Heavy Return Return Returns to Normal: What Changes on the Machine

• 1
  What a heavy return actually is at the tool tip
• 2
  How to tell a healthy heavy return from a damaging one
• 3
  When heavy return return returns to normal is not enough
• 4
  What the load curve tells you about part accuracy
• 5
  Adjusting feed and speed so the return settles cleanly
• 6
  Heavy return behaviour by material and cut type

A heavy return is a load spike that comes back through the tool during a reversal of feed direction, a re-entry into a cut, or the first moments of a deep pass. The spindle and the servo axis both see it. On a 3-axis mill the spike shows up as a torque bump on the spindle drive and a current bump on the axis drive. On a mill-turn center you also see it on the live tool drive. The spike is short. It may last only 20–80 ms. That is long enough to leave a mark on the surface and short enough to be missed by a slow data logger.

When the cut settles, the load drops back to its steady-state value. Operators describe this as the heavy return returning to normal. It is a description of a load curve, not a mode the control switches into. Nothing in the NC program changes at that moment. What changes is the deflection of the tool, the holder and the workpiece, and the heat going into the chip and the edge.

The reason this matters is that the spike and the settled state load the tool differently. A 12 mm carbide end mill in 6061-T6 at 0.5 mm radial engagement may run at 1.2 kW steady. During a heavy return it can briefly pull 2.5–3 kW. The tool survives the spike. The problem is what the spike does to the next 200 mm of cut: edge chipping, chatter marks, and a size shift the operator has to dial out.

So the useful question is not whether the load returns to normal. It is how fast, how far it overshoots, and what the part looks like when it does. Those three numbers separate a healthy cut from one that will scrap the next part.

• 1
  Spike durationUsually 20–80 ms at the reversal or re-entry point.
• 2
  Spike magnitudeCommonly 2–2.5× the steady-state spindle load.
• 3
  Settle time100–400 ms depending on axis acceleration and tool stiffness.
• 4
  Part riskEdge chipping, chatter, and a size shift on the following pass.

Start with sound. A healthy heavy return produces a short, dull thud at the reversal and then a steady cutting tone. A damaging one produces a sharp crack followed by a tone that wavers for a second or two. The second sound is chatter, and it means the settling motion is exciting a structural mode in the tool or the fixture.

Next, look at the load meter or the drive data. Most controls will show spindle load as a percentage. Note the steady value and the peak value. A peak below about 1.5× steady is comfortable for carbide in aluminium or brass. A peak above 2.5× steady is where edge chipping starts, especially on small-diameter tools and on tools with a thin web.

Then check the part. Measure the first 5 mm after the reversal and the last 5 mm before it. If the size difference is inside your tolerance band, the heavy return is returning to normal cleanly. If the difference is larger, the tool is pushing off and the axis is catching up. That is a setup problem, not a feed problem. Look at tool overhang, holder runout, and how the part is supported.

Finally, check the chip. A healthy chip is consistent in thickness and colour after the reversal. A chip that suddenly goes blue or thin means the load settled at a different engagement than the one you programmed. On deep pockets in 7075 or 17-4PH this is often the first sign that the tool is rubbing rather than cutting.

• 1
  Peak below 1.5× steadyNormal for carbide in aluminium and brass.
• 2
  Peak above 2.5× steadyChipping risk on small tools and thin webs.
• 3
  Wavering tone after the thudChatter, usually from tool overhang or weak fixturing.
• 4
  Size shift over 10 mmDeflection problem, fix the setup before the speeds.

There are cuts where letting the load settle naturally is the wrong choice. Deep slots in titanium and Inconel are the clearest case. The material work-hardens at the surface, and a tool that dwells during the settle rubs instead of cutting. The next tooth then bites into a harder layer. That cycle repeats until the edge fails. In TC4 and Inconel 718 we keep the load steady by ramping into the cut rather than dropping in, and by using trochoidal paths that keep radial engagement constant.

Thin-wall parts are the second case. A 1.5 mm wall in 6061 will deflect under a heavy return and spring back after the load settles. The wall ends up tapered even though the load returned to normal. Here the fix is to support the wall, reduce radial engagement, and take the finishing pass at a lower feed so the settling motion cannot push the wall past its elastic limit.

The third case is a tool with a long overhang. Anything over 4× diameter in a shrink-fit holder will flex during the spike. The load returns to normal, but the tool does not return to the same position. The result is a size that drifts over the length of the cut. Reducing overhang, or stepping up to a larger shank, does more than any feed change.

If your part falls into one of these three groups, treat the heavy return as a design input. Choose the path strategy, the tool and the fixturing around it. Do not rely on the load curve to save the cut.

• 1
  Titanium and InconelWork-hardening makes dwell during the settle destructive.
• 2
  Thin wallsSpring-back leaves a taper even after the load normalises.
• 3
  Long overhangOver 4× diameter the tool does not return to the same position.

The load curve is a proxy for deflection. When the load rises, the tool pushes away from the workpiece, the workpiece pushes away from the tool, and the machine structure stretches a little. When the load settles, everything springs back. The size you measure on the part is the sum of those two states plus the thermal growth of the spindle.

This is why a part can measure in tolerance at the start of a cut and out of tolerance 300 mm later. The heavy return that returns to normal at the start of the pass leaves the tool slightly off position. The next pass starts from that offset. Over a long cut the offset accumulates into a taper.

For parts held at ±0.005 mm, we plan for this. We keep radial engagement below 8% of tool diameter on finishing passes, use 5-axis paths that keep the tool normal to the surface, and take a spring pass at the end. On a 16-station 5-axis cell we can hold ±0.005 mm on aluminium and ±0.01 mm on stainless across a 300 mm part.

The same logic applies to surface finish. A heavy return that settles cleanly leaves a uniform Ra 0.8–1.6 μm. A heavy return that overshoots leaves a band at the reversal that can be 2–3× rougher. If the drawing calls for Ra 0.2–0.8 μm, that band is a reject. We finish those surfaces with a separate light pass rather than trusting the load to settle.

• 1
  Finishing engagementKeep radial engagement below 8% of tool diameter.
• 2
  Spring passOne light pass at the end removes the settling offset.
• 3
  Tolerance on aluminium±0.005 mm across 300 mm on a 5-axis cell.
• 4
  Finish band riskOvershoot can make the reversal 2–3× rougher.

When a heavy return is causing trouble, the first change is usually the entry. A straight plunge into a deep cut creates the largest spike. A ramped entry at 3–5° spreads the load over a longer distance. On aluminium we often use a 5° ramp; on stainless, 2–3° with a slower feed.

The second change is the reversal itself. Keep the tool moving through the corner. A sharp reversal at full feed creates a spike. A small arc, typically 0.5–1 mm radius, keeps the load continuous. This is why high-feed toolpaths look so smooth on the load meter.

The third change is spindle speed. Counter-intuitive as it sounds, raising the speed slightly can reduce the spike because the chip gets thinner and the cutting force drops. We test this in steps of 10%, watching the peak load. If the peak falls and the surface stays clean, keep it. If the tone gets sharper, back off.

The last change is feed. Reducing feed per tooth reduces the spike but also reduces the chip thickness, which can push the tool into rubbing. On 6061-T6 we rarely go below 0.05 mm per tooth for a 12 mm end mill. Below that the edge rubs and the finish gets worse, not better.

• 1
  Ramped entry3–5° in aluminium, 2–3° in stainless.
• 2
  Corner arc0.5–1 mm keeps the load continuous through the reversal.
• 3
  Speed testStep up in 10% increments and watch the peak load.
• 4
  Feed floorStay above 0.05 mm per tooth for a 12 mm end mill.