I ruined an Ultimaker 3 Extended print core in 2019 because I didn't know what a cold pull was. In 2022, I wasted $3,200 on an injection molding run because I'd never thought about gate design while prototyping. Same root cause, two different technologies: I treated the machine as the solution and ignored the process around it.
So here's the direct answer to anyone searching "what is cold pull in 3D printing" or wondering whether gate types matter for their prototype work: yes, they do, and both subjects are more connected than you'd think. Learn the cold pull procedure before your nozzle clogs, and learn the basic gate types before you approve a mold design. Both lessons are cheap if you learn them from my mistakes instead of your own.
I've been running FDM printers for manufacturing support since 2017, and specifically the Ultimaker 3 Extended since 2019. In that time, I've personally documented 17 significant printing mistakes — including the two above — that totaled roughly $6,800 in wasted budget. I now maintain our team's maintenance checklist, and the cold pull procedure is the first item on it.
What Is a Cold Pull in 3D Printing?
A cold pull is a nozzle-cleaning procedure. You heat the hotend to a temperature below the filament's full melt point, push a small amount of filament through, let it cool to a specific range, then pull it out with firm, steady force. The softened filament grabs contaminants, carbonized resin, and other residue that's been building up inside the nozzle and heat break. When you look at the tip of the pulled filament, you can literally see the dark crud that was restricting your flow and causing under-extrusion, stringing, or intermittent jams.
For an Ultimaker — and the 3 Extended specifically — here's the procedure I've settled on after refining it:
- Remove the loaded filament from the print core completely.
- Insert a cleaning filament or a piece of nylon (I've had the most reliable results with nylon; Ultimaker's own cleaning filament also works).
- Heat the nozzle to around 150°C. You want the filament soft, not fully melted.
- Push the filament down manually until you see the tiniest amount of material emerge from the nozzle tip.
- Wait for the hotend to cool to roughly 80–90°C. This takes a few minutes on the 3 Extended; don't rush it.
- Pull the filament out with a controlled, steady tug — not a violent yank.
The tip should come out shaped like a cone, ideally about the same diameter as the nozzle bore. If you see a blackened or discolored end, that's your problem — and your solution is the two minutes you just spent.
From the outside, a clogged nozzle looks like a hardware failure. The reality is that most of the time, it's a maintenance failure. I've caught 14 potential clog-related issues using this procedure in the past 18 months, and each one of them would've been a print core replacement if I'd skipped it.
Why This Matters More on the Ultimaker 3 Extended
The 3 Extended's dual-extrusion setup makes cold pulls more critical, not less. The second extruder is the one that sits idle most of the time. It picks up moisture. It accumulates heat creep residue. And it's the one that produces the most bizarre, inconsistent print defects.
I used to blame the print cores. A support engineer pointed out that idle extruders are exactly where routine cold pulls make the biggest difference. Now I have a rule: any time I switch from a high-temperature material (Nylon, Polycarbonate) back to PLA on the 3 Extended, I do a cold pull on both extruders. Any time the second nozzle shows signs of under-extrusion that isn't explained by slicer settings, I do a cold pull before I touch anything else.
People assume a print quality problem means the profile is wrong or the material is bad. What they don't see is that the nozzle was accumulating residue for days, and a two-minute cold pull would've prevented a two-hour troubleshooting session.
What Gate Types in Injection Molding Have to Do With Your 3D Printed Prototypes
This is the part where I hope someone learns from my $3,200 mistake.
In Q1 2024, I designed a housing for a client. I printed it on the Ultimaker 3 Extended, validated the fit with a 12-piece test batch, approved the design, and sent the files to an injection molding shop. The mold engineer asked what gate type I wanted.
I said "whatever's standard." Big mistake.
The edge gate they chose left a visible witness mark exactly where the assembly spec required a flush surface. The client rejected every single piece: 800 units, $3,200 of molded parts, plus the embarrassment of a missed deadline. That's on me. Gate placement affects everything — witness marks, weld lines, residual stress, even the dimensional stability of the finished part. The prototype I printed on the Ultimaker showed none of those issues because FDM doesn't have a gate.
Here's what I wish someone had explained before that happened:
- Edge gate: The most common and cost-effective. Molded along the parting line. Simple, reliable, leaves a small surface mark that you can trim. Fine for most parts — but think about where that mark lands.
- Submarine (tunnel) gate: Placed below the parting line and sheared off automatically during ejection. Leaves a tiny mark that's easy to hide. Slightly higher mold cost.
- Fan gate: Spreads material across a wide area. Good for large, thin parts. Requires a trim operation.
- Pin point gate: The cleanest option from an aesthetic standpoint. Leaves a very small mark. Needs an extra mold plate, so it's more expensive.
- Tab gate: Plastic enters a small tab that gets removed in post-processing. Useful when you want zero direct gate marks on the part surface.
If you're coming from a 3D printing background, the mental shift is this: in FDM, you can place seams, supports, and transitions anywhere, and you can always iterate. In injection molding, the gate determines weld lines and stress concentrations that are baked into every single part you make. You can't "post-process" those away, and you certainly can't fix them after a rejected production run.
There's a second assumption I'd like to save you from: I assumed "same material" meant identical results across suppliers. We approved a prototype in one filament, and the molded part — same nominal resin — behaved differently because the gate flow pattern created a weak weld line exactly where the part flexed. The 3D printed prototype passed every test I threw at it. The molded part didn't.
When an Ultimaker 3D Printer Is the Right Answer (and When It Isn't)
I'll keep this short and honest. For short runs, design validation, jigs, fixtures, and low-volume functional parts, a professional FDM printer like the Ultimaker 3 Extended is the right tool. I've printed custom end-of-arm tooling, assembly fixtures, and functional housings that went straight into production. For quantities under roughly 100–200 units, 3D printing is usually faster and cheaper than injection molding — period.
But if you're scaling to 5,000 or 50,000 units, injection molding wins on cost per part and material properties. I've seen colleagues try to 3D print their way through a production order that should have been molded, and it always ends with someone paying overtime labor to remove supports or sand away layer lines. Don't do that. This is where I want to be honest about my limits: I'm not the expert for ultra-high-volume production economics or multi-cavity mold design. My experience sits in the 10-to-10,000-unit range, mostly with engineering thermoplastics. If you're outside that range, talk to a mold engineer.
Boundary Conditions and the Advice I'd Push Back On
If you follow injection molding industry news today — and you should, because material pricing and supply chains shift faster than most engineers expect — you'll notice a lot of coverage of bio-based resins and sustainability initiatives. Gate design questions don't change there, but material behavior absolutely does. Bio-based resins often have different shrinkage characteristics and their own weld-line sensitivity. My experience leans toward petroleum-based materials, so I'd flag that as an area where I'd consult a specialist before making assumptions.
One more piece of counter-intuitive advice: you don't need to replace a print core every time it clogs. A cold pull will resolve a surprising number of what look like hardware failures. And print speed is overrated. I'd rather run a correct 16-hour print on the 3 Extended than a rushed 14-hour print that gets rejected at QC because of a weak interlayer bond. That's not a popular opinion in some circles, but I've measured the cost of rework, and nobody's paying me to be fast twice.
For reference: replacement print cores for the Ultimaker 3 Extended run roughly $80–180 depending on size (based on public retailer listings, January 2025 — verify current pricing). A cold pull costs five minutes and a few grams of filament. Do the math, and do the cold pull.
