FRC Manufacturing and Fabrication: COTS vs Custom, Tools, Materials, and Tolerances

Every FRC robot is a negotiation between what you can buy and what you have to make. A modern drivetrain might be 90% commercial-off-the-shelf (COTS) gearboxes, wheels, and tube stock bolted together with hex hardware — and 10% custom brackets, gussets, and mounting plates that exist only for your specific geometry. Knowing where that line falls, and how to make the custom 10% correctly, is one of the biggest differences between a robot that comes together in the shop and one that fights you all season.

This guide walks through the COTS-versus-custom decision, the shop tools you'll actually use, when 3D printing is the right answer and when it's a trap, the handful of materials that cover almost everything in FRC, and the fastener, hole, and tolerance conventions that let parts from six different vendors bolt together without a fight. The goal is fluency in design for manufacture (DFM): designing parts that are cheap, fast, and reliable to actually build.

COTS vs custom: buy what you can, make what you must

The single most useful DFM rule in FRC is this: do not manufacture anything you can buy. A COTS gearbox from REV, WestCoast Products (WCP), AndyMark, or CTR Electronics has been designed, tolerance-analyzed, and tested by professional engineers and thousands of teams. A gearbox you machine yourself has been designed by a high-schooler under deadline. The COTS part is almost always lighter, more precise, and more reliable — and it frees your build team to spend its limited hours on the parts that are genuinely unique to your robot.

Buy COTS for:

• Gearboxes and reductions — VersaPlanetary, MAXPlanetary, and the integrated swerve/gearbox modules are cheaper and better than anything you'll build. See FRC gear ratios explained for how to pick the ratio.
• Swerve modules — MK4i, SDS, and the REV MAXSwerve are marvels of tolerancing; building your own is a multi-year R&D project.
• Wheels, bearings, sprockets, pulleys, belts, chain — standardized around 1/2" hex bore and 1.125" OD hex bearings.
• Motors and motor controllers — you literally cannot make these. See FRC motors: NEO, Kraken, Falcon.
• Structural tube — pre-drilled 2x1 aluminum box tube.

Make custom for:

• Mounting plates and brackets that tie COTS parts to your specific frame geometry.
• Intake rollers, hood plates, and mechanism-specific structure that has no standard equivalent.
• Anything where buying a part would force your whole mechanism to conform to someone else's dimensions.

A good heuristic: if you find yourself designing a part whose only job is to hold a motor at a specific angle relative to a specific tube, that's a custom part. If you're designing a part whose job is to reduce motor speed by 9:1, stop — buy it.

The FRC standard hole grid: why parts fit together

Almost the entire FRC ecosystem is built on a shared convention: #10 clearance holes on a 0.500" grid. Vendor 2x1 tube from WCP, AndyMark, and REV MAXTube all come pre-punched with #10 clearance holes spaced 0.5" apart along the 1" faces (per WCP and AndyMark product specs). Because everyone agreed on the same pitch, a gearbox from one vendor bolts onto a tube from another with no drilling.

This is why your custom parts should also live on the 0.500" grid wherever possible. When your bracket's hole pattern matches the tube it mounts to, you can bolt it anywhere along the tube and adjust later — hugely valuable when a mechanism ends up 3/8" from where the CAD said it would.

The dominant fasteners are #10-32 socket-head cap screws (SHCS) and, for higher loads, 1/4-20. #10-32 is a fine thread and 1/4-20 is a coarse thread in these common sizes. #8-32 shows up on smaller mechanisms and 3D-printed parts.

Common shop tools and what each is for

You do not need a full machine shop to build a competitive robot. Many top teams build almost everything with a drill press, a bandsaw, and a router. Here's what each tool does and where it fits.

Drill press

The workhorse. A drill press makes clean, perpendicular holes — far more accurate than a hand drill. With a machinist's vise and an X-Y table you can drill respectable hole patterns. Every team should have one. Use it for clearance holes, tapping guide holes, and light spot-facing. It's the first real tool a rookie team should buy after a bandsaw.

Bandsaw (horizontal and vertical)

A horizontal bandsaw cuts tube and bar stock to length with square ends — critical, because a tube cut 1/16" long throws off a whole frame. A vertical bandsaw cuts curves and profiles in plate. Bandsaws are safer and more forgiving than most tools, cut aluminum happily, and are indispensable for cutting 2x1 tube to length.

