3D Printingand the Power ofRapid Prototyping

Making custom parts. Improving robot mechanisms.

Evan Kim

Programming Lead, Aperture Science
FDM Printing Researcher

1 minute · 0:00–1:00

Open: “I’m Evan Kim, programming lead for Aperture Science and an FDM printing researcher. Today I’ll show how we can use 3D printing to make custom robot parts and improve them through testing.”

Connect with the room: Ask for a quick show of hands: “Who has used a 3D printer? Who has had a print finish perfectly and then discovered that the part didn’t fit?” Allow a few seconds, without opening a long discussion.

Set the goal: Explain that this is an introduction for FIRST Tech Challenge teams. The goal is to understand how a part gets made, recognize the design choices that matter, and shorten the time between an idea and a useful test. If you brought a printed part, hold it up and point to one custom feature.

Transition: “First, where does 3D printing fit among the ways we can make a part?”

Ways to manufacture a part

Subtractive

Remove material.

Milling · drilling · laser cutting

A milling cutter removing aluminum and producing metal chips

Additive

Build up material.

3D printing

A 3D printer nozzle depositing layers of orange plastic

2 minutes · 1:00–3:00

Start on the left: “Subtractive manufacturing starts with material and removes what we don’t need.” Point to the cutter and chips. Drilling removes material to make a hole; milling shapes stock with a rotating cutter; laser cutting cuts a profile from sheet material.

Click 1: reveal additive: “Additive manufacturing builds material into the shape we want.” Point to the nozzle laying down plastic. Compare a flat mounting plate cut from sheet stock with a printed sensor mount that wraps around another component. Choose the process around the geometry, equipment, and job.

Click 2: circle Additive: “This is our focus today: filament-based 3D printing.” Both processes can create waste: machining produces chips, while printing can produce supports and failed parts. Avoid implying that one process is always faster, stronger, or cheaper.

Transition: “Let’s look at what happens inside the printer.”

References: FDM printing photo via Obico; CNC milling photo via Roberson Machine Company.

What is FDM 3D printing?

A digital model becomes a physical part
one layer at a time.

01
Feed
Filament enters the print head.
02
Heat & deposit
A nozzle lays down softened plastic.
03
Repeat
New layers bond to the layers below.

Desktop FDM 3D printer with a filament spool, print head, and build plate

Diagram of an extruder and nozzle depositing filament in stacked layers on a build plate

This talk focuses on filament printing: FDM / FFF.

2 minutes · 3:00–5:00

Point out the hardware: Use the printer photo to identify the filament spool, print head, and build plate. “Filament is the plastic feedstock. The hot end heats it, and the nozzle deposits a narrow line along a planned path.”

Follow the diagram: Walk through Feed, Heat & deposit, and Repeat. Each layer creates another cross-section of the part, and the new material bonds to the layer below. Relative movement between the nozzle and bed builds the height of the object.

Explain the names: FDM means fused deposition modeling; FFF means fused filament fabrication. Both names refer to the filament process discussed here. Resin printing and other technologies work differently.

Plant two ideas: The printer follows instructions prepared by a slicer, which we will cover shortly. Also, a layered part can behave differently depending on the direction of a force. We will return to that when choosing print orientation.

Transition: “Now, where do these parts actually show up on an FTC robot?”

References: Prusa: Modeling with 3D printing in mind.

3D printing on an FTC robot

It’s everywhere!

Cyan robot mechanism with highlighted rack, gear, housing, and curved guides — **Gears, rack & guides**

Plywood robot with blue end covers, mounting brackets, and chassis corner parts highlighted — **Covers & mounting brackets**

FTC robot with green custom cradle, upright brackets, pulley-like components, and slide fittings highlighted — **Custom mounts & slide fittings**

3 minutes · 5:00–8:00

Invite observation: “It’s everywhere! Take a moment to look at how many different jobs these parts do.” All circles are visible immediately. Spend roughly 45 seconds on each photo, then connect the examples. Alt-click a photo if you want to zoom; Alt-click again to return.

Left: guide and transmit motion: Point to the cyan rack and gear, then the housings and curved guides. Explain how custom geometry can help position components or guide a game piece along a chosen path. Emphasize the function of a feature rather than naming every circle.

Center: connect an assembly: Point to the blue covers and mounting brackets on the plywood robot. Printed interfaces can connect materials and components with different shapes. A small custom bracket can solve a specific mounting problem without replacing the whole structure.

