100 mm Ultra Low Profile Robot 7th Axis Linear Track

Why did the seventh axis have to be 100 mm tall?

Motionwell Automation designed and built an ultra low profile robot seventh axis linear track, 100 mm in overall height, carrying an ABB IRB-series six-axis arm across 1.5 m or more of stroke on a 3 m track at up to 500 mm/s, repeat positioning to plus or minus 0.5 mm on a ground-grade rack and pinion, with a slide platform load of 1.5 tonnes. It bolts directly onto the factory floor.

The reason to put an arm on a rail at all is spread rather than weight. Either the part is longer than the robot’s reach, or one robot has to serve several stations that will not fit inside a single working envelope. A track answers both because it moves the whole arm: every point along the stroke is served by the same reach, the same payload curve and the same tooling. What it does not do is add payload. An arm that cannot lift the part at full extension still cannot lift it after you have put it on a rail.

That much is ordinary, and it is set out in full on our robot seventh axis and linear track page. The constraint that shaped this particular build was vertical. A track sits between the floor and the robot base, so the arm’s entire working envelope rises by the height of the track, and that height comes out of the process end rather than out of the ceiling. A conventional robot travel unit takes 200 to 400 mm of vertical working height before the arm has done anything, which is why tracks in this payload class are often dropped into a sunken floor pit to win the height back.

Where the loss bites is not obvious from a datasheet, and it is worth checking in three directions rather than one. Reaching down: can the wrist still get to the bottom of a tote, a machine table or the infeed conveyor, or has the base risen far enough that it bottoms out. Reaching up: raising the base helps the top of a stack and hurts ceiling clearance, and it is the arm at full extension that finds the sprinkler line. Reaching in: a machine tool door or a chamber opening sits at a fixed height, and a pedestal of that height can turn a straight entry into an approach angle the wrist cannot hold.

What is in the delivered track?

The carriage is a machined mounting plate with dowel-pin location, so the arm returns to a known position on the platform after it has been lifted off rather than being re-found by eye. Guidance is two rails rather than one, which is what carries the overturning moment an arm generates when it works at extension off the side of the carriage.

Services travel with the robot in an energy chain running the length of the stroke, under a protective cover, with power, signal and air routed in one managed run. Cable specification is the part that decides whether that installation survives: continuous-flex cable rated for the travel, bend radius respected at design time, and chain fill sized so cables move rather than grind. A jacket cracked at a tight bend fails in the middle of the stroke, where it is least accessible, and it reappears every time it is patched. Travel is also protected independently of software by end-of-travel limit switches on the axis, so an overrun is caught by hardware rather than by a parameter somebody can edit.

Why is the drive a ground rack and pinion?

Because of what is riding on it. A robot on a carriage is a large mass with a moving centre of gravity, and it feeds reaction torque back into the axis every time the arm accelerates. That is a stiffness problem, not a resolution problem. A drive whose compliance changes as the carriage travels gives you an axis that settles differently at each station, and settling time is a cost you only meet once the cell is running.

Two consequences of that choice belong in any specification written around this axis. Backlash at the pinion mesh is dealt with mechanically at build time, because the mesh carries the load and it cannot be tuned out in software afterwards. And a rack and pinion axis is open by design, since the pinion has to engage the rack and the bearing blocks have to engage the rails, so protection is about keeping contamination off the engagement rather than sealing it away inside a tube. The same drive logic runs through the rest of our linear work, compared side by side on the servo lift and linear positioning page.

What does the +/-0.5 mm figure actually tell you?

Less than a specification writer usually wants it to. Repeatability describes how tightly the axis returns to a position it has already been taught, and +/-0.5 mm on this track is that measurement, taken through the rack and pinion drive with encoder feedback. It is a track figure, not an arm figure: the six-axis arm riding on it holds a much tighter repeatability at the wrist, and the track exists to deliver the arm to the station, where the arm does the fine positioning. Accuracy is a different claim again: how close the axis gets to a coordinate it has been commanded to, anywhere along the stroke. On the column palletizing track we quote +/-0.5 mm positioning along the rail from an absolute linear encoder, which is a positioning figure rather than a repeat figure. Two suppliers quoting different measurements look like a price difference until somebody checks which number was being quoted.

Then the harder point, which applies to every machine we build and is argued in full on our mechanical design and simulation page: none of those figures is placement accuracy on your part. The axis says nothing about gripper slip, part tolerance, fixture wear or thermal drift, and thermal growth is the one a long axis adds by itself, because it accumulates with length in a hall whose temperature moves between the night shift and the afternoon. The working answer is to teach positions at each station rather than deriving them from a nominal track coordinate.

How does the track run as the robot’s external axis?

It is driven by the robot controller, not by the cell PLC. ABB IRC5 and OmniCore controllers treat the track as a coordinated external axis, so the track servo hangs off the robot controller, the track appears in the same motion planner and the same coordinate frame as the arm, and the transit is planned together with the arm’s own motion instead of being sequenced after it.

None of that makes the PLC route wrong everywhere. A track that parks at three fixed stations with a long dwell at each one works perfectly well as an indexed axis, and the end-of-line palletizing cell is close to that pattern. The moment transit has to hide inside the arm’s own motion, the coordinated axis is what makes it possible at all.

