Library › Part D · Article 31 in the series · Level 2 · 11 min read

Choosing and sizing grow lights

Standfirst: After reading this article you will be able to compare grow lights by the one number that matters for plants, photons per joule of electricity, and to work out how many watts a shelf or room needs for a chosen light level. You will also be able to estimate the daily electricity cost, the heat the lamp adds to the room, and whether a dearer, more efficient fixture pays for itself.

Before you read: Light: PPFD, DLI, spectrum and photoperiod.

A lamp is a photon source

Plants respond to the number of photons they absorb, not to the energy or the brightness of the light: one absorbed photon excites one electron¹. Photosynthetic photons are those between 400 and 700 nm, the band called photosynthetically active radiation, or PAR¹. A grow light is therefore best described by how many PAR photons it emits for each unit of electricity.

That number is the photon efficacy: photon output in micromoles per second divided by electrical input in watts. Because a watt is a joule per second, the unit is micromoles per joule, µmol J⁻¹¹. Efficiency, by contrast, compares watts of light out with watts of electricity in, as a percentage¹.

The two are linked by the energy of a photon, which is inversely proportional to its wavelength (Planck’s relation)¹. A red photon carries less energy than a blue one, so one joule can make more red photons. A perfect device would give 3.76 µmol J⁻¹ at 450 nm (blue) and 5.52 µmol J⁻¹ at 660 nm (red)¹.

Lumens and lux weight light by the sensitivity of the human eye, which peaks in the green^1,7. In 2014 tests of single LEDs, a cool white LED gave 111 lumens per watt and 1.52 µmol J⁻¹, while a red LED gave only 47 lumens per watt but 1.72 µmol J⁻¹². Even the factor that converts foot-candles to PAR changes with the light source⁷.

How efficient lamps are, and why

Efficacy has risen fast, so every figure needs its year. In a 2014 comparison of 22 fixtures, the best double-ended high-pressure sodium (HPS) lamps and the best LED fixtures both reached 1.66 to 1.70 µmol J⁻¹². Mogul-base HPS fixtures with magnetic ballasts, then in common use, gave 0.94 to 1.16, the best ceramic metal halide 1.46, and fluorescent fixtures 0.84 to 0.95².

By 2020 independently tested LED fixtures reached 2.5 to 2.8 µmol J⁻¹ for white plus red and 3.0 for blue plus red, while a 1,000 W double-ended HPS measured 1.72¹. Typical LED values were 2 to 3 µmol J⁻¹¹. In early 2022 the 600-plus LED products on the DesignLights Consortium’s qualified list ranged from 1.81 to 3.69 µmol J⁻¹, and only about 6% reached 3.0, according to a trade-magazine column by a Michigan State University extension specialist¹⁰.

Three mechanisms explain the order. First, red LEDs convert about 81% of their electricity into light at near-optimal current and 25 °C, and each joule makes many red photons: 4.5 µmol J⁻¹ against 3.5 for blue¹. Second, white LEDs are blue LEDs coated with a phosphor that re-emits some blue photons at longer wavelengths, a conversion only 80% to 95% efficient; warm whites reach about 2.6 µmol J⁻¹ and cool whites 2.9¹. Third, the fixture loses more: drivers (power supplies) are 85% to 95% efficient, a glass cover passes up to 92% of the photons, and efficacy falls about 10% as the LED junction warms from 25 to 85 °C¹.

With 2020 parts, the calculated ceiling for whole fixtures is 3.4 µmol J⁻¹ for white plus red and 4.1 for blue plus red, about 10% less with protection from water and humidity¹. Future gains will be smaller than past ones.

Where the watts go: heat

Every watt that does not leave the fixture as light becomes heat at the fixture. For an LED fixture at 2.5 µmol J⁻¹, the Planck ceilings imply that between 45% (all photons red) and 66% (all photons blue) of the electricity leaves as PAR; this is our calculation from¹. Older figures for discharge lamps are lower: 26% to 27% for sodium lamps and about 20% for fluorescent tubes⁵.

