Robot cells fail at the guard, not at the robot. Coolant mist, chips, static and a 6-axis arm that occasionally reaches the fence decide whether the panel survives two years or two months. Here is how we specify sheet, thickness and mounting for automation work.
Four things come up on almost every enquiry in this sector. None of them is solved by choosing a thicker sheet.
Guard panels crack from the bolt outward within months. We drill oversize, chamfer both faces and use a floating washer so the panel is never clamped hard against steel.
Emulsion builds a film that cannot be wiped off with solvent. Guards in wet machining get a wipe-down protocol and, where the mist is heavy, a polycarbonate window in an acrylic frame.
Anti-static sheet is specified for cells handling film, foam or powder. Standard sheet in a dry cell will hold a visible dust layer inside a shift.
A door 900 mm wide in 8 mm sheet bows. We either specify 10 mm or bond a stiffening return — a decision made at drawing stage, not at install.
All of the below run on the same machines, in the same plant, with no operation subcontracted. Full detail sits on the Industrial & machinery line page.
Three questions decide the specification, and they are asked in this order: what can hit the panel, what has to be seen through it, and how is it held. Skip the first and you buy the wrong polymer; skip the third and a correctly specified panel cracks at a bolt hole in six weeks.
A guard on the outside of a fence sees coolant mist, chips and a sleeve. A guard inside the reach envelope sees a 6-axis arm. Those two are not the same product. Acrylic has roughly 1/200th the impact energy absorption of polycarbonate, and it fails by shattering into sharp fragments, so the rule we apply is blunt: acrylic everywhere the hazard is splash or contact, polycarbonate wherever a mass can arrive at speed. A cell usually ends up with both — a polycarbonate impact pane at the operator interface, acrylic windows on the sides and top.
Thickness is a deflection question before it is a strength question. A 5 mm panel on a 500 mm opening bows visibly under its own weight and will chatter against the frame at the machine's running frequency; 8 mm on the same opening is dead flat and silent. Our standing numbers: 5 mm to 400 mm opening, 8 mm to 700 mm, 10 mm to 1,000 mm, and anything wider gets a machined stiffener or a division bar. A door is one step thicker than a fixed panel of the same size, because it is handled.
| Position | Material | Thickness | Why |
|---|---|---|---|
| Operator interface | Polycarbonate | 6–8 mm | Impact from a dropped or ejected part; acrylic would shatter |
| Side and top windows | Cast acrylic | 5–8 mm | Mist and chips only; clarity and scratch resistance matter more |
| Hinged door | Cast acrylic, annealed | 8–10 mm | Handled daily; needs stiffness and a stress-free hinge line |
| Electronics bay window | Static-dissipative acrylic | 4–6 mm | Standard PMMA holds several kV after a wipe; a bare board does not survive that |
Almost every guard failure we are asked to diagnose is a hole problem, not a material problem. Three rules stop it. Drill before bending, never after — a hole in a bent panel sits in the highest-stress zone of the sheet. Anneal after machining if the panel will be drilled within 15 mm of an edge or will be solvent-bonded, because cast sheet carries residual stress from manufacture that a drill turns into a crack starter. And leave at least 8 mm of material between the hole edge and the panel edge, with a 1 mm chamfer on the hole, so the bolt washer is not sitting on a knife edge.
Where the machine has a 3D model we work from the STEP file rather than from measurements, because a mounting face taken from the model is accurate to the model, and a face measured with a tape against a frame is accurate to about a millimetre. That difference is the difference between a panel that drops in and a panel that needs the frame re-drilled on site.
Buyers usually assume acrylic parts are priced by weight. They are not. On a typical guard set, sheet is 20–30 percent of the price; the rest is machining time, edge work, annealing and packing. Knowing the split tells you which design changes are worth asking for.
| Cost element | Share | What moves it |
|---|---|---|
| Sheet | 20–30 % | Grade (cast vs extruded, UV, ESD) and thickness; a jump from 5 to 10 mm roughly doubles material cost |
| Machining | 25–35 % | Outline length, number of holes, number of setups. Every extra setup is a re-clamp and a re-datum |
| Edge finishing | 10–25 % | Flame polish is fast and cheap; diamond polish is 2–3x the labour and only worth it on a visible edge |
| Annealing | 5–10 % | A 2–4 hour cycle plus a controlled ramp, charged per batch rather than per part |
| Packing | 5–15 % | Crate versus carton, foam versus film, and whether each panel ships individually |
The two levers worth pulling are part count and edge specification. A guard designed as one bent piece instead of three bonded pieces typically comes out 30–40 percent cheaper and is stronger, because the bonded joints were the weak points anyway. Specifying flame polish on the four edges that are visible and leaving the hidden edges sawn usually saves 10–15 percent against polishing everything, with no visible difference once the panel is on the machine.
What we do not recommend is saving on annealing. It is 5–10 percent of the price and it is the single thing that decides whether the panel is still intact in year two. On a 40-panel order the saving is a few hundred dollars; the failure it prevents is a line stop.
Guard enquiries that come back with a question instead of a price are almost always missing one of five things. Send these and the reply includes a price band and a schedule.
| Item | What we need | Why it changes the price |
|---|---|---|
| Opening size | Width and height of the frame opening, measured or from the model | Sets sheet thickness through the span rule |
| Hole pattern | Pitch and diameter, or the STEP file | Drilling on your pitch avoids a second setup and a re-drill on site |
| What can hit it | One line on the hazard: mist, chips, or a moving mass | Decides acrylic versus polycarbonate, which is a 1.4–1.8x material difference |
| Access | Fixed panel or door, and how often it is opened | A door needs a hinge, a stay and one step up in thickness |
| Environment | Coolant type, cleaning agent, static-sensitive electronics nearby | May require a chemical soak test or a dissipative grade |
If you have a photo of the machine with a tape measure against the frame, that is usually enough to quote. If the guard is a replacement for one that failed, send a photo of the failure — where a crack started tells us more about the correct specification than the drawing does.
The acceptance points below are recorded per batch, and the record travels with the packing list.
For a fixed fence panel up to 600 mm span, 5 mm is normal and 8 mm is used where a tool change could swing a part against it. Doors over 800 mm wide go to 10 mm or get a bonded stiffener. Above 1,200 mm span, polycarbonate at the same thickness deflects less and is the safer choice.
A cobot is force-limited by design, so the enclosure usually protects people from the workpiece rather than from the arm. Acrylic at 5–8 mm is clear, cheap to replace and easy to drill for sensors. If the cell handles a loose heavy part or a spring-loaded tool, polycarbonate is worth the faster yellowing.
Yes. Send the frame drawing or a photo with three measurements — panel opening, hole pitch and frame thickness. We cut to the opening with 1 mm clearance and drill on your pitch. Panels for retrofits are normally shipped in 7–15 business days after drawing sign-off.
Yes, and it is the default rather than an extra. Send the post pitch and the hole diameter and the panels arrive drilled to match, so the retrofit does not need a drill on site. Where the frame is not square we work from the diagonals rather than the sides, because a fence that is 20 mm out of square will not accept a panel cut to the nominal rectangle.
One panel. Retrofit work is usually a handful of pieces rather than a production run, and there is no tooling to amortise, so a single guard is a normal order. The economic entry point is around six panels, where the setup and the crating are shared. Below that, the packing often costs more than the machining, and we will say so before quoting.
