How to Select High-Power LEDs for a 300–1200 nm Solar Simulator
A specification guide for light-engine integrators and whole-vehicle solar simulation chamber builders
Prepared for publication on the Heyi Light Control corporate website · Product line: LED solar simulator light engines and OEM/ODM high-power LED emitters
Publishing metadata Suggested URL slug:
/technology/300-1200nm-solar-simulator-led-selection/Meta title: 300–1200 nm Solar Simulator LED Selection Guide Meta description: How to select high-power LEDs for a 300–1200 nm solar simulator: the six IEC 60904-9:2020 bands, package choice, thermal and binning limits, and a procurement checklist. Primary keywords: solar simulator LED, 300-1200nm LED solar simulator, high power LED for solar simulator, IEC 60904-9 Class A+, LED solar simulator manufacturer Recommended schema: TechArticle + FAQPage
Quick answer: the six decisions that determine whether your simulator reaches Class A+
If you only read one section, read this one.
- Plan the LED set against the six IEC 60904-9:2020 spectral bands — 300–470, 470–561, 561–657, 657–772, 772–919 and 919–1200 nm — not against the older 400–1100 nm scheme, whose band shares were unequal. Each of the six bands carries roughly the same share of AM1.5G irradiance (16.6–16.7%), so each band needs its own radiometric budget.
- Count channels, not just watts. A practical LED engine uses 18–30 independently driven channels: 2–4 peak wavelengths per band. The 300–470 nm band is the most demanding because LED emission bandwidth scales with wavelength — a 400 nm emitter is only ~40 nm wide, a 1000 nm emitter is ~100 nm wide.
- Convert irradiance into radiant flux before you talk to any supplier. Radiant flux = target irradiance × illuminated area. Everything downstream — die count, package type, cooling, driver current, cost — follows from that number, not from a datasheet’s nominal “1 W”.
- Choose the package by thermal path, not by footprint. Ceramica 3535/5050 for high-flux channels; COB where single-point colour mixing matters; avoid epoxy-lensed through-hole parts in the high-UV channels.
- Specify wavelength bins of 5 nm or tighter and demand the junction-temperature wavelength coefficient. An A+ spectral match certified at 25 °C will drift out of tolerance on a hot array if you mix wide bins or ignore thermal shift.
- For whole-vehicle chambers, decide up front how you handle the 1200–4000 nm tail. AM1.5G integrates to 1000.4 W/m² across 280–4000 nm (ASTM G173). A 300–1200 nm LED engine matches the photovoltaic-relevant window, but it does not reproduce the long-wave infrared thermal load that DIN 75220 style chambers also characterise.
Why 300–1200 nm is the correct design window in 2026
The third edition of IEC 60904-9, published in 2020, changed how a simulator’s spectrum is judged. The older edition assessed spectral match from 400 A 1100 nm. Edition 3 assesses 300 A 1200 nm, because modern cell architectures — perovskite–silicon tandems, HJT, CIGS, CdTe — respond further into the blue and the near infrared than plain crystalline silicon.
Two consequences matter for an LED buyer:
- The bands were redrawn. Six bands of near-equal irradiance share replaced the older six bands of unequal share.
- A new top class appeared. Class A+ is roughly twice as tight as Class A on all three criteria, and it can only be claimed if the simulator is evaluated across 300–1200 nm.
IEC 60904-9:2020 classification thresholds
| Class | Minimum range evaluated | Corrispondenza spettrale, all bands | Spatial non-uniformity | Instabilità a breve termine (STI) | Instabilità a lungo termine (LTI) |
|---|---|---|---|---|---|
| A+ | 300–1200 nm | 87.5–112.5% | 1% | 0.25% | 1% |
| UN | 400–1100 nm | 75–125% | 2% | 0.5% | 2% |
| B | 400–1100 nm | 60–140% | 5% | 2% | 5% |
| C | 400–1100 nm | 40–200% | 10% | 10% | 10% |
The six bands and their AM1.5G share
| Band | Wavelength interval | Share of AM1.5G irradiance (300–1200 nm window) |
|---|---|---|
| 1 | 300–470 nm | 16.61% |
| 2 | 470–561 nm | 16.74% |
| 3 | 561–657 nm | 16.67% |
| 4 | 657–772 nm | 16.63% |
| 5 | 772–919 nm | 16.66% |
| 6 | 919–1200 nm | 16.69% |
Two reporting metrics introduced in edition 3 are not yet part of the class letter, but they are the fastest way to separate a genuinely good LED engine from one that games the band ratios:
- Spectral coverage (SPC) — the share of 300–1200 nm where the simulator emits at least 10% of the reference irradiance. Higher is better.