Router / CNC router

A CNC router (a gantry moving a spinning bit over a flat sheet) is the highest-value machine most FRC teams can own. It cuts 2D profiles — plates, gussets, gearbox side plates — out of aluminum sheet and polycarbonate with real precision. An Omio, Avid, or shop-built router lets you go from CAD to finished plate in minutes. This is how most teams make the flat custom parts that COTS can't cover. Feeds and speeds matter: aluminum needs the right bit, coolant/lubricant, and conservative depth of cut to avoid gumming up.

Mill (manual or CNC)

A milling machine removes material with a rotating cutter and can produce true 3D features: pockets, steps, precise bosses, and accurate hole locations. A manual mill (Bridgeport-style) or a small CNC mill (Tormach, Haas Mini) lets you make gearbox plates, custom shafts, and pocketed structure. Mills are the most capable metal-cutting tool but also the most expensive and the most demanding of skill. Most teams don't need one; those that have one use it for the 5% of parts that need real 3D precision.

Lathe

A lathe spins the workpiece while a cutting tool shapes it, producing round parts: custom shafts, spacers, standoffs, and stepped hubs. In FRC, most round parts (hex shaft, standoffs, spacers) are COTS, so the lathe sees less use than a mill. It earns its keep making custom shafts and turning down bearing seats.

Waterjet and laser

An abrasive waterjet cuts almost any material — thick aluminum, steel, polycarbonate — by firing a high-pressure stream of water and garnet abrasive. It produces excellent 2D profiles with no heat-affected zone. Most teams don't own one, but many have access through a sponsor or a service like SendCutSend. A fiber laser cuts steel and thin metals fast but is uncommon in FRC because it doesn't cut thick aluminum well (aluminum reflects the beam). For flat custom parts, "design it in CAD, send it to a waterjet/laser service" is a completely legitimate manufacturing strategy — see best CAD software for FRC.

Rule of thumb: most FRC custom parts are flat plates. A router or waterjet handles the vast majority. Reserve the mill and lathe for the genuinely 3D or round parts.

3D printing: powerful, and easy to misuse

3D printing (specifically FDM — fused deposition modeling, where a nozzle lays down molten plastic layer by layer) has transformed FRC. It's fantastic for parts that would be miserable to machine: intake rollers, complex brackets, sensor mounts, wire guides, belt tensioners, and one-off prototypes. But printed parts have real limitations, and rookies routinely print things that should have been aluminum.

The materials, and where each belongs

PLA's Tg of ~55–60 °C is the classic trap: a PLA part left in a car trunk on a summer day, or bolted next to a hot motor, can sag and fail. PETG is the sensible default for most functional FRC prints because it keeps most of PLA's printability while being much tougher and more heat-tolerant. Onyx (printed on Markforged machines many teams access through sponsors) is the go-to when a printed part needs to be genuinely structural — its carbon-fiber fill gives it a heat deflection temperature around 145 °C per Markforged's data sheet.

When 3D printing is right — and when it isn't

Good candidates:

• Complex geometry that's hard to machine (curved intake rollers, organic brackets).
• Low-load parts (sensor mounts, wire management, bumper brackets).
• Rapid prototyping — print it, test it, iterate before committing to metal.
• Compliant parts (flexible TPU wheels, grippers).

Bad candidates:

• Primary structure under high load — printed layers delaminate under shear and impact. A drivetrain gearbox plate should be aluminum, not PLA.
• Parts that take fastener preload directly — bolt tension crushes and creeps plastic over time. Use metal or brass heat-set inserts, and add generous wall thickness.
• Anything near heat — motors, especially, run hot.

Design printed parts for the process: orient the part so layer lines run perpendicular to the main load (layer adhesion is the weak axis), add fillets to spread stress, and use heat-set threaded inserts rather than threading into raw plastic. For load-bearing prints, thicker walls and higher infill matter more than exotic filament.

Materials: the short list that covers almost everything

You can build a superb robot from about five materials. Learn these deeply rather than chasing exotic alloys.

