Should a Plasma Table Use a Robot or a Gantry Axis?

For the proposed plasma table, 300 ipm equals 5 in/s, while the ABB IRB6/2’s listed 1.3 m/s linear-axis maximum converts to about 3,071 in/min; the robot’s quoted speed alone does not disqualify it, but the roughly 500-lb one-sided gantry raised the decisive acceleration and screw-speed tradeoff. Plasma current and plate thickness were not specified, so a usable cut-rate limit must come from the selected power supply’s cut chart, not from a robot or servo maximum-speed figure.

Cut speed, axis speed, and heat input

Keep three limits separate: the process feed that produces a sound cut, the axis speed the mechanism can reach, and the acceleration needed to follow the programmed path. The builder was targeting about 300 ipm for thin-sheet cutting. That is a requested feed, not a demonstrated process setting; the torch, power supply, material, and thickness determine whether a given feed is appropriate. Too much feed for the selected process can prevent complete cutting, while an unsuitable low feed can increase heat input and degrade the edge.

The reported 1.3 m/s value is a maximum speed for the robot’s radial and vertical arm motions. Convert it as follows:

That is about ten times the 300 ipm target, not 30 ipm. It still does not establish torch-tip feed over the whole work area: robot configuration, coordinated motion, payload, path shape, and the other axes constrain tool motion. Conversely, a nominal axis speed above the process feed is not a reason to reject an architecture. Check the process rate and the complete path envelope separately.

Robot, cantilever, and two-sided gantry tradeoffs

For flat-sheet cutting, a gantry uses Cartesian axes directly and carries the torch over a planar work zone. A robot provides additional wrist and arm motion that can suit vertical or angled cuts on preassembled shapes, but that flexibility adds kinematic and control integration work when every cut lies on a flat sheet. A one-sided cantilever preserves an open side for loading, yet concentrates the moving structure and its bending loads on one side. A two-sided gantry supports the bridge at both sides and introduces a second side drive that must stay coordinated and square.

The mechanical direction ultimately selected for this machine was a double-sided gantry, after the one-sided design and robot concept were set aside. Reorienting the layout placed the heavier gantry movement across the shorter table direction and kept the long-direction carriage light; the design later moved from ball screws to helical rack and pinion. That is a sound direction for this flat-sheet use, provided the twin sides can be synchronized and the loading path remains practical. Keep the robot in consideration when the work actually needs multi-orientation cutting rather than treating its axis-speed number as the deciding comparison.

A robot installation also changes the safeguarding problem. A light curtain was considered, but choosing a guarding device alone does not validate the cell. Assess access from all directions, material loading, robot reach beyond the nominal table, and the safety-control response with a qualified machine-safety designer before powered operation. If those constraints make the open-front requirement difficult, compare a protected robot cell against the gantry layout instead of assuming that an open robot workspace is acceptable.

Moving mass and acceleration budget

Plasma path motion spends much of its time changing direction around corners and small features. The axis must accelerate and decelerate, not merely attain 300 ipm on a straight line. A controller that slows sharply on every contour can miss the intended process motion even when the motor reaches the requested straight-line speed. High moving mass raises the force required for a given acceleration; higher force then drives motor torque, gearbox, and structure requirements. Reducing practical moving mass can improve the acceleration margin without raising motor size and cost.

Use consistent units in the sizing calculations. In SI units, the translating-axis force begins with F = m × a + F_friction, where mass is in kilograms, acceleration in m/s², and force in newtons. Include the rotating inertia of the motor, coupling, screw or pinion, and gearbox, reflected through the transmission. For a ball screw with lead L in meters per revolution, idealized motor torque for the translating load is T = F × L / (2π × η), where efficiency η must come from the selected screw and operating conditions. For rack and pinion, the corresponding output torque is approximately T_out = F × r, then motor torque must account for the reduction ratio and gearbox efficiency. These equations do not size a drive without the acceleration profile, friction, rotating inertias, efficiency, duty cycle, and selected motor data.

The project considered whether to keep the steel bridge or use aluminum plate to reduce nearly 200 lb, and whether larger AC servos could compensate. There is no universal 150-lb gantry ceiling in the supplied design information. Compare a lighter structure against the stiffness, deflection, joining, and durability required for torch positioning. Upsizing motors without calculating the complete motion profile can mask a mechanical problem while increasing inertia, drive cost, and tuning difficulty. The later plan reported reducing weight and using two side-driven size-34 motors with gearboxes; treat that as the project’s drive direction, not a sizing prescription for another table.