- Spectral deviation (SPD) — the total absolute departure from AM1.5G. Lower is better.
An LED array can pass Class A+ on band ratios while leaving visible holes inside a band, because a single strong emitter can supply a whole band’s irradiance from a narrow slice of it. SPC and SPD expose exactly that failure mode. When you write a purchase specification, put SPC and SPD in it.
Step 1 — Build the wavelength plan before choosing any part number
Do not start from a catalogue. Start from the band table, and give every band a channel budget.
The governing physical constraint is LED linewidth: full-width at half-maximum is typically about 10–20% of the peak wavelength. UN 385 nm UVA emitter is narrow; a 1050 nm NIR emitter is broad. This means the short-wave bands need more, narrower emitters and the long-wave bands can be filled with fewer, broader ones.
Reference wavelength plan for a 300–1200 nm engine
| Band | Typical peak wavelengths to combine | Channels | Notes for procurement |
|---|---|---|---|
| 300–470 nm | 365, 385, 405, 420, 450 nm | 4–6 | Highest technical risk. Deep-UV flux and package reliability dominate. Quartz optics, ceramic substrate, no epoxy in the optical path. |
| 470–561 nm | 470, 490, 505, 530 nm | 3–4 | Blue–cyan fill matters for the 470 nm band edge. |
| 561–657 nm | 570, 590, 617, 630 nm | 3–4 | Amber/red gap here is a common cause of low SPC. |
| 657–772 nm | 660, 680, 700, 730 nm | 3–4 | 730 nm is essential for the 772 nm edge. |
| 772–919 nm | 780, 810, 850, 880 nm | 3–4 | Cheap, efficiente, high-flux; good place to put redundancy. |
| 919–1200 nm | 940, 970, 1050, 1200 nm | 3–5 | Must reach a real 1200 nm emitter, not a 1050 nm tail. Silicon detectors fall off near 1100 nm, so plan an InGaAs measurement channel too. |
This table is a design reference, not a fixed bill of materials. The exact peak wavelengths follow from your target spectrum, your spectrometer calibration, and the spectral response of the devices under test.
| Modello del prodotto | Product size | Color | Dominant wavelength | Peak Wavelength | Potenza ottica | Voltaggio | RdiantIntensity | Half-wave widthity | SE (mA) | Visualizzazione | ||
| (mm) | nm/CIE | nm/CIE | IV(mW) | VF(V) | (mW/sr) | (nm) | Angolo(º) | |||||
| min. | massimo. | min. | massimo. | |||||||||
| HYF50P40D250AG-X4A | 5.2*5.2*4.0 | ●UVC LED | 250-260 | 300-350 | 10-12 | 110.4 | 11 | 1000 | 60 | |||
| HYF50P40F260AG-X4A | 5.2*5.2*4.0 | ●UVC LED | 260-270 | 210-250 | 10-12 | 117.3 | 11 | 1000 | 60 | |||
| HYF50E41F270AG-X4A | 5.2*5.2*4.0 | ●UVC LED | 270-280 | 450-550 | 10-12 | 205 | 11 | 1400 | 60 | |||
| HYF50P40F280AG-X4A | 5.2*5.2*4.0 | ●UVB LED | 280-290 | 280-330 | 10-12 | 99.7 | 13 | 1000 | 60 | |||
| HYF50P40F290AG-X4A | 5.2*5.2*4.0 | ●UVB LED | 290-300 | 470-520 | 10-12 | 153.3 | 13 | 1000 | 60 | |||
| HYF50P40F300AG-X4A | 5.2*5.2*4.0 | ●UVB LED | 300-310 | 600-700 | 10-12 | 142 | 13 | 1000 | 60 | |||
| HYF50P40F320AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 320-330 | 470-520 | 10-12 | 173.3 | 13 | 1000 | 60 | |||
| HYF50K45B350AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 345-355 | 1600-1800 | 8-9 | 652.8 | 10 | 1000 | 60 | |||
| HYF50G45N360AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 360-365 | 3500-4200 | 6.8-7.6 | 3577 | 9 | 1400 | 60 | |||
| HYF50X45B365AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 365-370 | 5500-6500 | 6.8-7.6 | 5622 | 9 | 2000 | 60 | |||
| HYF50X45B380AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 380-390 | 5500-6500 | 6.0-6.8 | 5576 | 11 | 2000 | 60 | |||
| HYF50X45B395AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 395-400 | 5500-6500 | 6.0-6.8 | 5398 | 13 | 2000 | 60 | |||