6061-T6 aluminum — the default

6061-T6 is the backbone of FRC. It has a yield strength around 40 ksi (~276 MPa) and ultimate tensile strength around 45 ksi (~310 MPa) (per aluminum property references), machines beautifully, welds, and is inexpensive. Nearly all FRC tube, plate, and stock is 6061-T6. Unless you have a specific reason otherwise, aluminum in FRC means 6061-T6.

7075-T6 aluminum — strong, pricey, fussy

7075-T6 roughly doubles 6061's strength: yield around 73 ksi (~503 MPa) and UTS around 83 ksi (~572 MPa). That lets you make thinner, lighter parts at the same strength. The tradeoffs: it costs several times more, it's not weldable in structural applications, it's more brittle (elongation ~9–11% vs 6061's ~12–17%), and it's less forgiving to machine. Reserve 7075 for weight-critical, high-stress parts — swerve forks, highly loaded gearbox plates, elevator carriages — where the strength genuinely pays for the cost. For most brackets, 6061 is the right call.

Polycarbonate (Lexan) — the tough, clear plastic

Polycarbonate is FRC's favorite plastic sheet. Its headline property is extraordinary impact resistance — it bends instead of shattering, which is exactly what you want for parts that get slammed in matches: intake plates, bellypans, guards, hoppers, and shields. It's optically clear, easy to cut on a router or even score-and-snap, and tolerant of drilling (though it can crack if you drill too aggressively near an edge — use sharp bits and back it up). It's not stiff like aluminum, so it flexes under load; use it where compliance is a feature, not where you need rigidity. Acrylic looks similar but is brittle and shatters — do not substitute acrylic for polycarbonate on a robot.

Delrin (acetal / POM) — the low-friction workhorse

Delrin (DuPont's acetal homopolymer, POM) is a rigid, dimensionally stable engineering plastic with a low coefficient of friction against metal. In FRC it's used for wear surfaces, bushings, low-friction guides, spacers, rollers, and gears. It machines cleanly, holds tolerances well, and slides against aluminum and steel without galling. When a part needs to slide or pivot smoothly without a bearing, Delrin is often the answer.

Steel — where you need hardness or strength in a small package

Steel is ~3× denser than aluminum, so it's a weight penalty you pay deliberately. It shows up in FRC as hardened shafts, gears, sprockets, and high-load hardware — places where aluminum would wear or yield. Most steel in FRC is COTS (hex shaft, gears, bolts). You rarely machine structural steel, but you rely on it constantly in the drivetrain. Grade matters: use proper alloy-steel SHCS (Grade 8 / class 12.9), not hardware-store bolts, for anything load-bearing.

Hole and tap standards

Getting holes right is where DFM becomes concrete. There are three kinds of holes you'll design.

1. Clearance holes — the bolt passes through freely. Size to the fastener with a little slop for alignment; a clearance hole is always larger than the screw's major diameter (0.164" for #8, 0.190" for #10, 0.250" for 1/4). Per standard tap/clearance charts:

FRC tube is punched with #10 clearance holes (nominally ~0.196–0.201") so #10-32 bolts drop through easily.

2. Tapped holes — you cut internal threads so a bolt screws directly in, no nut. The hole must be drilled to the tap drill size first (roughly 75% thread engagement):

Tapping into aluminum plate edges or gearbox plates is common and saves weight (no nut, no wrench access needed on the back). Tap straight — a crooked tap ruins the hole — and use tapping fluid on aluminum to avoid a broken tap.

3. Press-fit / bearing bores — holes sized so a bearing or dowel is retained by friction. A standard 1.125" OD hex bearing press-fits into a 1.125"-diameter bore cut slightly under for interference. This is where tolerance matters most (below).

Tolerances and design-for-manufacture rules of thumb

Tolerance is how much a real part is allowed to deviate from the CAD dimension. Tighter tolerances cost more time and money, so specify tight tolerances only where they matter.

• Bearing bores and press fits are the tight ones: a 1.125" bearing bore typically wants roughly a few thousandths of an inch of interference — too loose and the bearing spins in the bore; too tight and you can't press it in or you crack the plate. This is why teams cut bearing bores on a router or waterjet (repeatable) rather than a hand drill.
• Bolt-pattern holes can be looser — that's the whole point of clearance holes. A pattern that's a few thousandths off still bolts up fine.
• Cut-to-length tube wants square, accurate ends, but ±0.01" is plenty; frames tolerate a little slop because bolted joints have clearance.

