Fans, Power Supplies and Motor Drives: How Appliance OEMs Buy Harnesses

The harness is rarely the reason an appliance program is late, and it is frequently the thing that makes it late. A fan, a switch-mode power supply and a motor drive all have the same underlying shape from a purchasing point of view: a small number of expensive engineered parts, a moderate number of mechanical parts, and one or two internal wiring assemblies that cost a few dollars and are ordered as an afterthought. When an appliance OEM harness program for fans, power supplies and motor drives is run lot by lot — a purchase order raised when stock hits the reorder point, quoted fresh each time — the harness becomes the item that stops the line, because it is the only item nobody scheduled.

The fix is not a better supplier. It is a different buying structure. Appliance OEMs that run harness supply well treat the harness the way they treat a stamped bracket: a controlled part number with a drawing, a revision history, a forecast against it, and a release mechanism that does not require a new negotiation every eight weeks.

Why lot-by-lot buying fails specifically on harnesses

Three properties of a wire harness conspire against spot buying, and none of them are obvious from the unit price.

Its bill of materials is long and shallow. A ten-circuit loom for a motor drive might carry four wire types, six terminal part numbers, two housings, a length of sleeving, tape and a label. Each of those has its own minimum, its own lead time and its own allocation risk. The assembly takes days; assembling the inputs takes weeks. A supplier quoting you a two- or three-week production lead is quoting from stock or from a standing component position — the moment your order needs a terminal nobody has, the quoted lead time stops describing reality. That gap is the subject of the lead time and connector allocation article.

Its cost moves with copper and with labour, on different clocks. Copper is a traded commodity that reprices daily; assembly labour reprices annually, in steps, and regionally. A spot quote blends the two into one number and hides both. Programs that survive separate them: a labour and overhead element fixed for a period, and a metal element indexed to a published copper reference with an agreed reset cadence. The cost anatomy is broken out in the harness quote cost drivers piece, and the reason two suppliers quoting the same drawing land far apart is treated in why China quotes for the same spec differ.

It carries tooling that is specific and cheap enough to be forgotten. Every terminal in your harness runs on an applicator — a die set fitted to a crimp press for that specific terminal. Applicators cost real money, sit on the supplier's floor, and are usually paid for once and never mentioned again. Whose asset they are, and whether they can be moved, is a question your program should answer at quote stage rather than during a dispute; see applicator and die ownership on harness orders.

The part number is the contract

The single largest improvement most appliance buyers can make costs nothing: stop buying descriptions and start buying part numbers with revisions.

A harness part number is properly defined by four documents, and a supplier who has all four can build the same thing twice. A supplier working from a photograph and a sample cannot, and neither can their second shift.

• The assembly drawing — physical layout, branch points, dimensioned lengths between breakouts with tolerances, tape and sleeve coverage, tie positions, and the free length that decides whether the loom meets its mating connector once the chassis is closed.
• The wire list — every circuit as a row: from-pin, to-pin, wire size, insulation type, colour, and any twisting or shielding requirement. This is the document that makes a harness testable, because it is what a continuity fixture is programmed from.
• The connector and terminal BOM — housings, terminals, seals and cavity plugs by part number and revision, with approved alternates listed explicitly rather than left implied.
• The workmanship and test standard — the class of workmanship you are buying, and the electrical tests you require on every unit.

That last one is where appliance buyers most often leave money and risk on the table. Harness workmanship has an industry standard with defined acceptance classes; naming the class you are buying on the purchase order turns a subjective argument about a stray strand into a document lookup. The selection logic is in choosing a workmanship class for the purchase order. Pair it with a stated electrical test regime — continuity on 100% of units, and insulation resistance or a withstand-voltage check where the application justifies it, as discussed in the harness electrical testing article. For control-cabinet work, add wire numbering and ferrule marking to the drawing package; it is the difference between a serviceable cabinet and a puzzle, and it is covered in the wire numbering and ferrule labelling piece.

What goes into the enquiry itself — and what a supplier needs before they can quote rather than guess — is set out in the harness RFQ package article and, more generally, in writing a product spec sheet for a Chinese factory.

Blanket orders, call-offs and where the buffer sits

Once the part number is stable, the buying structure decides your lead time far more than the factory does. Three structures dominate this category, and they trade money against responsiveness in different places.

The important word in that table is liability. A buffer only exists because somebody is paying to carry it. The negotiation that matters is not "will you hold stock" — most suppliers will say yes — but "what happens to that stock if I change the design or stop ordering". Write the answer down: a defined buffer quantity, a defined component liability window, and an agreed treatment of obsolete material on cancellation or engineering change. Without that clause, "we hold stock for you" is a courtesy that evaporates the first time it costs the supplier money.

Whether any particular supplier operates call-off scheduling or carries buffer stock is not something you should assume from a catalogue. Published trading terms usually describe a spot transaction — a stated minimum, a sample dispatch window and a production lead time — and say nothing about programme structures. The profile for Huisheng Electronics, for instance, states an MOQ of 1,000 pieces, sample dispatch in 7–10 days and 15–25 days to warehouse, and describes six product lines with OEM, ODM and build-to-drawing service, including multi-branch looms and control-cabinet wiring for machinery, fans, power supplies and motor drives — the company's own catalogue places its products into exactly those applications. What that published information does not tell you is whether a blanket order with call-offs is available, what notice a call-off needs, or whether component stock would be staged against a forecast. Those are questions to put in writing, not inferences to draw. The same profile notes that stated capacity figures are supplier-stated and should be confirmed in an audit or a live video walkthrough before you commit to a schedule, which is the correct order of operations for any programme decision. How much published MOQ and lead-time data is worth in general is discussed in china factory sourcing data on MOQ and lead times.

