What the ramp has to accomplish
Ramp-up takes a line that passed its pilot run at limited volume and brings it to the output your demand plan needs, without losing yield or quality on the way. Plan it as a curve with gates: each step up in volume happens when the line has shown it can hold yield and quality at the current step, not when the calendar says so.
Three things have to scale together: line capacity, yield and material supply. A line that can start 2,000 units a day, with parts for only 1,500 and a first-pass yield of 85%, delivers about 1,275 good units a day before rework. Work out which of the three limits you, and build the ramp plan around it.
Volume also changes the problems. Component lot variation, tool wear, newly hired operators and a second shift rarely show up in a pilot run. Expect new failure modes in the first weeks of mass production, and plan the people and the time to deal with them.
Planning the ramp curve
Work backward from demand. Start with the dates and quantities the business needs: launch date, channel fill and expected weekly sales after launch. Subtract transit, customs clearance, outgoing inspection and packing to find when each quantity must leave the line. That gives the required output curve.
Then plan starts with realistic yield, not target yield. Planned starts equal required good units divided by expected yield: 10,000 good units a week needs about 11,100 starts at 90% yield and about 10,500 at 95%. Planning at target yield from week one creates a shortfall in exactly the weeks that matter most for a launch.
A step plan keeps the ramp honest:
| Ramp step | Line state | Gate to the next step |
|---|---|---|
| First production lots | One line, one shift, below full rate | Yield at or above the pilot exit level; first outgoing lots pass; no open critical defects |
| Full rate, one shift | Line at target cycle time | Yield stable for an agreed number of consecutive days; output per hour at target |
| Added shifts or lines | Second shift or line running | New operators trained and signed off per station; any new line passes its own validation build |
| Steady state | Planned capacity | Weekly measures within agreed limits; ramp actions closed |
Treat a second line or a second factory as a new pilot run, not a copy of the first. Equipment, fixtures, operators and sometimes sub-suppliers differ, and its yield usually starts lower.
Automotive programs often formalize the step to full rate with a “run at rate” exercise: the line runs at planned speed for a defined period with production staff, and the output is counted and inspected. The method transfers well to consumer and industrial hardware, even without the automotive paperwork.
Capacity and materials
Line capacity
A line’s output is set by its slowest station. Calculate each station rather than trusting a single line-rate figure:
Station capacity per hour = parallel stations × 3,600 ÷ cycle time in seconds
A functional test that takes 90 seconds, run on two fixtures, handles 80 units an hour. If assembly runs at 120 an hour, test is the bottleneck, and the fix is more fixtures, not more operators. Steps with long, fixed cycle times are the usual suspects: functional test, firmware programming, calibration, adhesive curing and burn-in. Check those first.
Upstream processes need the same check. Molding capacity depends on cavities, cycle time and the press hours the molder actually commits to your parts. A second tool can take weeks to build, so order it from the forecast, not when the shortage appears.
Materials
A ramp can stall on one missing part while the line itself is ready. Ask the factory for a clear-to-build report: for each BOM line, the quantity on hand, on order and confirmed by delivery date, compared against the build plan. Review it weekly throughout the ramp. Look for:
- Long-lead parts, such as some ICs, displays and custom batteries, ordered early enough against forecast
- Minimum order and reel quantities that force larger purchases than the ramp needs
- Sub-tier suppliers of custom parts (molded, machined, die-cut, printed) that also have to ramp and may never have been checked
- Single-source parts with no qualified alternative
- End-of-life and change notices from component makers
- Who carries the material risk: in a turnkey model the factory buys, in consignment you do
Accessories, manuals and printed packaging are easy to forget and can stop a shipment as surely as a missing chip.
Yield learning
Yield often dips at the start of mass production, even after a good pilot run, because volume exposes variation the pilot never saw. Treat yield as something to improve on a plan, with the factory, every week:
- Track first-pass yield by station every day and the ranked defect list every week, as in the pilot run.
- Use parametric test data, not just pass or fail. A measurement drifting toward its limit shows a problem before it starts failing units.
- For critical dimensions and electrical parameters, track process capability (Cpk). Many quality teams treat 1.33 as a minimum for important characteristics; automotive programs often require more.
- Link failures to component lots through traceability, so a bad lot can be contained quickly.
- Set yield improvement targets with dates and owners.
The cost of a yield gap is easy to underestimate. At 10,000 units a week, each percentage point of yield is about 100 units of rework or scrap, every week.
Quality gates during the ramp
Incoming, in-process and outgoing inspection (IQC, IPQC and OQC) continue from the pilot run. The ramp is the time to tighten them, not relax them.