| HYB50G45N400AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 400-405 | 5000-6000 | 6.0-6.8 | 4066 | 15 | 2000 | 60 | |||
| HYB50G45N410AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 410-420 | 4000-4500 | 6.0-6.8 | 3442 | 17 | 1400 | 60 | |||
| HYB50G45N420AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 420-430 | 4000-4500 | 6.0-6.8 | 2646 | 17 | 1400 | 60 | |||
| HYB50G45N430AG-X4A | 5.2*5.2*4.0 | ●UVA LED | 430-440 | 4000-4500 | 6.0-6.8 | 2221 | 19 | 1400 | 60 | |||
| HYB50X38B450AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 450-460 | 450.4 | 7000-8000 | 6.4-7.2 | 7202 | 20 | 3000 | 60 | ||
| HYB50G45P460AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 460-470 | 459 | 2500-3000 | 6.0-6.8 | 2825 | 20 | 1400 | 60 | ||
| HYB50G45P470AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 470-480 | 467 | 2300-2800 | 6.0-6.8 | 2062 | 25 | 1400 | 60 | ||
| HYB50G45P480AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 480-490 | 479 | 1800-2200 | 6.0-6.8 | 2133 | 24 | 1400 | 60 | ||
| HYB50G45P500AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 500-510 | 498.8 | 1200-1500 | 6.0-6.8 | 1430 | 25 | 1400 | 60 | ||
| HYB50G45P520AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 520-530 | 514.5 | 1200-1500 | 6.0-6.8 | 1941 | 27 | 1400 | 60 | ||
| HYB50G45P530AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 530-540 | 522.6 | 800-1000 | 6.0-6.8 | 1358 | 30 | 1400 | 60 | ||
| HYB50G45P540AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 540-550 | 538.1 | 500-600 | 6.0-6.8 | 1868 | 36 | 1400 | 60 | ||
| HYB50G45N550AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 550-560 | 541 | 600-700 | 5.2-6.0 | 1074 | 35 | 1400 | 60 | ||
| HYB50F45N560AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 560-570 | 555.1 | 700-900 | 4.8-5.6 | 2296 | 40 | 1400 | 60 | ||
| HYB50E42N580AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 580-590 | 584.6 | 200-300 | 4.4-5.2 | 310 | 15 | 1400 | 60 | ||
| HYB50E42N590AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 590-600 | 595.2 | 350-450 | 4.4-5.2 | 910 | 15 | 1400 | 60 | ||
| HYB50G42N600AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 600-610 | 608 | 1000-1200 | 4.4-5.2 | 2003 | 15 | 1400 | 60 | ||
| HYB50G42N610AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 610-620 | 615.2 | 1000-1200 | 4.4-5.2 | 1711 | 26 | 1400 | 60 | ||
| HYB50E42P620AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 620-630 | 634.1 | 1700-2000 | 4.4-5.2 | 3778 | 17 | 1400 | 60 | ||
| HYB50E42P630AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 630-640 | 646.1 | 2100-2500 | 4.4-5.2 | 5421 | 15 | 1400 | 60 | ||
| HYB50E42N660AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 655-665 | 2500-3000 | 4.4-5.2 | 11696 | 17 | 1400 | 60 | |||
| HYB50G42N670AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 670-680 | 2500-3000 | 4.0-4.8 | 14571 | 21 | 1400 | 60 | |||
| HYB50G42N680AG-X4A | 5.2*5.2*4.0 | ●Visible LED | 680-690 | 2500-3000 | 4.0-4.8 | 9101 | 27 | 1400 | 60 | |||
| HYB50G42N700AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 700-710 | 2200-2600 | 4.0-4.8 | 1462 | 24 | 1400 | 60 | |||
| HYB50X45N720AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 720-730 | 3000-3500 | 4.0-4.8 | 1963 | 21 | 2000 | 60 | |||
| HYB50E42N730AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 730-740 | 3000-3500 | 4.0-4.8 | 2134 | 37 | 2000 | 60 | |||
| HYB50G42N740AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 740-750 | 2500-3000 | 4.0-4.8 | 1264 | 22 | 2000 | 60 | |||
| HYB50E42N750AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 750-760 | 2500-3000 | 4.0-4.8 | 1789 | 24 | 2000 | 60 | |||
| HYB50G42N760AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 760-770 | 2500-3000 | 3.2-4.0 | 1717 | 24 | 2000 | 60 | |||
