I review incoming industrial equipment before it goes anywhere near the production floor. Roughly 200 robot arms and major components a year, across maybe a dozen suppliers. Over the last four years, I've rejected close to a fifth of first deliveries from new vendors — not because the arms were unsafe, but because the paperwork, cabling, or mounting flange didn't match what we'd signed for.

These are the seven questions that come up most when someone on our team asks me to look at an industrial robot arm sourcing decision. Short answers, in the order I'd ask them.

  • Is this an actual robot arm factory or a trading desk?
  • Which datasheet numbers should I trust?
  • Whose components are inside the arm?
  • How do I read the price quote?
  • What gets missed on the assembly line side?
  • What actually fails on first delivery?
  • How do I write acceptance criteria?

1. What does an "industrial robot arm factory" actually mean — and how do I tell one from a trading desk?

Plenty of suppliers describe themselves as an industrial robot arm factory when they're really a trading company with a showroom. That's not automatically disqualifying, but you should know which one you're dealing with, because it changes who owns your warranty claim and who carries spares.

What I ask for: a live factory walkthrough video with a visible timestamp, not an edited reel. The serial number and nameplate photo of a unit that actually shipped in the last 90 days. And the production floor of the model you're buying — not a demo bay set up for visitors.

A real factory can show you the test bench and tell you how many units run through it per shift. A trading company shows you a conference room and a sample arm. Ask for the raw video — or rather, ask for the raw video and a photo of the same unit's nameplate. Those two together are hard to fake.

2. Which numbers on a robot arm datasheet should I actually trust?

Four numbers drive the decision: payload, reach, repeatability, and cycle time. The rest is usually marketing.

The trap is that repeatability is quoted under ideal conditions. ISO 9283 defines how repeatability and path accuracy should be measured, and the number on the datasheet is typically taken at rated payload, rated speed, and a specific pose near the center of the working envelope. Move to full extension at max payload and that ±0.02 mm can quietly become ±0.05 mm or worse.

Two things to check before you compare two arms:

  • Rated vs. max payload. Rated payload is the number you can run continuously. Max payload is a peak figure, often 60–70% higher, and it's the one printed in bold.
  • Whether the degree-of-freedom count includes the wrist. "6-axis" should mean six driven joints. I've seen 4-axis + 2-axis wrist assemblies described as 6-axis units. Technically defensible, practically misleading.

If the supplier can't tell you the measurement standard behind the repeatability figure, treat that figure as decorative.

3. Do I need to know whose components are inside the arm?

Yes, and this is the question people skip. Ask for the BOM on the drivetrain: harmonic or RV reducer brand, servo motor brand, controller platform, encoder manufacturer.

This isn't about brand snobbery. It's about what happens in year four when a joint reducer fails. If the reducer is a branded, catalogued part, you or your integrator can source a replacement in weeks. If it's an unbranded part made for that specific arm, your maintenance window depends entirely on the original supplier's willingness to sell you one part.

I'm not saying unbranded is automatically bad — some of it holds up fine. But you should make that choice with your eyes open, not discover it during a line-down event.

4. How do I read an industrial robot arm price quote without getting surprised later?

I've learned to ask "what's NOT included" before I ask "what's the price." That single question has saved us more money than any negotiation.

Typical exclusions on robot arm pricing: the controller, the teach pendant, interconnecting cables, base mounting plate, crating, freight, import duties, on-site commissioning, operator training, spare parts kit, and warranty labor. Any one of those can add 5–20% to the number you thought you were agreeing to.

So when you compare quotes, compare landed cost, not line price. The vendor who lists every fee upfront — even when the total looks higher — usually costs less in the end, because there's nothing left to negotiate after the PO is signed.

On actual figures: as of Q4 2024, quotes we've seen for 6-axis industrial arms in the 6–20 kg payload class ranged widely depending on origin, controller, and reducer brand. I'm deliberately not putting a single number on it, because the range is too wide to be useful and rates move. Get a written quote with a validity date.

5. What gets missed in an industrial robot arm assembly line project?

The arm is usually 30–40% of the project. The rest is end-of-arm tooling, fixtures, safety fencing or light curtains, PLC and cell integration, vision if you need it, and engineering hours from whoever is stitching it together.

Two things bite people. First, cycle time promises usually assume ideal part presentation. If your incoming parts vary in position by even a few millimeters, the gripper has to compensate, and that eats the margin. Second, mechanical interface mismatch. ISO 9409-1 defines the standard round mounting flange dimensions for robot arms, and I once said "standard flange" on a spec sheet and had a vendor interpret that as their own house pattern. The gripper we'd already machined didn't bolt on. That was a six-week delay.

Put the flange standard, the bolt circle, and the pilot diameter in writing. Every time.

6. What quality problems actually show up in first deliveries?

I don't have hard data on industry-wide defect rates — nobody publishes that honestly. But based on our incoming inspections over the past three years, the issues cluster in three places.

Cable and connector quality is number one. Harness routing that rubs against the housing, undersized strain relief, connectors that aren't the spec'd IP rating. This stuff passes a visual check and fails at month six.

Documentation is number two. Missing calibration certificates, a manual that describes the previous model revision, no wiring diagram for the controller I/O. Annoying, but it also tells you something about the factory's process discipline.

Cosmetics and hardware is third — paint runs, mismatched fastener plating, a nameplate with the wrong voltage.

Actual motion accuracy failures are rarer than people fear. The arm usually moves fine. It's everything around it that slips through.

Case in point: I skipped the incoming inspection on a repeat order once because, you know, "same model, same factory." One unit arrived with a different reducer series — same envelope, different backlash spec. That was the one time it mattered.

7. How do I write acceptance criteria that hold up?

Define the acceptance test before you place the order, and attach it to the PO.

What it needs to specify:

  1. The test program. Exact motion sequence, speeds, payloads, dwell times.
  2. The measurement method. Instrument type, measurement positions, and the standard you're referencing. ISO 9283 is the usual one for performance criteria; ISO 10218-1 and -2 cover safety requirements for the robot and the integrated cell — check you're on the current edition, since these have been revised and the older versions are still floating around in some contract templates.
  3. Sampling. Every unit or lot sample? If sampling, define the lot size and accept/reject number.
  4. Thresholds. Numeric pass/fail, not "acceptable quality."
  5. Remedy. Rework, credit, or replacement — and who pays freight.
  6. Dispute window. How many days after delivery you have to raise an issue.

The datasheet should be a version-stamped attachment to the contract, and the version you approved should be the version that governs. If a supplier sends a revised datasheet after the fact, that's a conversation, not an update.

If this is your first time sourcing robot arms, get the acceptance criteria reviewed before you get the quote. It's much cheaper to argue about a bolt circle on paper than on a loading dock.