Ball-screw speed and rotary-nut alternatives

The one-sided proposal used a direct-drive, 1.5-in-diameter ball screw with a 0.500-in lead for approximately 106 in of X travel. At the requested 300 ipm, that lead requires:

300 in/min ÷ 0.500 in/rev = 600 rev/min

With direct drive, the screw and motor turn at the same nominal speed. Actual operation also needs acceleration and rapid-motion margin, and the screw’s critical speed depends on its unsupported length, diameter, end support, and installation. The long screw raised a ball-screw whip concern. A fine lead reduces the rotational speed needed for a given feed but requires more input torque for a given axial force; a coarse lead reduces torque demand for the same axial force but turns fewer inches per revolution, requiring more shaft speed to reach a given linear feed. Select the combination by checking both motor torque over the motion profile and screw speed against the manufacturer’s allowable limits.

A rotating-nut assembly was considered to avoid spinning the long screw. The design discussion described a 20-mm screw paired with a 20-mm lead in the THK BLR line, and the proposed timing-pulley reduction was meant to alter the motor-to-nut speed relationship. A reduction can change torque and output speed; it does not remove the need to verify nut speed, screw support, preload, and backlash. The project abandoned this route because it found no straightforward way to add a second nut for preload in the BLR configuration. The THK DIR option was described as having a preloaded nut only with a precision-ground screw; the unpreloaded option was reported at about 0.015 in backlash. Confirm current product data before treating those design-era details as a present specification.

A driven-nut concept is therefore not an automatic solution to the long-axis tradeoff. It substitutes rotating-nut bearing and preload questions for screw whip and motor-speed questions. If reconsidered, compare the full catalog speed, load, support, preload, and accuracy data for the selected assembly against the rack-and-pinion option.

Rack-and-pinion backlash and gear reduction

The selected drive direction used helical rack and pinion with a 39:1 gearbox reported at less than 5 arcminutes backlash. The project estimated about 0.0025 in of drive backlash in an idealized calculation. That value is a theoretical drive estimate, not demonstrated machine positioning accuracy: it does not include rack-to-pinion mesh, mounting errors, bearing clearance, bridge deflection, servo following error, thermal movement, or calibration.

Backlash matters most when the direction reverses, such as on a contour or a short feature. Check reversal error at the torch carriage with a dial indicator or other suitable metrology, approach the same test position from opposite directions, and repeat with the mechanism under representative load. Also check both sides of a dual-drive gantry. A low gearbox backlash figure cannot by itself prove the two sides remain square, that the pinion mesh stays consistent along the whole rack, or that the torch tracks the programmed path.

The design discussion favored helical rack and pinion after finding a practical ball-screw compromise difficult: the long, fine-lead screw raised speed concerns, while the coarse-lead option raised torque concerns. The rack drive can avoid long screw rotation, but it still needs correctly selected reduction, rack support, pinion engagement, lubrication and contamination control, and a measurable accuracy target. Specify repeatability and reversal performance at the tool, rather than selecting a gearbox from backlash alone.

Rails, contamination, and loading access

The proposed X support used four 1.5-in Thomson Twin Pillow Block bearings, spaced 24 in in both directions; the Y proposal used paired 1.5-in supported shafts with bearings separated vertically by 9 in. The Z proposal used THK 12-series square-slide bearings, 350 mm long, for approximately 4.5 in of travel. Those dimensions describe the project design, not a verified load-rating calculation. Check the manufacturer’s allowable load and moment for the actual orientation, shaft support, bearing spacing, acceleration, and cable/torch loads.

Orientation can change capacity. A bearing user reported that an open bearing’s rating was reduced by more than 50% when it was pulled upward rather than pushed onto its rail. Treat that as a warning to inspect the manufacturer’s load cases for the installed direction, not as a substitute for current catalog data. Another operating detail was that wiper seals added more friction than expected. Include the actual wiper drag and preload in the axis force budget, then confirm drive current and motor temperature through repeated motion tests.

Plasma cutting produces metal dust and slag. The table design angled the bed so slag could slide toward the front for cleaning; it did not tilt, and ball transfers could pop up for material handling. A production layout also needs a practical way to remove accumulated scrap and contain dust. Under-table extraction may require temporary coverage around the sheet, so the cover arrangement must not obstruct the torch path, material supports, or loading access. Protect rails, screws, racks, pinions, and bearing wipers from debris; contamination can add friction or damage a rolling element, changing both motion load and positioning behavior.

Motion control and robot integration

Camsoft Professional with a six-axis DSP Ethernet card was considered, but the control system was not reported as selected. The key selection is not whether a package can import a DXF file; it is whether the complete chain can turn the intended part geometry into coordinated, bounded motion with usable homing, limits, torch-height control, fault handling, and safety behavior. DXF import does not by itself provide a plasma process plan, a postprocessor, robot kinematics, or validated axis tuning.