| HYB50G42N780AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 780-790 | 3000-3500 | 4.0-4.8 | 1922 | 23 | 2000 | 60 | |||
| HYB50G42N800AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 800-810 | 2000-2400 | 3.2-4.0 | 1872 | 30 | 2000 | 60 | |||
| HYB50G42N810AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 810-820 | 2000-2400 | 3.2-4.0 | 1985 | 30 | 2000 | 60 | |||
| HYB50G42N820AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 820-830 | 2500-3000 | 3.2-4.0 | 2025 | 35 | 2000 | 60 | |||
| HYB50E35N850AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 840-860 | 2500-3000 | 3.2-4.0 | 1598 | 35 | 2000 | 60 | |||
| HYB50G42N880AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 880-890 | 2500-2800 | 3.2-4.0 | 1563 | 39 | 2000 | 60 | |||
| HYB50G42N900AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 900-910 | 2000-2400 | 3.2-4.0 | 1460 | 33 | 2000 | 60 | |||
| HYB50G42N910AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 910-920 | 2000-2400 | 3.2-4.0 | 1169 | 39 | 2000 | 60 | |||
| HYB50G42N920AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 920-930 | 2000-2400 | 2.8-3.6 | 1368 | 47 | 2000 | 60 | |||
| HYB50E42N940AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 935-945 | 2000-2400 | 2.8-3.6 | 2069 | 40 | 2000 | 60 | |||
| HYB50E42N950AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 950-960 | 2000-2400 | 2.4-3.2 | 2383 | 47 | 2000 | 60 | |||
| HYB50G42N970AG-X4A | 5.2*5.2*4.0 | ●Infrared LED | 970-990 | 1500-1800 | 2.4-3.2 | 2489 | 50 | 2000 | 60 | |||
| HYF50H45N1000AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1000-1020 | 2000-2400 | 2.4-3.2 | 2416 | 45 | 2000 | 60 | |||
| HYF50G42N1020AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1020-1040 | 1500-1800 | 2.4-3.2 | 2091 | 55 | 2000 | 60 | |||
| HYF50E42N1040AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1040-1060 | 2000-2400 | 2.4-3.2 | 2466 | 55 | 2000 | 60 | |||
| HYF50E42N1060AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1060-1080 | 2000-2400 | 2.4-3.2 | 2440 | 50 | 2000 | 60 | |||
| HYF50X42N1080AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1080-1100 | 2000-2400 | 2.0-2.8 | 2403 | 55 | 2000 | 60 | |||
| HYF50X42N1100AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1100-1120 | 2000-2400 | 2.0-2.8 | 2345 | 55 | 2000 | 60 | |||
| HYF50G42N1120AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1120-1140 | 300-400 | 1.6-2.4 | 829 | 56 | 1400 | 60 | |||
| HYF50G42N1140AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1140-1160 | 300-400 | 1.6-2.4 | 836 | 65 | 1400 | 60 | |||
| HYF50G42N1160AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1160-1180 | 300-400 | 1.6-2.4 | 810 | 67 | 1400 | 60 | |||
| HYF50X42N1180AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1180-1200 | 700-900 | 1.6-2.4 | 1650 | 66 | 2000 | 60 | |||
| HYF50G42N1200AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1200-1250 | 300-400 | 1.6-2.4 | 822 | 68 | 1400 | 60 | |||
| HYF50G42N1250AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1250-1300 | 300-400 | 1.6-2.4 | 816 | 75 | 1400 | 60 | |||
| HYF50E42N1300AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1300-1350 | 500-600 | 1.6-2.4 | 1422 | 85 | 2000 | 60 | |||
| HYF50G42N1350AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1350-1400 | 300-400 | 1.6-2.4 | 787 | 83 | 1400 | 60 | |||
| HYF50G42N1400AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1400-1450 | 250-350 | 1.6-2.4 | 595 | 108 | 1400 | 60 | |||
| HYF50G42N1450AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1450-1500 | 250-350 | 1.6-2.4 | 694 | 115 | 1400 | 60 | |||
