CNC Dry Run Explained
A dry run executes the program with the spindle clear of the stock, so you can watch the motion before any material is removed. This page covers what it verifies, what it cannot catch, and how to set one up on a 3-axis or 5-axis machine. Written for engineers and buyers who need to judge whether a shop's first-article process is sound.
What a CNC dry run actually does
A dry run is a program execution with no cutting. The control reads the same blocks it will read in production, moves the same axes, and changes the same tools, but the tool tip stays above the stock or the spindle stays off. Nothing enters the material.
That distinction matters. A dry run proves the program and the setup agree with each other. It does not prove the cutting parameters are right, and it does not tell you whether the part will hold tolerance once real cutting forces appear.
Most shops run one of three variants. The simplest lifts the Z axis by a fixed clearance value, often 50 to 100 mm, and runs the whole program. A second variant uses the machine's graphics simulation with no axis motion at all. A third runs the program with rapids reduced to a low feed so the operator can watch each approach.
Which one a shop picks depends on the part. A first-article run of a 4,000 mm gantry part gets the slow approach. A repeat job with a proven program often gets nothing at all, and that is a deliberate decision rather than an oversight.
How a dry run catches errors before the spindle turns
The control has no idea where the stock ends and air begins. It only knows the coordinates in the program and the offsets in the table. A dry run puts those two sets of numbers in front of a human eye at a speed the eye can follow.
The classic failure it catches is a wrong work offset. If G54 was set from the wrong corner, the tool will drive to where the part is not. Watch the first approach move and the error is obvious before the insert touches anything.
It also catches missing or duplicated tool changes. A program that calls T07 twice because of a copy-paste error in the CAM post will show up when the carousel indexes to a station that should already be empty.
Retract plane errors are the third common catch. If the clearance plane sits below a clamp, the rapid between pockets will pass through the clamp body. On a 5-axis machine the same check covers the rotary table envelope, where a tilted tool holder can swing into the trunnion.
Setting up a dry run on the machine
Start by zeroing the work offset against a known reference, not against the stock face. A sawn face has 0.5 mm of variation or more, so the offset will be wrong by that amount before the first cut.
Enter the tool lengths from the presetter, then add a safe clearance in the Z offset table. A common approach is to add the clearance to the tool length itself rather than editing the program, so the CAM file stays untouched and the change is reversible in one edit.
Set the rapid override to 25 percent or lower for the first pass. Feed override does not affect G00 moves, so rapid override is the control that actually slows the approach moves you are trying to watch.
Keep one hand on feed hold and watch the distance-to-go display, not the part. Distance-to-go tells you what the control is about to do. The part tells you what it already did, which is too late.
- 1Clearance value50 to 100 mm above the highest stock point is typical for 3-axis work.
- 2Rapid overrideDrop to 25 percent or lower for the first approach of every tool.
- 3Single blockUse it through tool changes and any rotary indexing move.
- 4Distance-to-goWatch this readout instead of the cutting zone.
What a dry run cannot tell you
A dry run never loads the tool. Without cutting force, the machine cannot show you chatter, deflection, or thermal growth. A program that runs clean in air can still scrap the part on the first real pass.
It also cannot validate stock condition. If the casting arrived with 2 mm of extra material on one wall, the dry run will not see it. The tool will simply cut deeper than planned, and the wall will come out thin.
Tool wear and holder runout stay invisible. A dry run with a worn 12 mm end mill looks identical to one with a fresh tool, because neither one touches anything.
The last gap is measurement. A dry run confirms motion, not geometry. Only a cut part on a CMM or a height gauge confirms that the program produces the dimensions on the drawing.
Where dry runs fit in a tolerance-driven process
Tight tolerances change the value of a dry run. On a part held to ±0.005 mm, a dry run confirms the setup is repeatable before you spend a billet on it. The cutting parameters still have to be proven separately.
The order matters. Prove the setup with a dry run, prove the parameters with a test cut on scrap or on a slug of the same material, then run the first article. Skipping the middle step is how shops end up with a good program and a bad surface finish.
Surface finish targets are a useful sanity check. A program that leaves Ra 1.6–3.2 μm as-machined may need a separate finishing pass to reach Ra 0.8–1.6 μm or Ra 0.2–0.8 μm. A dry run will not show you which of those you are about to get.
For thin-wall parts, the dry run also hides the deflection that comes from the wall itself. Air cutting a 1 mm wall looks perfect. Cutting it with a 16 mm cutter at full radial engagement does not.
Dry run methods compared
Pick the method that matches the risk on the part.
| Method | What it verifies | Best for | Typical limit |
|---|---|---|---|
| Graphics simulation | Toolpath geometry and stock removal | New CAM output, simple 3-axis parts | No machine kinematics |
| Z-lifted dry run | Offsets, tool changes, clearance planes | First article on a proven machine | No rotary or trunnion motion |
| Full machine simulation | Kinematics, holder and fixture collision | 5-axis and mill-turn work | Needs an accurate machine model |
| Slow rapid on first pass | Real approach moves under load-free motion | High-value or long-cycle parts | Operator attention is the constraint |
| Air cutting with feed | Feed rates, servo response, cycle timing | Tuning and cycle-time checks | No cutting force or deflection |
| No dry run at all | Nothing | Repeat jobs with an unchanged program | One wrong offset scraps the part |
When to dry run and when to cut
For a new program, a new fixture, or a first article on an expensive blank, run the dry run and slow the first approach. For a repeat job with a proven program, a proven fixture, and an unchanged tool list, skip it and spend the time on the first-piece inspection instead.
Frequently asked questions
Does a dry run replace a first-article inspection?
No. A dry run checks motion and offsets. A first-article inspection checks dimensions on a real cut part, usually on a CMM or with hand tools depending on the feature.
The two answer different questions. You need both before a production run starts.
How much Z clearance should I add for a dry run?
50 to 100 mm above the highest point of the stock is a common range for 3-axis work. For tall fixtures or deep pockets, base the number on the tallest obstruction, not on the stock.
Add the clearance to the tool length offset rather than editing the program, so the change is easy to undo.
Can I dry run a 5-axis program without a machine model?
Not safely. The collision risk on a 5-axis machine comes from the holder, the trunnion, and the rotary table, none of which appear in a simple toolpath plot.
Use a full machine simulation with an accurate kinematic model, or run the dry run on the machine at low rapid override and watch the rotary moves closely.
Why did my dry run pass but the part still scrapped?
The most common causes are stock variation, a wrong tool length, or cutting forces that the dry run never applied. A dry run cannot see any of those.
Check the raw material against the drawing envelope, re-measure the tool, and review the cutting parameters before blaming the program.
Is a dry run the same as air cutting?
No. Air cutting runs the program at real feed rates with the spindle on and the tool in the air. A dry run usually keeps the spindle off and may lift Z or reduce rapids.
Air cutting tests servo response and cycle time. A dry run tests setup and clearance.
How long should a dry run take?
As long as the cycle, or longer if you slow the rapids. On a 40-minute cycle, a full dry run at 25 percent rapid override can stretch past an hour.
That is why shops dry run the first article and skip it on repeats. The time is only worth spending once per setup.
Send us your drawing and we will plan the setup
Upload your CAD file and get a quotation with a free DFM analysis within 12 hours. We dry run every new setup before the first cut and inspect 100 percent of parts before shipment.
12-hour quote100% inspectionNDA on request
More machining notes
We publish setup notes, tooling trials and inspection data from the factory floor.