Engineering change on a running part number

This is the section most appliance OEMs write only after it has cost them something. A fan or a power supply is not a static product: a connector goes end-of-life, a motor supplier changes a terminal, EMC testing sends you back for a ferrite, or cost reduction moves two circuits from 0.5 to 0.35 mm². Every one of those is a harness change on a part number that is already in production, already in transit and already in your finished goods.

An uncontrolled change is the most expensive quality event in this category, because the failure is silent. A harness that is wrong in a way you can see gets caught at goods-in. A harness with the correct connectors and one circuit re-routed passes every incoming check you have and fails in the field.

Control comes down to four mechanics, all of which are ordinary and none of which happen by default with a supplier who has never been asked.

1. Revision in the part number. The revision must appear on the label, on the packing list and on the invoice — not only on the drawing. If your goods-in system cannot see the revision, you do not have change control, you have hope.
2. A written change notice with an effectivity point. Not "from next month" but "from serial batch X" or "from PO line Y", agreed in writing before the change is cut in. Effectivity is what lets you plan the drain-down of old stock.
3. A stated position on unauthorised substitution. The supplier must not change a wire, a terminal, a housing or a supplier of any of them without notice — including changes their own sub-supplier makes to them. Ask specifically how the supplier learns about, and passes on, a sub-supplier change. That is the leak most programmes never close.
4. Re-qualification scope agreed in advance. Decide now which changes require a new first article and a fresh test report and which do not. A colour change probably does not. A terminal change does, because it changes the crimp, and the crimp is the joint that fails.

Ask, too, what the supplier does with the old revision's material. A change that leaves 4,000 obsolete housings on their floor will come back to you as a price conversation, and it is cheaper to agree the principle before the change than after. The introduction discipline that surrounds all of this — first article, pilot lot, sign-off gates — is the same discipline used on board assemblies, and the new product introduction article is a reasonable template to adapt. The realistic calendar from approved sample to steady production is sketched in sample to mass production timeline, and the approval loop specific to cable assemblies in the sample approval article.

Common questions

How many harness part numbers justify a programme rather than spot buying?

Fewer than most buyers assume. The trigger is not part-number count but repeat frequency: two reorders a year on the same drawing is enough to make a blanket order worth negotiating, because the saving is in avoided requotes, staged components and stable price, not in volume discount. A single part number ordered quarterly benefits more than ten part numbers ordered once.

Can I put several low-volume harnesses under one minimum?

Sometimes, and the answer depends on shared content rather than on goodwill. Looms that share wire types, terminals and housings and differ in length or branch count are genuinely cheaper to batch than looms that need a different connector family, because the second case adds purchased-part minimums rather than just a changeover. Go into that conversation with a grouping already proposed — the arithmetic is worked through in splitting a 1,000-piece harness MOQ across SKUs.

Should the harness supplier hold my connector stock, or should I?

Whoever can carry it at lower cost and lower obsolescence risk — which is usually the supplier for common parts and you for anything specified uniquely for your product. The failure mode to avoid is neither party holding it while both assume the other is. Write the position for each connector part number into the programme document, and revisit it whenever an engineering change touches that part. Wider supply-continuity thinking is set out in china supply chain risk management.

What to ask the supplier next

1. Programme structure: whether a blanket order with call-offs is available, what release notice a call-off requires, and over what horizon a forecast must be given.
2. Buffer and liability: whether finished or component stock can be staged against a forecast, at what level, and who owns it if the design changes or the programme ends.
3. Change control: how a revision is carried on the label, packing list and invoice; how effectivity is agreed; and how the supplier notifies you of a change made by their own sub-supplier.
4. Re-qualification scope: which categories of change trigger a new first article and a fresh test report, agreed in writing before the first change happens.
5. Test regime per lot: what is tested on 100% of units versus by sample, and whether the report ships with the goods by default.
6. Applicator and tooling position for each terminal in your BOM: who owns it, where it sits, and what happens to it if you dual-source.
7. Component position: which terminals and housings in your BOM the supplier stocks routinely and which are bought to order — this is your real lead time.
8. Price mechanics: how the copper element is indexed, to what published reference, and how often labour and overhead reset.
9. Capacity for your programme, expressed as build stations and hours available rather than as a monthly figure — and confirmed on a live walkthrough rather than taken from a catalogue.
10. A second source plan for the two or three part numbers that would stop your line, prepared before you need it rather than during the week you do.

Take that list into the first commercial conversation rather than the third. A supplier that answers items one through four crisply is a supplier who has run programmes before; one who treats them as unusual questions is telling you something useful about how your part number will be managed once the novelty of the first order has worn off. The same discipline applied to a smaller, higher-churn product family is covered in the aftermarket clip-and-plug lead article.