| HYF50G42N1500AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1500-1550 | 180-220 | 1.6-2.4 | 404 | 137 | 1400 | 60 | |||
| HYF50G42N1550AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1550-1600 | 100-140 | 1.6-2.4 | 341 | 143 | 1400 | 60 | |||
| HYF50G42N1600AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1600-1700 | 80-120 | 1.6-2.4 | 276 | 138 | 1400 | 60 | |||
| HYF50T42N1700AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1700-1800 | 180-210 | 1.6-2.4 | 531 | 129 | 2000 | 60 | |||
| HYF50T42N1900AG-X4A | 5.2*5.2*4.0 | ●SWIR LED | 1870-1930 | 140-180 | 1.6-2.4 | 404 | 109 | 2000 | 60 | |||
Step 2 — Choose the package by thermal path and optical role
Heyi Light Control supplies 3535 E 5050 ceramic SMD, high-density COB, 3528 SMD and 3 mm / 5 mm through-hole emitters across 200–2000 nm, so the selection question is genuinely about which format fits which channel — not about what is available.
| Pacchetto | Best role in a solar simulator | Strengths | Watch-outs |
|---|---|---|---|
| 3535 ceramic SMD | Workhorse high-flux channel, 1–3 W class | Low thermal resistance ceramic substrate, tight footprint repeatability, easy to array on MCPCB with predictable pitch | Single emitter per package, so channel count grows; needs good secondary optics for mixing |
| 5050 ceramic SMD | Higher-flux channels, multi-die options | More die area per footprint, better radiant flux per placement | Thermal pad design must be respected or junction temperature drifts |
| High-density COB | Single-point mixing, collimated engine cores, uniformity-critical designs | 6–16 wavelengths integrated on one board, secondary lens gives seamless mixing with no ghosting, high radiant density per unit area | Concentrated heat load — requires a real cold plate, not a heatsink; per-channel addressability must be designed into the driver |
| 3528 SMD | Low-power trim channels, 657–919 nm fill | Low cost, small pitch, good for closing narrow spectral gaps | Limited flux; not for the main radiometric budget |
| 3 mm / 5 mm through-hole | Legacy instruments, sensor reference lamps, low-flux retrofits | Familiar footprint, cheap, easy to socket and replace | Epoxy lens degrades under sustained high UV flux — do not use these as the primary 300–470 nm source in a Class A+ engine |
A practical architecture is hybrid: COB cores for the visible and NIR bands where mixing uniformity drives the A+ number, discrete 3535/5050 arrays for the UV band and for bands where you want independent tuning, and small 3528 or through-hole parts for gap filling.
Step 3 — Turn irradiance into radiant flux, then into electrical power
This is the step most specification documents skip, and it is where projects fail late.
Radiant flux required (W) = target irradiance (W/m²) × illuminated area (m²)
Electrical power (W) = radiant flux (W) ÷ wall-plug efficiency of the chosen emitters
Illustrative arithmetic — substitute your own measured values:
- UN 0.1 m² test plane (32 cm × 32 cm) at 1000 W/m² needs 100 W of optical power.
- UN 2 m × 3 m vehicle chamber test plane at 1000 W/m² needs 6,000 W of optical power.
- At 30% wall-plug efficiency, those become roughly 333 W and 20 kW of electrical input respectively.
Wall-plug efficiency is not a constant. It varies by wavelength — deep-UV and SWIR emitters convert far less efficiently than blue or NIR — and it falls as junction temperature rises. Ask every supplier for measured radiant flux at a stated drive current and stated junction temperature, and build your budget with the worst-case figure, not the marketing number.
That single number determines cooling architecture (passive heatsink vs. forced air vs. liquid cold plate), driver current per channel, and the number of emitters per band.
