Nov 01, 2026
Grinding Robot Rigidity, Payload and Heavy Gate Cutting: Three Specs Foundries Usually Misread
When a foundry compares grinding cells, the first glance is almost always the same: how many axes, what envelope, what rated payload. Those numbers sit neatly in a brochure. What shows up after start-up is different. The wheel chatters on a gate stub, the spindle bogs, the path drifts a fraction, and everyone argues about the program.
Iron and steel make this obvious. Aluminum flash is thin; you buy speed and a sharp cutter. Iron and steel gates, risers and parting-line fins — especially stubs that still carry scale before heat treat — buy rigidity and sustained cutting force.
Three numbers that share a spec sheet are not the same thing on the floor.
Rated payload, table capacity and cutting force
A 200 kg arm rating does not mean the machine can push a wheel with 200 kg of force. Payload is how much the arm can carry: spindle, dressing, gripper. Table capacity is fixture plus casting. Cutting force is what the wheel can keep on the stub without the path floating.
A general-purpose six-axis with a spindle hung on the flange often has little payload left. Ask it to chew a thick gate and the gearbox and arm give first. The programmed path still runs. The actual depth of cut does not. Operators call it “inaccurate teaching.” The structure is yielding.
NEVIEW states this without dressing it up on the iron DR line. On the DR-750B, table and arm are both in the 200 kg class. On heavy gate stubs, sensor trials have seen more than 165 kg of acting force. That figure is not a slogan. It says the load during stub cutting is real and it lands on the machine and the fixture. Selecting only because “the casting weighs less than 200 kg” usually means extra passes later, and the cycle time falls apart.
Why heavy gates punish a soft structure
Flash on a flat face is polite. Contact area is predictable. A gate stub is not. The section is ragged, sometimes with shrinkage or burnt-on sand. The instant the wheel bites, contact jumps and the force spikes.
A skilled grinder eases off, changes angle, takes two or three bites. A robot that keeps a fixed normal force on a soft machine shows three familiar faults: chatter that kills wheels; path give that leaves a ring at the root; fixture creep that throws every later hole and parting line off.
On heavy iron, the first question is not “can we teach it.” It is “can this structure take the force at the stub.” Dual stations, enclosures and vision matter. They come after the force path is closed.
Material changes the demand
High-volume iron with reasonably regular gates fits a rigid dual-station DR cell: DR-550B for smaller parts, DR-750B for mid-size, DR-950B and DR-1200R/B when envelope and mass climb. Cut first, then argue takt.
Steel is harder on the same stub. GR-750B and GR-950B raise spindle power and capacity for a reason. Wheel cost on steel already hurts; add chatter and the consumable bill outruns depreciation.
Aluminum is a different job. You rarely lean on a stub with that kind of force. You trim flash, gate roots and cavity corners. The MR line is about cutter speed and compliant contact. Thin walls dent if you shove. You still need stiffness — to hold the path, not to ram the part.
Giga-castings are a third case. Once the envelope approaches something like φ2300 mm, the problem is stiffness at long reach. The same tip force becomes a much larger moment at the base. Those cells are not catalogue machines. Floor space, number of spindles, floating versus rigid heads are all part-specific.
Spindle kilowatts are not cutting capacity
Motor power is one link. Holder stiffness, wheel diameter, surface speed, feed and whether the fixture backs the cut all sit in between. Oversized wheels raise surface speed and also the overhang. Undersized wheels cut but eat changeovers. Iron wants wheels and burrs; aluminum wants mills and disc cutters. Swap the logic and you load the cutter or cut through a wall.
Compensation is separate. Stub height varies with the casting. Without a probe or displacement check, a rigid machine only repeats the same wrong depth. Path correction from measured offset, and current/position/torque loops for parts that flex, solve different shop problems. Do not mash them into one phrase.
Why a converted general-purpose arm struggles on thick stubs
A stock robot, an electric spindle and a spark shield quote well and ship fast. Shallow aluminum flash in small batches can live with that. Mid-to-large iron with continuous stub cutting runs out of structural margin first.
A foundry-built cell sizes table, arm, spindle and enclosure for the same force path. The dual-station table has to hold during the cut. The cabin has to live with dust and vibration, not only hide sparks. Skip that and the machine will run on day one. Three months later you watch gearbox temperature, repeatability and wheel spend.
Not every plant needs the heaviest frame. Small, regular iron with gates already sawn back can live on a DR-550B envelope. Buying a DR-1200 “for later” drags floor space, power and fixture cost with it. Rigidity should be enough. Bigger is not automatically better.
What to ask and what to try
Try the worst stub, not the pretty face.
Ask whether rated payload includes the spindle and the toolholder.
Check that the fixture supports the cut direction so the part is not lifted.
Run a dozen pieces and watch whether root stock grows. Heat, wheel wear and micro-shift stack up; a single try-out hides them.
On steel, watch wheel spend. If the machine is quiet, consumables become a number you can plan.
If changeover is frequent, budget fixtures and vision on their own line. Do not hide that cost inside “cutting capacity.”
Close
Whether a grinding robot holds up in a foundry fettling bay is not decided by axis count on a slide. It is decided at the gate stub — hard, irregular, unforgiving. Get rigidity, payload and cutting force straight. Dual stations, sealing and compensation are worth discussing after that. If those three are wrong, editing the program will not stiffen the machine.
Read More
IPv6 network supported