What did the installation ask of the floor?

The civil scope was six adjustable feet and a levelling exercise. That is cheap compared with excavation, and it is the reason the 100 mm height is worth engineering for, but it is not nothing.

A track is a straightedge bolted onto a surface that is not straight, so levelling and shimming is what converts a good axis into a good installation. Flatness and level are measured along the whole run rather than assumed, and the run is levelled as a whole rather than foot by foot, because correcting one foot moves its neighbours. Anchoring is the second half of it. The anchors do not only carry the static weight of the arm and the carriage; they take the reaction every time the arm accelerates and every time a 1.5 tonne platform load decelerates on top of them, and those are horizontal loads into the slab as well as vertical ones. Distributed load along the track and point loads at the feet both get checked against the slab, particularly on an upper floor.

The last practical item is the route in. The track arrives in sections that have to reach the installation position through the doors and aisles that actually exist, which is what decides how the frame is split rather than any preference for fewer joints.

How does a moving robot base change the safety case?

The safeguarded space grows with the stroke. A fixed arm has one safeguarded area; an arm on a rail has one as long as the rail, plus the crushing and shearing hazards of the carriage and the track itself, which exist whether or not the arm is moving. The gap between the moving carriage and anything fixed beside it is a trapping point that travels along the run, and it is designed out with clearance and covers at concept stage or guarded around forever afterwards.

The method does not change. A risk assessment sets the limits of the machine and covers every mode, including teaching and fault clearing, when guards get opened and hands go where the designer never pictured them. The robot and the integrated system are covered by ISO 10218-1:2025, with each safety function rated for performance level under ISO 13849-1. One track-specific point is worth stating plainly: stopping performance now includes up to 1.5 tonnes of travelling mass, so the distances that set scanner and fence positions have to be measured on the built machine rather than calculated from the arm alone. That scope, and the LVD and CE testing around it, is described on our machine safety and compliance page.

Constraints and trade-offs

Distance, not weight. The track moves the arm; it does not raise the payload curve. If the arm is marginal on the part at reach, it is marginal at every point along the stroke, and the answer is a different arm rather than an axis under it.

One rail, one point of failure. A fault on the track, or on the robot, stops every station the rail serves. Two arms on fixed bases fail independently. Where a stoppage at one station cannot be allowed to stop the others, the second arm is the right purchase and we will say so.

A traverse on every cycle buys a slower cell. A track pays for itself when the transit hides inside a wait that already exists. If the robot has to run the full stroke every single cycle, transit time has been added to every cycle.

A track is one dimension of freedom. Stations on both sides of an aisle, or scattered around a cell, are the case a linear axis cannot serve, and no amount of stroke turns a line into a plane. Where the floor itself is the binding constraint, an overhead beam keeps the aisle clear underneath, which is the comparison run on the gantry versus six-axis machine tending page.

Where a catalogue travel unit fits, buy it. If your robot vendor’s standard unit covers the stroke, the payload and the height you have available, that is the useful answer. This track exists for the case where the height budget is the thing that does not fit. We also do not build production welding cells, which is one of the standard applications for a rail of this kind, so that exclusion belongs here rather than in a later conversation.

Related work

The heavy end of the same engineering, at platform loads to 2,000 kg and strokes beyond 20 m, is on our heavy duty robot track page, and the precision work this track class supports is described under aerospace and precision engineering. Loading and unloading machine tools with an arm on a rail is covered under machine tending automation, and the wider cell scope around any arm sits with robot integration services.

To size a seventh axis, send us the robot model, the stroke, the heaviest part with its tooling, and the height you have to work in.

Frequently Asked Questions

Can the stroke be extended after the track is installed?

Yes, and that is a property of the drive rather than a concession. Rack sections splice end to end, so the stroke is not limited by the drive element itself the way a ballscrew is limited by its own length. Extending the axis means adding rack, rail and frame to what is already there rather than replacing it, then re-levelling the run as a whole and re-teaching the positions beyond the old end stop. What has to be re-checked is everything the extra length touches: cable and energy chain length, the safeguarded area, and the floor under the new section.

What does a 1.5 tonne platform load actually cover?

Everything the carriage carries, not the weight of the part being handled. Platform load is the sum of the arm itself, its base plate, the dress pack, the end-of-arm tool and the workpiece at its heaviest. On a track of this class the arm is usually the largest term in that sum, which is why the arm has to be chosen before the track is sized rather than after. Platform load and robot payload are different numbers and they get confused in quotations, so state which one your specification means.

Does the low profile track work with robot brands other than ABB?

The mechanical scope is largely brand-independent and the controller integration is what changes. This track was delivered under an ABB IRB-series six-axis arm and runs as a coordinated external axis on ABB IRC5 and OmniCore controllers. The frame, carriage, rails, rack and pinion and cable management do not care whose arm sits on the mounting plate; what changes is the mounting pattern, the drive package the controller can take as an external axis, and how the axis is commissioned. Our heavy-duty track series is built to mount FANUC, ABB, KUKA and Yaskawa platforms.