Robot control requires a specific architecture decision. Establish whether the ABB robot’s original controller and amplifiers will remain, or whether another motion controller will command the axes. The installation had robot control cabinets, while reuse of the existing amplifiers remained undecided. Before purchasing software or drives, identify each amplifier’s command and feedback interfaces, motor and encoder compatibility, axis limits, homing method, and how the controller handles the robot’s kinematic coordinates. A statement that software supports G-code is not proof that a robot can accept ordinary Cartesian G-code without a suitable kinematic transform and coordinated servo interface.

A gantry has fewer kinematic translation questions, but a double-sided bridge requires a defined method to home and square the two sides and prevent skew under unequal loads. Confirm whether the controller supports the required coordinated axes and, where used, a gantry-slaving or squaring routine. Test the path interpreter, feed override, limit handling, and torch-height I/O with the plasma output disabled before making cuts. Select control hardware only after these interfaces and functions are documented for the actual software, controller, drives, and motors.

Torch touch-off and THC

Programmed sheet thickness alone does not establish the top-surface Z datum when plate thickness varies, the sheet is bowed, or it sits unevenly on supports. Touch-off and torch-height control (THC) solve different tasks. A touch-off sensor finds the plate surface and establishes a repeatable Z reference before a pierce; THC then adjusts torch height during cutting using the selected system’s sensing method. THC cannot correct an unknown starting datum merely because the nominal plate thickness was entered in software.

The proposed Z arrangement included a spring- or shock-mounted torch slider and a limit switch to detect the plate. If that is the chosen sensing method, account for its compliance and switch repeatability in the Z travel and reference logic. Check that the torch mount can sense a surface without driving the torch into the plate, that the switch state is visible to the controller, and that the Z axis can retract to the required pierce and cut heights. The exact pierce height, cut height, dwell, and THC enable point depend on the selected torch and power-source cut data; read those values from the equipment documentation rather than estimating them.

Commission the sequence with outputs inhibited first: home Z, approach a test plate, verify touch detection, establish the datum, retract, and confirm that the programmed pierce and cut positions are referenced to the measured surface. Then test arc-voltage or other THC feedback under the manufacturer’s prescribed procedure, and verify that THC does not chase the plate during pierce or act on unstable feedback. The table builder had not yet selected software or a THC setup, so this sequence is an implementation requirement, not a reported completed test.

Design and commissioning sequence

Use a gated procedure so a speed target does not substitute for a mechanical, electrical, or process calculation.

1. Define the cut envelope. Record sheet dimensions, required torch orientations, loading side, access openings, and the full work area. Compare that envelope with the robot’s reach or each gantry axis stroke. For the reported robot-on-slider concept, the builder described a 5-ft slider and about 42 in of reach across a 48-in-wide sheet; verify the actual usable envelope at every required point.
2. Set process requirements. Choose the plasma power source and torch, then obtain its cut data for each material and thickness. Record cut feed, current, pierce and cut heights, and any timing values from that equipment data. Do not use the 300-ipm motion target as a universal cutting parameter.
3. Model moving mass and motion. Weigh or calculate each moving assembly, specify cutting and rapid speeds plus the acceleration profile, and calculate linear force and reflected rotary inertia. Include wiper friction, cables, torch mount, gearbox, and transmission losses. Compare lighter structure against required stiffness.
4. Check the transmission at both ends of the speed range. For a screw, calculate rpm from lead and feed, then compare the complete speed range to screw critical-speed data and motor limits. For rack and pinion, calculate output torque and linear force through the reduction, then check pinion speed, backlash, engagement, and rated loads.
5. Close the control architecture. Choose robot-controller reuse or replacement, confirm amplifier and encoder compatibility, define homing and limits, and verify interpolation, gantry squaring if applicable, THC I/O, and fault responses. Confirm these functions against the selected controller and software, not a generic feature list.
6. Validate the structure and safety design. Check bearing load cases in the installed orientation, rail and rack alignment, gantry stiffness, scrap removal, dust extraction, and material handling. Have a qualified machine-safety designer validate guarding and safety functions for the actual robot or gantry work envelope.
7. Test motion without cutting. Verify homing, travel limits, axis direction, repeatability, reversal error, acceleration, drive current, following behavior, and two-side gantry squareness at low risk before increasing speed. Confirm the torch path stays inside limits.
8. Prove the cut sequence. With suitable test material and the selected process data, validate touch-off, datum reset, pierce, cut-height transition, THC response, corner motion, edge quality, and repeatability. Compare measured cuts with the process chart and revise mechanical tuning or process settings based on which limit is failing.

Separate heat faults from logic faults during commissioning. A poor edge with stable axis following points first toward process setup, consumables, height, and feed; stalls, skew, position loss, or drive trips point toward load, friction, tuning, feedback, or controller configuration. Trend drive current and inspect fault logs alongside cut quality so a thermal/process correction does not conceal a motion fault.

Frequently asked questions