Waterjet Troubleshooting — Loss of Accuracy, Poor Cuts and the Most Common Faults

Loss of Accuracy — Parts Come Out Skewed or Out of Size

A machine that used to hold ±0.1 mm tolerance now deviates by ±0.3-0.5 mm. The cause may be mechanical wear, calibration drift or a software error.

Calibration Drift — Symptoms, Causes and Intervention

The most common cause of lost accuracy is not mechanical wear but "drift" in the calibration parameters. The CNC controller knows the axis position from three things together — encoder, home switch and software offset; if any one of them changes, the position the machine "believes" to be correct is in fact wrong.

• SYMPTOM 1 — The same G-code file cuts parts of different sizes on two different machines (or on the same machine two weeks apart)
• SYMPTOM 2 — A test square (100×100 mm) measures something like 99.7 × 100.3 mm — an axis calibration difference
• SYMPTOM 3 — A diagonal cut comes out at 44.7° / 45.3° instead of 45° — the axes are not orthogonal (squareness)
• SYMPTOM 4 — A circular cut (R50 circle) comes out as an ellipse instead of a true circle — an X-Y axis scale difference
• SYMPTOM 5 — The positional deviation grows when the same cut is repeated 10 times (drift) — an encoder or home switch problem
• SYMPTOM 6 — The return-to-park command parks the machine in a different place — the home switch position has shifted

The 6 Most Common Causes of Calibration Drift

A Quick Test the Operator Can Run (5 Minutes)

1. Cut a 100 × 100 mm test square in a slab (a simple square from CAM)
2. Measure the X and Y dimensions with callipers or a micrometer — ±0.05 mm is normal, above ±0.1 mm is a calibration problem
3. Rotate the same file by 90° and cut it again — the dimensions should stay the same (axis orthogonality)
4. Measure the diagonal of the test piece (Pythagoras: √(x²+y²) = 141.42 mm) — a deviation indicates a squareness error
5. If there is a deviation, report it to USEL Authorised Service for a calibration check

On a USEL Authorised Service calibration visit the true positional accuracy of the X/Y/Z axes is measured with a laser interferometer (typically 1-2 µm resolution). Backlash compensation, pitch error compensation and squareness compensation parameters are written to the controller. Typically a 4-6 hour job; afterwards ±0.05 mm accuracy is the reference.

On a waterjet the cut surface is graded into 5 classes (Q1 rough separation, Q5 mirror quality). If you have been getting Q5 and have suddenly dropped to Q3, one of the parameters below has changed.

• CAUSE 1 — Worn orifice: a sapphire or diamond orifice wears out in 200-500 hours. The water stream spreads and the cutting stream weakens. SOLUTION: change the orifice
• CAUSE 2 — Worn focusing tube (mixing tube): it wears out in 80-150 hours. The internal diameter grows and the cutting stream is no longer focused. SOLUTION: change the focusing tube
• CAUSE 3 — Wrong garnet size: using 120 mesh instead of 80 mesh raises the speed but lowers the quality. SOLUTION: choose the garnet to suit the job (thick plate → 80 mesh, fine work and quality → 120 mesh)
• CAUSE 4 — Damp garnet: clumped garnet feeds irregularly and the stream vibrates. SOLUTION: fill the hopper with dry garnet and add silica gel
• CAUSE 5 — Wrong standoff (head height): beyond 2-3 mm the stream spreads. SOLUTION: calibrate the standoff sensor
• CAUSE 6 — Traverse speed too high: reduce the feed rate in CAM to match the quality class (Q5 → 100-150 mm/min)
• CAUSE 7 — Pressure has dropped: the gauge shows 3,500 bar instead of 4,000 bar. The pump and accumulator must be checked

Along the line of cut the top surface is wider than the bottom — the part has tapered into a V shape. A certain amount of taper is unavoidable because of waterjet physics, but too much of it means there is a problem.

• CAUSE: The traverse speed is faster than the cutting stream can fully penetrate the material → the stream spreads downwards and the lower part of the cut stays narrower
• CAUSE: Worn focusing tube (the parallel symptom of falling quality)
• CAUSE: Excessive standoff (more than 3 mm)
• SOLUTION 1 — Reduce the traverse speed by 10-20% (especially over the last 5 mm at corners)
• SOLUTION 2 — Activate "taper compensation" in IGEMS/Contranest CAM software
• SOLUTION 3 — On a 5-axis machine, tilt the head with the A/B axis (dynamic taper compensation)
• NOTE: On a 2D 3-axis machine taper cannot be eliminated, ±0.5° is accepted; with a 5-axis head it drops below ±0.05°

There is no burnt edge on the top surface of the slab as there is in laser cutting; but there can be a protruding burr (frosting) on the underside or small projections on the top edge.

• BOTTOM BURR — The stream has not fully cut the material as it leaves the underside of the slab. Cause: high traverse speed, low pressure, worn orifice
• TOP BURR — A pierce blow-out at the lead-in. Cause: the lead-in speed is too slow, or the pressure has not been given a soft start
• SOLUTION: Optimise the lead-in/out parameters — a 30-50 mm lead-in for thick plate, 5-10 mm for thin plate
• SOLUTION: Increase the "pierce delay" — piercing time on thick plate is 3-8 s

Wavy lines running up and down appear on the cut surface — a zigzag on an edge that should look straight.

• CAUSE: The accumulator (pressure vessel) gas pre-charge has dropped — the pressure fluctuation shows on the surface
• CAUSE: Irregular pump piston stroke (mechanical wear)
• CAUSE: Irregular garnet feed (vibrator fault)
• SOLUTION: Check the accumulator nitrogen pre-charge (an annual service job)
• SOLUTION: Clean the garnet lines and test the vibrator
• NOTE: Light striation is accepted as normal at Q3-Q4 quality; for a Q5 mirror cut the accumulator and pump must be working at their optimum

Before cutting begins there is a piercing operation. At this step the slab cracks, the stream deflects or the hole forms irregularly.

• BRITTLE MATERIAL (glass, marble, ceramic) — a direct pierce cracks it. SOLUTION: a "low-pressure pierce" (1,000 bar low pressure at the start → raised in stages)
• COMPOSITE MATERIAL (engineered stone, CFRP) — risk of delamination. SOLUTION: a pre-drilled hole (a starting hole drilled beforehand)
• THICK STEEL (above 40 mm) — piercing takes a long time and the stream spreads. SOLUTION: a "dynamic pierce" (pressure and standoff raised in stages)
• PIERCE DELAY setting: 0.5 s for 3 mm steel, 5-7 s for 25 mm steel, 12-15 s for 50 mm

Consumables wear out over time — observation matters when answering the question of when the head needs changing.

• SYMPTOM: The tank water is black or dark — the garnet sediment has built up beyond its limit. SOLUTION: clean the tank (weekly vacuum removal of garnet)
• SYMPTOM: The drain is blocked — the water level is rising. SOLUTION: clean the drain filter
• SYMPTOM: The pH value is below 6.5 or above 9 — stainless components corrode
• SOLUTION: pH stabilisation (target pH 7-8) plus a biocide additive to prevent bacterial growth

The 7 Things the Operator Should Check in 5 Minutes at the Start of a Shift

Annual Inspection — USEL Authorised Service

Operator-level daily and weekly checks prevent 60-70% of breakdowns. The remaining 30% — pump seals, accumulator, valve set, linear guideway lubrication, calibration — is carried out by authorised service. The detailed maintenance schedule is set out in the USEL Care maintenance documentation.