. Not a PDF, not a screenshot. When the machine's accuracy slips two years from now and the supplier suddenly changes their story, that signed document is your only leverage.

Notes and Common Mistakes

Don't compare only the quote sheet. Comparing quotes across suppliers is fine—just remember that the quote sheet rarely includes the parts that matter later: repeatability records, turret type, post-processor availability. Those are the items that decide whether a machine actually holds up on your floor.

Don't turn the "three-quote rule" into an endless comparison. The "always get three quotes" advice is fine for commodities, but on a CNC lathe it ignores the real transaction cost of supplier evaluation—site visits, test cuts, back-and-forth. On a first purchase, evaluating two or three genuinely-matched suppliers is more practical than trying to cover seven.

Don't pay before the acceptance signature. Price, delivery, and payment terms are negotiable—but land the final payment after the acceptance test passes and is signed. Put this at the top of the contract, not the bottom of a payment schedule.

Don't ignore maintenance—including your own side. A machine with no maintenance plan on your side will lose accuracy in the first year faster than the spec sheet's calibration interval suggests. We had a trial-cut issue a few years back that turned out to be a coolant nozzle aimed wrong—simple fix, but everything looked like a problem while it was unresolved.

One last note: the checklist itself evolves. What was best practice in 2020 doesn't automatically apply in 2025. The fundamentals haven't changed—define the part, define the spec, verify the spec—but the verification has transformed. Laser interferometry, cutting tests, post-processor validation—these are now table stakes, not optional extras. Keep the fundamentals, embrace the updated verification. That's how you get a CNC lathe that actually does the job.