Step 4 — Lock the thermal design before you lock the spectrum
LED peak wavelength shifts with junction temperature, and the coefficient is material-dependent: InGaN violet and blue devices shift less than AlGaInP amber and red. In a solar simulator this is not a cosmetic issue — a wavelength shift moves irradiance between bands and changes your spectral match ratio.
Three consequences for selection:
- Treat thermal resistance as a first-class specification. Ceramica 3535/5050 packages in the Heyi high-power line are built on high-thermal-conductivity ceramic substrates with thermal resistance below 2.5 K/W; that figure, not the nominal wattage, sets how hard you can drive a channel.
- Ask for the measured wavelength-versus-junction-temperature curve for each wavelength family you plan to use, and integrate it into your control model. If a supplier cannot produce the curve, the part is not ready for a Class A+ engine.
- Design for thermal equilibrium, not for switch-on. Short-term instability of 0.25% (Class A+) is not achievable while the array is still warming up. Specify a stabilisation period, and require the driver to hold constant current while the cold plate holds constant temperature.
Step 5 — Demand tight wavelength binning, and bin by channel
Spectral match is computed per band, so binning discipline is where a cheap BOM quietly destroys an expensive calibration.
- Wavelength bin: 5 nm or tighter for the 300–470 nm band and for any band edge where you are close to tolerance. Wider bins may be acceptable in 772–919 nm, where the band is 147 nm wide and the emitter linewidth is broad.
- Flux bin: one flux bin per channel, or the drive current per channel has to be trimmed individually at production.
- Forward voltage bin: narrow Vf bins keep series-connected strings balanced and make constant-current drivers behave predictably.
- Traceability: every reel should carry wavelength bin, flux bin, Vf bin, lot and LM-80 reference. This is what lets you re-order the same engine two years later and keep the same classification report.
Step 6 — Driver and control architecture
The LED is necessary but not sufficient. Class A+ temporal instability of 0.25% is a system-level number, and it comes from three things:
- Independent, high-resolution dimming per channel. 16-bit or better resolution, so you can trim a band ratio by a fraction of a percent rather than by a visible step.
- Optical feedback. A monitoring photodiode per band, or at minimum per band group, closed on the driver. Open-loop current control cannot hold 0.25% over a multi-hour ageing test.
- Communication and logging. CAN, RS-485 or Ethernet control with a documented register map, so the spectrum recipe, drive currents and calibration date travel with the data. This is also what makes the engine reusable across an AM1.5G pump, a stabilised light-soaking mode, and a variable-spectrum research mode.
Step 7 — Do not buy emitters without a metrology plan
To claim anything under IEC 60904-9:2020 edition 3, you must measure to 1200 nm. Silicon spectrometers fall off near 1100 nm, so a single-channel Si measurement will silently under-report the top of band 6.
A credible verification chain has three parts:
- Spectroradiometer with a silicon plus InGaAs channel, calibrated and traceable, covering 300–1200 nm.
- Spatial uniformity mapping at the number of positions your target standard requires (ASTM E927-19 specifies a minimum of 64 positions for the uniformity calculation).
- A reference cell or detector for irradiance level, plus a black-panel or equivalent temperature reference for thermal-load work in chambers.
If your supplier’s classification report does not state which edition of the standard it follows, which spectral range was measured, and which detector technology was used above 1100 nm, treat the class letter as unverified.
Whole-vehicle solar simulation chambers: what changes
A whole-vehicle chamber is the same physics at a very different scale, and four constraints change the emitter selection.
1. Area scales the flux budget brutally. UN 6 m² test plane at 1000 W/m² is 6 kW of optical power. That is an array of hundreds of high-power emitters and tens of kilowatts of electrical input. Efficiency stops being a nice-to-have and becomes the dominant cost line — both capital and running.
2. Uniformity across metres is an optical design problem, not an emitter problem. At 1% spatial non-uniformity, array pitch, mixing distance, collimation and baffling dominate. This is where high-density COB cores with secondary optics earn their cost: a single mixing point with a tailored lens produces a smoother field than a sparse grid of discrete emitters.
3. The spectrum gap is real. DIN 75220 style vehicle ageing work characterises broad-band solar load, e CEI 60068-2-5 Test Sa describes simulated ground-level solar radiation across a broad spectrum. A 300–1200 nm LED engine is the right answer for the photovoltaic-relevant window and for visible/UV material degradation, but the long-wave infrared contribution above 1200 nm has to come from somewhere — either supplementary IR emitters or a hybrid lamp arrangement. Decide this in the architecture phase; retrofitting it later means rebuilding the optical bench.