Right: custom interfaces: Point to the green upper brackets, central cradle, side pivot and pulley-like components, and the small fittings along the slides. These illustrate how custom printed geometry can connect and guide a mechanism. The highlighted parts appear to be printed based on their visible geometry; confirm the manufacturing method with the team before describing an uncertain component as printed. The metal rails, black tracks, cable carriers, and intake wheels are not highlighted.

Ask and transition: “Where on your robot do you need a shape that you can’t easily buy?” Take one quick example. “The first version of that shape probably won’t be the last. That’s where rapid prototyping helps.”

What is rapid prototyping?

Each version builds on what you learned.

Early blue Pentos cradle design with a simple flat center — Initial design

Orange Pentos cradle revision with changed mounting geometry — Revised geometry

Pentos V2 cradle with reinforcing ribs and an integrated bearing seat — PENTOS V2

Design→Build→Test→Revise→Repeat

3 minutes · 8:00–11:00

Define the idea: “Rapid prototyping means making a version quickly enough to learn from it, then using that result to improve the next version.” A prototype should answer a question about the design.

Walk left to right: Use the Pentos cradle images to trace the changing geometry: the simple blue design, the orange revision, and the more detailed PENTOS V2 with ribs and a central bearing seat. Point to one visible change at a time. If you share the project’s actual history, explain the specific observation that prompted a change. The images alone do not establish what failed or how much performance improved.

Read the cycle with an example: “Design: choose the geometry. Build: make a version. Test: check the question you started with. Revise: change the design using that result. Repeat: test the new version.” For a cradle, possible questions include whether the hardware fits, whether moving parts clear it, and whether it flexes too much.

Make learning reusable: Record the version, the test conditions, what happened, and what you changed. Keep comparable conditions between versions so the result means something. A version that reveals a problem can still be a useful prototype.

Transition: “Here is the workflow for making each printed version.”

From CAD model to physical part

1. Print the smallest thing that answers your question

A small test piece can answer a big design question.

Test only the critical section. Print the interface at its real dimensions before committing to the full part.
Quick tip: print one layer, then cancel. Check the footprint and hole positions. Use a deeper sample for actual fit.

Small printed sample marked 22.15 with a metal bearing fitted into its pocket — A small sample tests bearing fit.

2 minutes · 13:00–15:00

Start with the question: “Before printing a large part, ask what you need to learn first.” Point to the bearing sample. If the uncertain feature is a bearing pocket, print that region at its real size, with representative wall thickness and orientation.

Explain the payoff: A small sample lets you try nearby dimensions before committing to the full part. Keep the critical interface full size; shrinking the entire design changes the fit you are testing. The 22.15 label belongs to this sample and is not a universal bearing-pocket recommendation.

Give the quick tip: For a rough footprint or hole-position check, print one layer, cancel using the printer controls, and let it cool before removing it. Compare the outline with the assembly. First-layer squish and distortion can affect the result, so use a deeper sample to check an actual bearing fit or clearance through a hole.

Close and transition: “A small fit sample answers a fit question. It doesn’t prove the full part’s strength.” Next, explain how the direction of the printed layers affects that strength.

2. Orient your prints right!

Balance strength, support material, and surface finish.

Diagram comparing tension across layer bonds with tension along the layers

Respect the layers

Orient the part so working loads are less likely to pull layers apart.

Person holding a blue printed part with extensive tree supports underneath

Minimize supports

Rotate to reduce unsupported overhangs. Check cleanup time and access.

White printed part with visible extrusion lines on its top surface

Choose the top face

Keep critical faces clear of support contact. Inspect top layers in the slicer.

2 minutes · 15:00–17:00

First 40 seconds: load direction: Point to the layer diagram. “Think about which way the force will pull on this part.” Loads that pull layer interfaces apart can cause separation. Choose orientation around the actual load path, rather than assuming upright or flat is always best. The important direction is the force relative to the layers.

Next 40 seconds: supports: Point to the supported blue part. Rotate a model in the slicer before deciding where supports are needed. Compare unsupported areas, bed stability, print time, and access for removal. Supports can be necessary; reducing them should not sacrifice a suitable orientation for strength.

Last 40 seconds: functional surfaces: Point to the white part. Orientation determines which surfaces face upward, touch the bed, or contact supports. Supported faces can have marks, bed-contact faces pick up plate texture, and top faces show extrusion lines. Decide which face must slide smoothly or mate with another part, and preview it carefully.

Transition: “Even with a good orientation, two dimensions that match in CAD may not fit in real life.”

References: Prusa design guidance.

3. Measure real parts and test clearances

Rule of thumb: start with 0.2 mm clearance.

