Swirled 1000-1750nm LED
A reliable SWIR LED factory is not proven by a nice website, “1,000–1,700 nm” claim, or a sample that lights up. Evaluate the specific device, production line, and batch-control system.
Use this due-diligence scorecard:
| Area | What to verify | Strong evidence | Red flags |
|---|---|---|---|
| Technical capability | Exact wavelength, spectral width, radiant power, package, drive conditions | Measured spectrum and power-vs-current/temperature data for the exact part | Only peak wavelength stated; no test conditions |
| Chip source & process | In-house epitaxy/chip, outsourced die, packaging-only, or trader | Clear supply-chain disclosure, die traceability, wafer/die lot records | Claims to be “manufacturer” but cannot identify its manufacturing scope |
| Measurement system | Spectrometer, integrating sphere, calibrated photodiode, thermal test, reliability equipment | Calibration certificates, test SOPs, repeatability data | Handheld meter screenshots or no calibration traceability |
| Batch consistency | Bin definitions and outgoing limits | Data from 3+ production lots: wavelength, output, Vf, leakage, thermal resistance distributions | Only one “golden sample” report |
| Reliability | Lifetime under actual current and junction temperature | LM-80-style or equivalent stress report, HTOL, temperature/humidity, thermal cycling, failure analysis | “50,000 hours” with no test current, temperature, or degradation curve |
| Thermal design | Junction temperature control and mounting guidance | RθJC/RθJA, recommended PCB and heat-sink layout, derating curve | High-power claims with no thermal data |
| Optical design | Beam angle, uniformity, lens compatibility, optical power at working distance | Far-field pattern, angular intensity, application-level illumination data | Radiant-flux data treated as proof of system brightness |
| Quality system | Process control and traceability | ISO 9001, incoming/process/final inspection plans, lot-level CoC/CoA, CAPA process | Certificate only; no real quality documents |
| Commercial execution | Engineering support, MOQs, lead times, change control | Sample-to-pilot-to-mass-production plan; PCN policy; written warranty/RMA rules | Won’t commit specs or change notifications in writing |
| Compliance | Safety and restricted substances | RoHS/REACH reports; IEC 62471 assessment when relevant | Unrelated CE certificates used as product proof |
The most important distinction: many Chinese suppliers are competent package houses or trading companies, but that is different from owning epitaxy, chip fabrication, and full reliability engineering. Neither model is automatically bad; you just need pricing, continuity, IP, and quality expectations aligned with the real model.
For qualification, request these before placing a meaningful order:
- Datasheet with every measurement condition stated.
- Lot-level test report for at least three recent lots.
- Spectrum at nominal current and across operating temperature.
- Radiant power distribution, forward voltage distribution, and binning rules.
- Reliability report showing test conditions and degradation curves.
- Calibration certificates for key optical and electrical instruments.
- Factory audit evidence: production flow, ESD controls, traceability, incoming inspection, aging/burn-in, final inspection.
- PCN/change-control policy and written acceptance criteria.
- Ten to thirty samples from a normal production lot—not hand-selected engineering samples.
- A pilot order with your own incoming test plan.
Do not compare vendors solely by optical power. For SWIR illumination, validate at the intended wavelength, distance, beam geometry, duty cycle, sensor response, heat sink, and ambient temperature. A 1550 nm LED with higher radiant flux may still lose to a lower-flux 1300 nm part if your camera sensitivity, optics, or thermal constraints favor the latter.
A practical decision gate:
- Pass: 3 lots meet your spec, measurements agree within your and supplier’s uncertainty, and reliability/traceability documents are credible.
- Conditional pass: Sample works, but lot consistency or reliability evidence is incomplete; limit to engineering/pilot use.
- Reject: Supplier refuses raw test conditions, cannot explain its manufacturing role, or gives inconsistent answers about wavelength/power/lifetime.
For a China visit, spend more time in test, reliability, and outgoing-quality areas than in the showroom. Ask to trace one finished reel back through package lot, die lot, incoming inspection, process records, and final test data. If they cannot do that quickly, their claimed quality system is not mature.