4. Serviceability decides total cost of ownership. A vehicle chamber may run thousands of hours. Specify socketed or connectorised emitter modules, per-channel fault detection, and a spare-part wavelength bin commitment from the supplier, so a single failed channel can be replaced without recalibrating the whole field.
Heyi Light Control’s OEM/ODM model is built around this: custom wavelength sets from 200 nm a 2000 nm, ceramic and COB formats, and ready-to-ship standard wavelengths so that a production line is not held hostage by a single custom run.
Procurement acceptance checklist
Put these twelve items in the purchase order. They are the difference between a simulator that passes on the day of acceptance and one that still passes two years later.
| # | Item to specify | Why it matters |
|---|---|---|
| 1 | Target IEC 60904-9:2020 edition, target class per metric (per esempio. A+A+A+) | Edition 2 and edition 3 classes are not interchangeable |
| 2 | Peak wavelength and bin width per channel (≤5 nm in short-wave bands) | Band ratio tolerance is consumed by bin spread |
| 3 | Radiant flux at stated drive current and stated junction temperature | Nominal wattage is not radiant flux |
| 4 | Thermal resistance (K/W) and recommended Tc | Sets achievable drive current |
| 5 | Wavelength temperature coefficient, measured | Needed for spectral stability modelling |
| 6 | Spectral coverage (SPC) and spectral deviation (SPD) targets | Exposes band-ratio gaming |
| 7 | Spatial non-uniformity target and measurement position count | ASTM E927-19 requires ≥64 positions |
| 8 | Short-term and long-term instability targets and stabilisation time | A+ requires 0.25% STI |
| 9 | LM-80 / TM-21 data or an equivalent lumen and radiant flux maintenance curve | Predicts recalibration interval |
| 10 | Driver interface: canali, resolution, feedback, protocol, register map | Determines whether 0.25% is reachable |
| 11 | Optical materials in the UV path (quartz vs. polymer) | Polymer yellows and shifts the UV band |
| 12 | Re-order durability: bin commitment, lot traceability, change-notification clause | Protects the classification report over time |
Five failure modes that show up after commissioning
- The comb spectrum that still passes. A set of narrow emitters can hit every band ratio while leaving visible gaps inside each band. SPC and SPD catch it; a class letter alone does not.
- Bin drift on re-order. The second production batch uses a wider wavelength bin, the band ratio moves, and the original classification report no longer applies.
- Thermal creep. Junction temperature rises as the cold plate fouls, peak wavelengths shift, and long-duration tests drift out of tolerance while short tests still pass.
- UV path degradation. Epoxy or silicone in the 300–400 nm optical path yellows, cutting UV flux and destroying the short-wave bands first.
- Metrology blind spot above 1100 nm. The simulator is fine but the measurement is not, so band 6 is reported from a detector that cannot see it.
Eight questions to ask any LED supplier
- Which wavelength families do you manufacture, and what is your measured wall-plug efficiency at each?
- What is the actual wavelength bin width you can commit to, per family, in writing?
- Can you provide a measured peak-wavelength versus junction-temperature curve?
- What is the thermal resistance of the package, and what Tc do you recommend at my drive current?
- Do you have LM-80 or equivalent long-term radiant flux data, and over what duration?
- Which optical materials do you use in the UV path, and how do they age under high flux?
- Can you supply a mixed-wavelength COB with 6–16 channels on one board, and what is the die placement tolerance?
- What is your change-notification process when a die or phosphor changes?
A supplier who answers all eight in writing is a supplier you can put on a vehicle chamber programme. Heyi Light Control answers them per project, because spectral customisation is the core of the business rather than an exception.
FAQ
What LED wavelengths do I need for a 300–1200 nm solar simulator?
Plan for 18–30 channels grouped into the six IEC 60904-9:2020 bands. Typical peaks are 365/385/405/420/450 nm, 470/490/505/530 nm, 570/590/617/630 nm, 660/680/700/730 nm, 780/810/850/880 nm and 940/970/1050/1200 nm. The short-wave bands need more, narrower emitters because LED linewidth scales with wavelength.