Start around 0.2 mm. A starting point for a well-tuned 3D printer.
Aim for a snug fit. Parts should go together without noticeable wiggle.
If fit matters, test it. Print a small sample and adjust the clearance.

Yellow and gray interlocking printed fit samples with different labeled clearances — Compare clearances. Choose the fit your assembly needs.

2 minutes · 17:00–19:00

Start with actual hardware: “Measure the part you have, then decide how the joint should behave.” A sliding joint needs room to move; a locating fit should limit wobble; a press fit has a different purpose. The slide’s snug-fit advice applies when that is the behavior you want.

Define 0.2 mm clearly: Treat 0.2 mm as a starting point to test, not a promise. For a round example, a 10 mm shaft in a 10.2 mm hole has 0.2 mm diametral clearance, or 0.1 mm per side. State which convention you are using before changing dimensions.

Use the sample photo: Point to the labeled fit samples. Print several nearby sizes and compare them with the actual hardware. Material, printer calibration, orientation, and geometry can change the result. Choose the fit that serves the assembly rather than automatically choosing the tightest one.

Keep the result: Record the successful dimension along with the printer, material, and orientation. Recheck when those change.

Transition: “Once the parts fit, we need a way to hold them together.”

References: Prusa design guidance.

4. Three ways to screw into printed parts

Design the fastening method into the part.

Captive nutTrap a metal nut in a shaped pocket so it cannot spin.
Heat-set insertHeat a threaded insert into the plastic for reusable metal threads.
Tapless / self-tappingUse a screw designed for plastic to form threads in a pilot hole.

Printed demonstration part containing captive nuts, brass threaded inserts, and screw holes

2 minutes · 19:00–21:00

Frame the choice: “Choose the fastening method while designing the part, so you can leave space for the hardware and tools.” Spend about 30 seconds on each option, then compare when you would use them.

Captive nut: A shaped pocket keeps a metal nut from turning. Leave a way to insert the nut, a route for the screw, and enough surrounding plastic to transfer the load. Think about whether the nut will fall out during assembly.

Heat-set insert: An insert provides reusable metal threads. Use an insert intended for heat installation in thermoplastic and follow its specified hole dimensions and installation guidance. Leave enough material around it and allow it to cool before loading the joint.

Tapless or self-tapping: Use a screw designed for plastic with the recommended pilot hole. It forms or cuts its thread without a separate tapping operation. This can be convenient for simple assemblies, but repeated removal can wear the plastic thread. Avoid overtightening.

Compare and transition: “If we open this assembly often, reusable metal threads may be useful. If access is limited, installation space may decide the method.” The photo shows a variety of features, not an exact three-item key. Next, choose the material around the job.

References: Formlabs: Adding screw threads to 3D-printed parts.

5. Match the material and settings to the job

Bambu filament examples · values apply to these specific products

Material comparison		PLA	PETG HF	TPU 95A HF
Material properties	Impact strengthXY · kJ/m²	26.6	31.5	123.2
	StiffnessBending modulus · XY · MPa	2750	2050	N/A in guide
	FTC applicationIllustrative uses	Rigid fit models	Functional mounts	Flexible contacts
Print preparation	Nozzle temperature	190–230 °C	230–260 °C	220–240 °C
	Dry before use	Optional	Required	Required
	Enclosure	Optional	Optional	Optional

Layer heightDetail & print time

WallsShell & load paths

InfillInternal structure

Source: Bambu Filament Guide. Bars compare values within each row. Use the matching printer and filament profile.

2 minutes · 21:00–23:00

First minute: choose a property: “What does the part need to do: stay rigid, absorb an impact, or flex?” Use rigid fit models, functional mounts, and flexible contacts as examples. These columns describe the specific Bambu products in the supplied guide, not every formulation of PLA, PETG, or TPU.

Explain the comparison: Impact strength and stiffness answer different questions. A flexible material can absorb an impact but still bend too much for a rigid bracket. The bars compare values within a row; do not compare bar lengths across different properties. XY describes the printed test orientation. These numbers do not directly predict the load capacity of the finished robot part. No stiffness value is supplied here for TPU.

Next 30 seconds: preparation: Use the matching printer and filament profile. The table gives the guide’s temperature ranges and drying guidance for these products. Follow the product’s drying instructions and check that the printer and feeder support the filament, particularly TPU.

Final 30 seconds: settings: Smaller layer heights generally add time while improving vertical detail. Walls form the shell and matter to many load paths. Infill supports internal regions and top surfaces. More infill alone will not solve weak geometry or poor orientation.

Transition: “Before spending that print time, inspect what the slicer actually generated.”

