CNC Programming Nightmares: The 7-Step Checklist I Wish I Had Before That $4,200 Rush Order

Who This Checklist is For (and When to Use It)

If you've ever stared at a finished part that's wrong—maybe a hole off by 0.010", maybe a thread pitch you thought was right but wasn't—and felt that sinking feeling in your stomach while the clock ticks down on a rush order... you're in the right place.

I run CNC programming for a mid-sized job shop. In my first year (2017), I crashed a spindle on a Hurco VM10i because I forgot to G43 my tool length. That was a $3,200 mistake. Three hours scratching my head, then one click of 'Run,' and the sound of a $1,500 tool shattering. I was certain I'd entered everything correctly.

Since then, I've personally documented 43 significant programming errors across our five Hurco machines. Combined cost? Roughly $11,000 in scrapped material, broken tooling, and rework. This checklist is the result. I use it before every program goes near a machine. Especially on those rush orders where the client needs the parts yesterday.

Why the rush emphasis? In August 2023, I paid $600 for expedited delivery on a batch of 6082 aluminum. We saved two days on shipping. Then we scrapped the whole batch because of an error in the post-processor. The $600 rush fee turned a one-week delay into two weeks. Certainty matters when the clock is ticking.

I've seen plenty of buyers ask "Can you reuse resin in 3D printer?" to save a few bucks. In CNC, the logic is different. The cost of a mistake scales fast. This checklist won't make you a master programmer, but it will catch the expensive stupid stuff. Here are the 7 steps.

Step 1: Verify the Work Coordinate System (WCS) Location

Most common error I see: Someone sets G54 to center of the part, but the CAM file assumes top-left corner. Or vice versa. The result is a crash, or a feature in the wrong spot.

Every time I load a new program, I check the WCS settings on the control. On the Hurco WinMax, I go to the Setup MDI screen and physically verify the coordinates. Not just the numbers—I jog the machine close and eyeball it. Takes 90 seconds. Saved me at least three crashes.

Step 2: Match Your Tool List to the Program

This sounds obvious. It's not. In September 2022, I programmed a part that called for a 0.500" endmill. The operator loaded a 0.475" endmill from the wrong drawer. The program ran fine until it didn't—pocket walls were 0.025" undersized. 47 parts, $680 in material, straight to scrap.

My rule: Print the tool list from the CAM post. Then physically lay out every tool on the bench next to the printout. Check diameter, length, and corner radius. Do not load anything until you've verified each one. I use a caliper for critical dimensions.

Step 3: Dry-Run the First Part (Without Cutting)

This isn't a suggestion. It's non-negotiable. On our Hurco lathes, I run the program in Single Block mode with the feed hold at 10%. No material loaded. Watch the tool path on the screen. Watch the actual machine motion.

The question everyone asks is "Is the program correct?" The question they should ask is "Did I visually verify that the tool moves where I expect?" If you've ever worked with laser fiber inside vein ultrasound guided applications in medical prototyping, you know that a 0.001" deviation can be the difference between a valid part and a failed inspection. On a standard job, maybe it's not that tight. But the habit matters everywhere.

Step 4: Check Your Feeds and Speeds for the Actual Material

I've seen a program set up for 6061 aluminum running at 10,000 RPM and 200 IPM. Great for the prototype. When the same program ran on a batch of 304 stainless steel? 10,000 RPM with a 1/2" endmill is 1,310 SFM—way too fast for stainless. The tool lasted 3 parts before edge failure.

Manufacturers publish recommended speeds for a reason. On Hurco, you can enter the material in WinMax and it suggests conservative settings. Use that. Then cross-reference with your tool supplier's chart. I keep a laminated card of common feeds and speeds taped to the monitor. Best $10 I ever spent.

One nuance: When it's a rush order, default to conservative settings. Adding 30 seconds per part is better than rescheduling the whole job. That extra $400 on expedited shipping doesn't matter if the parts are wrong.

Step 5: Verify the Post-Processor Output

This is the step most people ignore until it bites them. We use Fusion 360 at our shop. The post-processor for our older Hurco VM20 is different from the one for the newer VMX30i. I once used the wrong post. The program looked fine in the CAM preview, but the actual G-code had wrong IJK arc centers. Crashed a $700 fixture.

Before any production run: Open the posted G-code in a text editor. Look for the header comments—does it list the correct machine? Check a few lines of a simple move. Does it match what you expect? This takes 2 minutes and has caught errors on maybe 10% of my jobs. Worth it.

Step 6: Validate Workholding Clearance

In March 2024, I programmed a part that required a long reach tool. In CAM, the tool path looked fine. On the machine, the tool holder body hit the vice jaw at Z-1.500". The sound was awful. The spindle was okay (thankfully), but the tool holder was destroyed. Plus a 3-day delay for a replacement.

Most people check tool length. They forget that the holder, collet nut, and extension all add up. I now do a full clearance check in the CAM simulation, then again physically with a single block run. If it's a near-miss, I'll add 0.25" of clearance in the program. Which is to say: don't cut it close. Not ideal, but safe.

Step 7: Run a Test Part in Wax or Foam

This costs maybe $15 in material. It has saved me over $2,000 in scrapped metal. For the parts we cut in aluminum or steel, I keep some blocks of machinable wax. When I have a new program—especially one with tight tolerances or new features—I'll run it in wax first.

The best part of this process: you can measure the wax part. If the hole pattern is wrong, or the pocket depth is off, you caught it on a $3 block instead of a $50 aluminum billet. Take it from someone who learned the hard way: wax is cheaper than aluminum.

A note on 5-axis: If you're programming a 5-axis Hurco, this step is absolutely mandatory. The kinematics are complex. A single sign error in the rotation direction can send the tool through the workholding. I don't even load material until the wax part passes inspection.

What Happens When You Skip Steps (A True Story)

In April 2024, we had a rush order for a medical component. Stainless steel, ±0.005" tolerance. The client needed it in 4 days. We skipped Step 5 (post-processor check) and Step 7 (test part). The program was for our Hurco TM8i lathe, but the operator was on the TM10. The turret indexing offsets were different. The tool hit the part incorrectly on the first operational pass.

Part destroyed, tool damaged, 2 days of production lost. The client eventually got their parts—another shop saved us—but we paid $800 in rush fees to them. That 'time certainty premium' we usually charge? We paid it to someone else. The irony stung.

The lesson: you can't outsource your own process failures. The rush fee buys speed from others, not from physics.

Final Notes & Frequently Missed Details

• Always check G41/G42 direction. I've programmed climb milling as conventional. The surface finish was terrible. Took me 15 minutes to find the error.
• Don't trust tool offsets blindly. If a tool is 2.000" long in the offset table, but the actual tool is 1.995", you'll be 0.005" above the part. Not a crash, but a bad part.
• Pause and re-check after any CAM update. I once updated Fusion 360 and the post-processor silently changed a macro call. Found it by accident during a dry run.
• Backup your programs. We had a hard drive failure. Lost all the test programs for a custom 5-axis part. Had to redo everything.
• Resin reuse is a 3D printing problem, not a CNC problem. But if you are doing rapid prototyping with both techniques, the checklist mentality transfers. Verify your material. Verify your setup. Don't assume.

I have mixed feelings about sharing this list. On one hand, it makes me look like I make a ton of mistakes (I do). On the other, if it saves one person from the sound of a shattered tool, it's worth it. Seriously.

Prices as of February 2025; verify current rates for materials and rush shipping.