Sourcing CNC Components for Industrial Robotics in Vietnam: A Field Guide for Procurement and Manufacturing Engineers

The global supply base for robotics components is rebalancing. Robot builders and integrators who once specified suppliers in eastern China are now expected to qualify a second source in Southeast Asia, and Vietnam consistently lands on the shortlist for precision machined parts. The CNC market here is large, but the slice of it that can actually deliver robotics-grade components is narrow. The criteria that distinguish a good general-purpose machinist from a good robotics supplier are not obvious from a website or a factory tour.

This guide is to equip the person on the buying side who has been handed the task of qualifying a CNC supplier and who wants to know what to specify, what to ask, and what to expect back. It assumes you already understand your own product. What follows is the supply-side perspective from a manufacturer that runs robotics work day to day.

What "robotics CNC" actually means at the component level

Robotics is a misleading category for procurement because the parts inside a robot vary enormously in precision, function, and difficulty. A floor plate for a cell enclosure and a harmonic-drive flexspline are both "robotics components," but one is a sheet-metal job and the other is a hardened-steel grinding operation at the edge of what most facilities can do. Lumping them together in an RFQ tells the supplier almost nothing useful.

In practice, the CNC content of an industrial robot or robot cell clusters into a handful of part families:

• Actuator and joint housings are the most common, typically machined from 6061 or 7075 aluminum, with bearing seats that must hold concentricity within a few microns and surface finishes that drive seal life.

• Gearbox-adjacent components such as output flanges, pre-heat-treat gear blanks, and tone-ring carriers sit in a higher precision band and almost always involve grinding after heat treatment.

• End-effector parts including gripper jaws, custom flanges, and quick-change couplings are smaller-volume but higher-mix, where a supplier's setup time matters more than cycle time.

• Mounting hardware, sensor brackets, cable carriers, and base plates round out the bill of materials. They are the easiest parts to source and, for that reason, the parts buyers most often use to test whether a supplier can deliver consistency before being trusted with harder work.

  
  

What the supplier actually needs from your RFQ is the part family and the function the part performs in the assembly. A good supplier will ask follow-up questions on the first call about thermal environment, load case, mating components, and expected lifetime. A supplier who quotes the drawing without any of those questions is telling you something worth listening to.

Material specification is the first real decision

Material choice is where most of the friction between buyer and supplier originates, because the drawing usually inherits a material callout from a previous program and nobody revisited whether it still makes sense for the new application.

6061-T6 is the default for the majority of aluminum robotics parts. It machines predictably, anodizes well, has adequate strength for most housings and brackets, and is widely stocked in Vietnam in plate, bar, and other shapes. The friction shows up when a part inherits a 6061 callout but actually needs the strength of 7075-T6: joint housings under repeated dynamic loading, end-effector bases that cantilever heavy tooling, and any part that is locally thin for mass reduction. The raw material cost delta is real but rarely program-breaking. The bigger cost is qualifying the change after the fact when a 6061 part deflects in service.

Stainless and tool steel become the right answer when wear, corrosion, or fatigue dominates. Shafts and gear components are the obvious cases. Less obvious are end-of-arm tooling parts that contact wet or chemical environments in food, pharma, or battery assembly cells, where anodized aluminum looks fine on day one and pits within a quarter. If your robot cell is going into one of those environments, flag it on the RFQ and let the supplier propose the material, rather than copying the standard aerospace-jig-shop callout.

What the drawing says vs. what the part needs

Drawings inherited from prototype work often carry tolerances tighter than the function requires, and drawings inherited from cost-cutting exercises often carry tolerances looser than the function requires. Both are expensive for different reasons, and the supplier you want is the one who reads the drawing against the application and pushes back where it matters.

Achievable tolerances in a competent Vietnamese facility are not the bottleneck people sometimes assume. A well-equipped 5-axis line can hold ±0.005 mm on critical features routinely, and tighter with grinding operations. The harder question is whether the manufacturer holds the same tolerance on part #1 and part #2,400 of a production run. That is a process control question, not a machine capability question, and the two require different evidence to evaluate.

Surface finish callouts deserve their own attention because they are routinely under-specified on robotics drawings. A bearing bore left at the default Ra of a finish boring pass will not deliver the bearing life the calculation assumed. Anodize specifications similarly need more than "black anodize." They need a hardness class, a thickness, and a color tolerance if the parts will be visible to a customer or assembled into a matched set across multiple deliveries. Lot-to-lot anodize color variation is one of the most common quality complaints in robotics manufacturing, and it is preventable with the right specification on day one.

What to send your supplier on day one

A good RFQ packet for a robotics component does more than attach a drawing. It tells the supplier enough about the application that they can flag wrong decisions before the first chip is cut, and it gives them enough commercial context to quote a real number rather than a hedge.

The packet that gets the most useful response back contains:

• The drawing in a current revision, with critical features clearly identified rather than left for the supplier to guess from tolerance callouts alone.

• A short application brief: what the part does, what mates to it, what environment it operates in (temperature range, humidity, cleanliness, chemical exposure), and the expected service life.

• Annual volume and lot size assumptions, including whether you expect a single run or repeat orders, because the right manufacturing approach for 200 pieces once is different from 200 pieces every quarter.

• The functional acceptance criteria, separate from the drawing tolerances. If a feature is dimensioned at ±0.02 mm but the part fails functionally above ±0.01 mm, the supplier needs to know.

• The inspection expectations: first-article requirements, the sampling plan for production lots, and whether you want SPC data on critical features reported with each shipment.

• The secondary process expectations: heat treat specification, surface coating with hardness/thickness/color requirements, and any cleanliness or packaging requirements for the final part.

• Any prior supplier history on the part, including known failure modes, scrap reasons, or process improvements from previous production. This is the single most useful piece of context a new supplier can receive, and it is the one buyers are most reluctant to share.

  
  

A question to ask back in the first technical review is worth more than a dozen capability slides. Ask the supplier to see the inspection records for the first and last part of their last production run. Crisp answers mean the supplier has a process. Vague answers mean you will be debugging the process for them after the first production lot.

What the top manufacturers do differently

The Vietnamese precision machining base is deeper than it was five years ago, and the gap between the top-tier and the average manufacturer has widened. The top-tier facilities invest in 5-axis machining centers, multi-tasking turning, internal and cylindrical grinding, and wire EDM, and they run those machines under documented process control rather than operator instinct. They also build partner networks for casting, forging, heat treatment, anodizing, and coating that they manage as extensions of their own quality system, not as separate vendors the customer has to chase.

The practical implication for robotics buyers is that the supplier shortlist is shorter than the directory suggests, but the suppliers that make the shortlist can deliver work that competes with anywhere in the region. The qualification effort up front is real. The reward is a second source that holds up under volume and under the kind of program scrutiny robotics buyers run after the first field failure.