How Hobbyists Are Sourcing Used Industrial Robots

There's a category of YouTube video that exists somewhere between a genuine how-to and a "look what I have" flex, and the Excessive Overkill channel has found a comfortable home there. The creator's latest — nearly 47 minutes of driving to Ohio, hauling robots on trailers, and debugging decades-old Fanuc controllers — is ostensibly an answer to the question he says he gets more than any other: where do you actually find industrial robots for cheap?

It's a fair question. Industrial robot arms show up in headlines as the thing that's going to take everyone's job, but in secondhand form they're largely invisible to anyone outside the manufacturing world. The video functions as a field guide to that gap — partly a flex, partly a genuinely useful map of a market most people have no idea how to access.

The market exists, and it's stranger than you'd think

The creator drives two hours to an Ohio robotics integrator called SixAxisSolutions, where an auction had cleared out a batch of machines. Most of the better-condition units were already spoken for. What he came back with were "parts robots" — units sold without controllers, missing pieces, probably flagged as broken before they were decommissioned.

That's the actual entry point for this hobby: not the clean machines under fluorescent lights, but the ones covered in grease in the outdoor lot.

The sourcing hierarchy he lays out is pretty sensible. Industrial auctions are accessible but come with friction — 15-20% buyer's premiums on top of the final bid price, plus loading fees or mandatory rigging companies that can easily cost more than the robot itself. Industrial surplus stores are easier to deal with. eBay and Facebook Marketplace occasionally surface deals. And then there's the top of the pyramid: "by far the most powerful source of used industrial goodies is knowing someone who happens to be getting rid of stuff." More often than you'd think, he says, decommissioned robots go straight to the scrap yard. Knowing the right people gets you a robot for free.

That last point is both the most useful and the most frustrating thing in the video. It's true — industrial equipment disappears into scrap constantly, and personal networks are genuinely how a lot of this stuff moves. It's also advice that only helps you after you've already spent years building the right contacts. Which, to be fair, he acknowledges. The network is the result of doing this for a while, not a shortcut into it.

What to check before you buy

No obvious damage to motors or encoders. Cut cables are a red flag — he notes with genuine exasperation that "apparently, the 5 seconds to unplug the cable is too long compared to the 3 seconds to cut it off." Tags on faulty equipment sometimes tell you what's wrong, sometimes just say "NFG." Open or missing access covers usually mean someone was already inside trying to fix something or cannibalize parts.

The serial number matching issue is one he learned the hard way. Controllers are built around specific robot configurations with software calibration baked in, and those settings aren't always in the standard documentation. He bought a Fanuc R-2000 years ago, got it paired with a controller from a different model robot, couldn't get it running — and that failure is apparently what kicked off his open-source controller project. The match matters. Check the numbers.

One counterintuitive point worth holding: a robot coated in grease and dirt isn't necessarily a bad robot. Some production environments just do that to machines regardless of condition. The creator points out that the worse something looks cosmetically, the less competition you'll face on price — if you're willing to clean it up and troubleshoot whatever's wrong, ugly machines are where the deals are.

You already know the three-phase power thing is going to be annoying

It is. Most industrial robots want three-phase power, which isn't standard in US residential wiring. The creator's solution is a rotary phase converter — essentially a three-phase motor spun up on single-phase that generates the remaining phases. It works, it's buildable from surplus parts, and it only gets you to 220V three-phase. Some machines want 480V or 560V, which requires additional step-up transformers, also sourced from industrial surplus.

The actual power consumption is more manageable than the supply requirements suggest. According to the creator, his large robots average around 2.5 kW — significantly less than their peak draw, since robots are "very peaky loads" that spike hard during fast moves and barely draw anything while stationary.

Safety, honestly

He gets points for not glossing over this. Industrial robots — anything that isn't specifically a collaborative robot designed to work around people — can cause serious harm. His words: it ranges from "breaking your hand to throwing you through a wall."

He also admits he doesn't run a full safety enclosure. His personal setup relies on staying in teach mode, which requires him to physically hold the pendant's safety switch for the robot to move. The moment he lets go, it stops. It's a reasonable personal practice. It is not the same as a proper safety cage with light curtains and access-controlled e-stops, which he says you should absolutely build if you're running the robot in auto mode or if anyone else might be in the space.

One specific, useful piece of advice: buy safety-rated e-stop buttons from actual automation suppliers, not Amazon, not AliExpress. He's seen those fail. "I've seen way too many of those fail for me to bet my life on them" is the kind of sentence that should appear early in anyone's research on this.

Getting a decades-old Fanuc to move

The commissioning sequence is the most entertaining part of the video and also the most educational. The robot that came with its original controller powers on fine. Immediately throws a wall of errors. All of them trace back to dead encoder batteries — the small lithium cells that keep the motors' position memory alive when the controller is unpowered. Standard issue on any Fanuc that's been sitting for a while, apparently not a serious problem.

Getting those batteries replaced becomes a whole sequence: improvised battery pack from mismatched holders, swapped connectors, a brief acrid smell, a melted contact, and eventually a bench power supply standing in for the encoder backup circuit. Then more errors, a software configuration rabbit hole involving system variables with names like "op work you op disable," and eventually — all six joints moving.

The moment he cranks it to 100% speed and screams is genuinely funny. The joints are loud in a way that sounds alarming but apparently isn't — older servo drives run at lower switching frequencies, which just makes them noisier. Not a sign of damaged gear reducers.

The hardware teardown of the failed components is genuinely informative. Harmonic drives — the gear reducers used in smaller robot joints — fail catastrophically when they've been through a crash. The flex spline that gives them their precision just buckles. There's no repairing it. He pulls apart a motor with a "bad brake" that turns out to have a failed rotor: magnetic dust has separated from the rotor body and jammed against the stator. Also unrepairable. The parts-robots he bought will get stripped for usable components, and he'll build as many functional machines as the good parts allow.

What this is actually building toward

The open-source controller project is the through-line underneath all of this. Every robot without a working controller — which is most of what he buys, because they're cheaper — is currently waiting on that project. He's building custom servo drives capable of running three motors per board, getting PCBs made through a sponsor, working through hardware bugs in real time on camera. It's real engineering progress, documented openly.

Here's my actual read on where this goes: the open-source controller reaches the Discord-and-Patreon crowd and stops there. That's not a knock — it's a description of what it is. The people who can follow this project well enough to use the controller when it's done are already the people who know where to find surplus robots, who can wire a rotary phase converter, who understand why matching serial numbers matters. The information gap doesn't close so much as it moves up one level of abstraction. Two years from now, there will be a working open-source controller with documentation good enough for a technically fluent hobbyist to use — and the bottleneck will shift from "I can't run this robot without a $10,000 Fanuc controller license" to "I can't find the mating connectors" or "I can't source replacement harmonic drives" or some other piece of institutional knowledge that lives in forums and Discord servers and nowhere else.

The secondhand industrial robot market is real and more accessible than it looks. The knowledge required to navigate it is also real, and it doesn't live in any manual Fanuc will give you for free.