Can an LED simulator reach Class A+?
Yes, on all three metrics, but only if it is evaluated across 300–1200 nm. Class A+ requires spectral match within 87.5–112.5% in every band, spatial non-uniformity of 1%, short-term instability of 0.25% and long-term instability of 1%. The hard parts are the 300–470 nm band and the measurement chain above 1100 nm.
Is 300–1200 nm enough for a whole-vehicle solar simulation chamber?
It is enough for the photovoltaic-relevant window and for visible and UV material degradation, but not for the long-wave infrared thermal load above 1200 nm that DIN 75220 style ageing work also characterises. Decide in the architecture phase whether you supplement with IR emitters or use a hybrid source.
3535 or 5050 or COB — which package should I use?
Use 3535 ceramic for general high-flux channels, 5050 where you need more die area per placement, and high-density COB where single-point mixing and uniformity are the binding constraint. Most real engines are hybrids. Avoid epoxy-lensed through-hole parts as the primary source in the 300–470 nm band.
How many LEDs do I need for 1000 W/m²?
Divide: required radiant flux = 1000 W/m² × illuminated area. UN 0.1 m² plane needs 100 W of optical power; a 6 m² vehicle chamber plane needs 6,000 W. Then divide by the measured wall-plug efficiency of each wavelength family to get electrical input, and size cooling from there.
Why do spectral coverage and spectral deviation matter if they are not part of the class letter?
Because band ratios average over a band, a narrow but intense emitter can satisfy a whole band while leaving the rest of it dark. SPC and SPD quantify what the class letter hides. A high SPC and a low SPD are the practical definition of a simulator that behaves like sunlight rather than merely scoring like it.
Does wavelength binning really change the classification?
Yes. Spectral match is computed per band, so a ±10 nm bin spread inside a band only 91 nm wide (470–561 nm) consumes a meaningful fraction of the tolerance before any thermal shift is added. Specify 5 nm or tighter in the short-wave bands.
What is the minimum verification setup I need in-house?
A calibrated spectroradiometer covering 300–1200 nm with both silicon and InGaAs channels, a uniformity mapping rig at the position count your target standard requires, and a reference detector for irradiance level. Without the InGaAs channel, band 6 is unverified regardless of what the report says.
Standards and references
- IEC 60904-9:2020, Photovoltaic devices — Part 9: Classification of solar simulator characteristics, IEC Webstore: https://webstore.iec.ch/publication/28973
- ASTM G173-03(2020), Standard Tables for Reference Solar Spectral Irradiances: Direct Normal and Hemispherical on 37° Tilted Surface: https://www.astm.org/Standards/G173.htm
- ASTM E927-19, Standard Classification for Solar Simulators for Electrical Performance Testing of Photovoltaic Devices: https://www.astm.org/Standards/E927.htm
- Pasan, IEC 60904-9 edition 3: 300 A 1200 nm, new bands, SPC and SPD: https://www.pasan.ch/systems/guides/iec-60904-9-edition-3-spc-spd
- G2V Optics, IEC 60904-9:2020 Standard for Solar Simulators: https://g2voptics.com/iec-60904-9-2020/
- DIN 75220, Ageing of automotive components in solar simulation units; SAE J2412 / J2527 for North American interior and exterior exposure practice; IEC 60068-2-5 Test Sa, Simulated solar radiation at ground level.
About the manufacturer
Heyi Light Control (Shenzhen Heyi Optical Control Co., Ltd.) manufactures high-power LED emitters and spectral modules across 200–2000 nm, compreso 3535 E 5050 ceramic SMD, high-density COB, 3528 SMD and 3 mm / 5 mm through-hole formats. The company supports OEM and ODM programmes with custom wavelength sets, mixed-wavelength COB integration and ready-to-ship standard wavelengths for fast sampling.
For product data, sample requests and spectral customisation enquiries, contact the team through the company’s Alibaba.com storefront: https://led-solarlight.com/
https://hongkongheyi.en.alibaba.com/
All standard thresholds and band shares quoted in this article are taken from the published standards referenced above. Application-specific parameters — thermal resistance, radiant flux and wavelength temperature coefficients — should be confirmed against the current datasheet and test report for the exact part number used in your design.
Modulo di scheda PCBA SIMULATORE SOLAR LED 200-1750NM