References: User-supplied Bambu Filament Guide, page 1; Prusa infill guidance.

6. Inspect the sliced file before pressing print

The preview shows what the printer will actually build.

Confirm the profile. Match the printer, nozzle, and filament.
Scan the layers. Check for missing walls and unsupported starts.
Inspect support contact. Check access for removal and surface finish.
Review time and material. Compare orientations before increasing speed.

Slicer preview showing the model, layer slider, filament usage, print-time estimate, and G-code

2 minutes · 23:00–25:00

Use the screenshot: Point to the layer slider, toolpaths, time estimate, and material use. Alt-click to zoom if needed. “The shaded model shows the intended shape. The layer preview shows the paths the printer is being asked to follow.”

Check the setup: Confirm the printer, nozzle, and filament profile. Then scan from the first layer upward. Look for a sensible bed footprint, thin features that disappear, and new regions that start without material or support underneath.

Inspect critical areas: Pause at mounting holes, thin walls, and the faces that must contact other parts. Find the support contact points and consider how you will remove the supports. An inaccessible support inside a feature can make an otherwise printable design difficult to use.

Compare before committing: Check predicted time and material use. If either seems excessive, compare another orientation or a smaller test section. Treat time as an estimate and balance it with strength and surface quality.

Transition: “Sometimes the best improvement is to change which parts of the assembly we print.”

References: Prusa design guidance.

7. Use 3D printing to its strengths

Custom printed geometry. Metal where the structure needs it.

Robot arm assembly with a long gold metal tube and black printed end components

3D-printed partCustom shape and mounting features.

Metal tubeA structural member for the arm.

Try cardboard before committing to a print.Use cardboard, foam board, scrap wood, or tape to test size and motion. Print when custom geometry adds value.

2 minutes · 25:00–27:00

Start with the unmarked photo: “This assembly combines a printed interface with a metal structural member.” Give the audience a moment to locate the two parts before revealing the circles.

Click 1: printed end component: Point to the orange circle. Explain how printing can combine custom shape and mounting features in one part. It also makes it easier to revise an interface when a neighboring component changes. Leave room for assembly, tools, and replacement of worn components.

Click 2: metal tube: Point to the blue circle. The tube provides a structural member for the arm, while the printed part handles the custom connection. The printed interface still needs appropriate material, orientation, and testing under its working loads.

Click 3: quick prototype materials: The captions change to the cardboard tip. “If the question is about size, reach, or motion, cardboard might give us an answer sooner.” Foam board, scrap wood, tape, and existing hardware can also help check a layout before detailed CAD and printing. A mockup checks those questions without establishing final strength or durability.

Transition: “Use printing where the geometry helps. Keep its practical limits in mind.”

What 3D printing can’t do

Practical limits of desktop FDM printing

Replace metal in every job

Loads, stiffness, wear, and heat may call for metal parts or standard hardware.

Print every shape unsupported

Overhangs may need supports, a different orientation, or a redesign.

Guarantee perfect fit and finish

Allow for process variation, layer lines, and cleanup. Test critical fits.

Prove that your design works

A finished print still needs testing on the robot under real conditions.

1 minute · 27:00–28:00

Move briskly through the four limits: These are limits of the desktop filament process covered today. Metal or standard hardware may suit stiffness, heat, wear, or other loading requirements. Difficult overhangs may require supports, a different orientation, splitting the part, or redesign. Critical fits and surfaces still need checking and sometimes cleanup.

Land on the last point: “A finished print still has to work on the robot.” Check assembly, motion, and performance under representative conditions. A part can look correct and still flex, bind, wear, or fail during use.

Close the teaching portion: “Choose a question, make a useful test, and use the result to improve the next version.” Return to the rapid prototyping cycle, then advance to questions.

References: Prusa design guidance; Bambu Filament Guide.

Thank you.

Questions?

fdm.evankim.me

2 minutes · 28:00–30:00

Invite questions: “Thank you. What are you working on, and where are you getting stuck?” Point to the QR code and fdm.evankim.me so people can open the linked site.

If the room is quiet: Ask: “Does anyone have a part that won’t fit, keeps breaking, or takes too long to print?” Pick one example and identify a useful next test using the ideas from the talk.

Keep answers concrete: For a fit question, ask about the hardware and measured dimensions. For a broken part, ask where it failed, the load direction, material, and orientation. For a long print, ask which feature needs testing first. When a setting depends on the printer or product, direct them to the matching profile rather than guessing a universal number.

Finish on time: At 30 minutes, thank the audience again. If questions remain, invite people to continue the discussion afterward.