- Outgoing inspection. Sample each lot against an agreed AQL plan with defined defect classes. Critical defects, those affecting safety or regulatory compliance, normally have zero acceptance. Major and minor levels are set per product: AQL 2.5 for major and 4.0 for minor defects is a common starting point for consumer goods, and many electronics brands set tighter major levels such as 1.0 or 0.65. These are commercial choices, not fixed standards.
- Tightened inspection at the start. Larger samples, or 100% inspection of specific features, for the first lots, relaxed only on evidence.
- Independent inspection. Pre-shipment inspection by a third party gives you a record that does not depend on the factory’s own reports.
- Ongoing reliability testing. Pull a few units from production at a set frequency and repeat key reliability tests, to catch drift in materials and processes that functional test misses.
- Stop-ship rules. Agree in advance what holds shipments: any critical defect, a lot rejection, yield below a set threshold or a safety-related field report.
- Field feedback. Send returns and complaints back to the factory with serial numbers, so they can be traced to lots, shifts and stations.
Put these rules in the quality agreement before the first production order. The guide to building a practical hardware quality plan covers the wider structure.
Change control (ECN)
Once mass production starts, every change to design, material, supplier, process, firmware or test goes through a formal engineering change notice (ECN or ECO), whoever proposes it. That includes the factory. Factories and their suppliers sometimes substitute a component, move a process to another site or switch a sub-supplier to solve a shortage or a cost problem. The agreement should require written notice and your approval first.
A usable ECN states what changes and why; the affected part numbers, drawings and new BOM revision; effectivity, meaning the serial number, date or lot where the change cuts in; what happens to existing stock (use up, rework or scrap); the validation required and its results; the impact on certification and cost; and approvals from both sides.
The validation should match the risk of the change:
| Change | Typical validation | Certification question |
|---|---|---|
| Cosmetic: color, texture, print | Cosmetic review; new golden or limit sample | Usually none |
| Mechanical part or material | Fit, assembly, drop and relevant reliability tests | Does the material change a safety rating, such as flammability? |
| Electrical component substitution | Functional test, parametric comparison, targeted reliability tests | Does it touch power, radio, battery or emissions-sensitive circuits? |
| PCB layout | Repeat design validation for the affected functions | Often needs lab assessment, and possibly retest |
| Firmware | Regression test; version check on the line | Do radio parameters or safety functions change? |
| New factory, line or sub-supplier | Validation build treated as a new pilot | Is the certified configuration unchanged? |
The certification questions are the ones most often skipped. See product certification planning for why a change after approval can require new testing.
Supplier performance management
During the ramp, review supplier performance weekly; once it is stable, monthly or quarterly. A short scorecard is enough if the numbers are real:
| Measure | What it shows |
|---|---|
| On-time delivery against confirmed dates | Planning and capacity discipline |
| Lot acceptance rate and defect rate at your inspection | Quality as you receive it |
| Corrective action response time and closure | Whether problems get fixed or only contained |
| Field returns by root cause | Quality that inspection did not catch |
| Quote accuracy, change-order pricing, agreed cost reductions | Commercial behavior |
Set response times in the quality agreement, for example containment within a day or two and a root cause report within an agreed number of days, and follow each corrective action through to verified effectiveness (CAPA). The factory should pass the same expectations down to its key sub-suppliers, and you should be able to see the results.
Once the ramp is stable, the scorecard becomes the basis for business reviews and cost-down discussions. Cost reductions that change the product go through the ECN process like any other change.
Where Zefinity helps
During a ramp, problems tend to arrive together: a yield dip at one station, a late sub-supplier, a substitution request from the factory. Zefinity’s Launch service supports this stage with the OEM/ODM partner, coordinated from Hong Kong. The work includes supplier coordination and issue resolution, QA and corrective action follow-up, and supplier performance management once volumes settle.
Frequently asked questions
How long does a production ramp take?
It depends on product complexity, the demand curve and how clean the pilot run was. A simple product on one line may reach full rate within weeks; a complex product with long test times, several lines or new sub-suppliers can take a few months. Plan the ramp by gates rather than dates, and protect launch quantities by planning with realistic early yield.
What is a clear-to-build report?
A part-by-part comparison of material availability against the build plan: for every BOM line, the quantity on hand, on order and confirmed by delivery date, and the first build that would be short. It shows which part will stop the line and when, early enough to expedite, find an alternative or change the plan. Review it weekly during the ramp.
Can a factory substitute a component without approval?
It should not, and the supply and quality agreements should say so explicitly. Require written notice and approval before any change of component, material, sub-supplier, process or production site, including substitutions the factory considers equivalent. Some substitutions are harmless; others affect reliability, safety or certification, and only a controlled review tells you which.