Practical DFM rules of thumb:

1. Design on the 0.500" grid. Match the FRC hole standard so custom parts interchange with COTS tube and each other.
2. Prefer flat parts. A part you can make on a router or waterjet is faster, cheaper, and more repeatable than one needing a mill.
3. Don't over-tolerance. Call out tight tolerances only on bearing bores and mating surfaces; leave everything else loose.
4. Add fillets on internal corners. Sharp internal corners concentrate stress and crack; a router bit has a radius anyway, so design the radius in.
5. Leave wrench access. A beautifully placed bolt you can't reach with a tool is a bad bolt. Check assembly clearance in CAD.
6. Lighten intelligently, not everywhere. Pocketing and lightening holes save weight, but material near loaded holes and along bending axes earns its place. Use the FRC deflection calculator to see whether a lightened part still meets its stiffness target before you cut it, and mind the overall robot weight and size rules.
7. Prototype in cheap material. Cut the first version from scrap or print it in PLA to check fit before committing to 7075 or a long machine job.

Manufacturing well is mostly about restraint: buying the COTS parts that are better than anything you'd build, keeping custom parts flat and simple, choosing the ordinary material that's correct rather than the exotic one that's impressive, and specifying tight tolerances only where physics demands them. A team that internalizes those habits spends its build season assembling a robot instead of remaking parts — which, when you're staring down a build season timeline, is the whole game.

Frequently asked questions

Should a rookie team buy a CNC router or a mill first?

A CNC router, without question — if you buy any CNC at all. The overwhelming majority of FRC custom parts are flat plates and gussets, which a router cuts perfectly from aluminum sheet and polycarbonate. A mill is more capable in 3D but far more expensive and skill-intensive, and most teams never need its extra capability. Realistically, a rookie team's first three tools should be a horizontal bandsaw (cut tube to length), a drill press (clean holes), and a hand drill — you can build a competitive robot with just those plus COTS parts.

When should I 3D print a part instead of machining it from aluminum?

Print when the geometry is complex and the load is low: sensor mounts, wire guides, intake rollers, prototypes, and compliant grippers. Machine from aluminum when the part is primary structure, takes bolt preload directly, or sits near a heat source. Printed layers delaminate under shear and impact, and PLA softens above roughly 55–60 °C, so a hot, highly loaded printed part is asking to fail. Use PETG or Onyx (not PLA) for anything functional, and use heat-set inserts wherever a bolt threads into plastic.

When is 7075 aluminum worth it over 6061?

Only when weight is critical and the part is highly stressed. 7075-T6 has roughly double the yield strength of 6061-T6 (about 73 ksi vs 40 ksi), so you can make a thinner, lighter part at equal strength — worthwhile for swerve forks, elevator carriages, and heavily loaded gearbox plates. But 7075 costs several times more, isn't weldable, and is more brittle, so for ordinary brackets and structure, 6061-T6 is the correct, cheaper choice. Default to 6061 and reach for 7075 only when a weight budget forces it.

Why do so many FRC parts use #10 clearance holes on a half-inch grid?

It's a de-facto industry standard that makes parts from different vendors interchangeable. Pre-drilled 2x1 tube from WCP, AndyMark, and REV all come with #10 clearance holes on a 0.500" pitch, so a gearbox from one company bolts onto tube from another with no drilling. Designing your custom parts on the same grid means they mount anywhere along the tube and can be repositioned later — invaluable when a mechanism lands a few millimeters off from the CAD. The dominant fasteners are #10-32 and, for higher loads, 1/4-20 socket-head cap screws.

What tolerances actually matter on an FRC robot?

Almost none of them tightly — with one big exception: bearing bores and press fits. A 1.125" hex-bearing bore needs to be within a few thousandths of an inch or the bearing either spins loose or won't press in, which is why teams cut them on a router or waterjet rather than by hand. Bolt-pattern holes are deliberately loose (that's what clearance holes are for), and cut-to-length tube only needs square ends to about ±0.01". Spend your precision budget on mating surfaces and bearing bores, and let everything else run loose.