The light then becomes heat too. At best about 27% of the energy of an absorbed photon is stored as chemical energy in the leaf; the rest becomes heat⁴. Our inference from these two steps: nearly all the electricity a lamp uses ends up as heat in the growing space.

Where the heat appears differs. LEDs shed much of their heat by convection from heat sinks, away from the plants, while hot HPS lamps radiate long-wave heat toward the canopy³. At equal PPFD, modelled leaves under LEDs were about 1.3 °C cooler than under HPS indoors, and well-watered leaves under every source stayed within about 2 °C of air temperature³. So, as this article’s inference, a switch from HPS to LED may call for a slightly warmer air setpoint, which changes the vapour pressure deficit; see Temperature, humidity and VPD.

Sizing a fixture: the chain

Four inputs size a lamp: the target PPFD and photoperiod, which come from your crop (see Light: PPFD, DLI, spectrum and photoperiod), the lit area, and the fixture efficacy from its test report.

1. Photon flux needed (µmol s⁻¹) = PPFD (µmol m⁻² s⁻¹) × area (m²).
2. Electrical power (W) = photon flux ÷ efficacy (µmol J⁻¹).
3. Daily energy (kWh) = power (W) × hours of light ÷ 1,000.
4. Daily cost = daily energy × your electricity tariff.

Step 1 assumes every photon lands on the crop. Some always fall outside, so divide the flux by the fraction you expect to capture. This canopy photon capture efficiency can exceed 90% in large greenhouses with narrow aisles²; a single shelf loses a larger share at its edges.

One kilowatt-hour is 3.6 million joules, so a fixture of efficacy E delivers E × 3.6 moles of photons per kWh; divided by the tariff, that is moles per unit of money.

Worked example: a lettuce shelf

Take a shelf 1.2 m × 0.6 m, lit at a PPFD of 250 µmol m⁻² s⁻¹ for 16 hours a day. Romaine lettuce has been grown at this PPFD under red and blue LEDs in small household plant factories¹³. The tariff of EUR 0.25 per kWh is the reader’s own figure; replace it with yours.

• Area = 1.2 × 0.6 = 0.72 m².
• Photon flux = 250 × 0.72 = 180 µmol s⁻¹.
• Daily light integral = 250 × 3,600 × 16 ÷ 1,000,000 = 14.4 mol m⁻² d⁻¹ (the arithmetic of⁷).

Compare an LED fixture at 2.5 µmol J⁻¹, within the typical 2020 range of 2 to 3 µmol J⁻¹¹, with a double-ended HPS at 1.7 µmol J⁻¹^1,2. All numbers in the table are this article’s calculation.

The LED uses about a third less electricity for the same photons. Both figures are lower limits. If only 80% of the photons reach the shelf (an assumed figure), the flux rises to 225 µmol s⁻¹ and the LED to 90 W (our calculation). And no 106 W HPS reaches 1.7 µmol J⁻¹: that figure comes from 1,000 W double-ended lamps, while a 400 W magnetic-ballast HPS measured 0.94², which would need 191 W here (our calculation). All 72 W of the LED end up as heat in the room, like a 72 W heater running 16 hours a day.

Getting the photons onto the leaves

A small bare lamp far from the canopy behaves like a point source: its photons spread over an ever larger sphere, so the PPFD falls with the square of the distance. This is geometry, not plant science. A wide LED panel hung close to a shelf does not follow that rule closely, because each spot on the canopy receives light from many parts of the panel. With either, raising the fixture spreads the light wider and sends more of it past the edges.

Fixtures of equal efficacy spread their light very differently. In the 2014 tests one LED fixture concentrated most of its photons directly below it, while a ceramic metal halide fixture spread them evenly over a large area². High-wattage HPS fixtures must hang higher above the canopy than LEDs to light it evenly¹; Resh gives 61 to 91 cm for 100 to 250 W discharge lamps and 1.2 to 1.8 m for 1,000 W⁶. Edges are hardest: keeping a greenhouse perimeter as bright as its centre needs more fixtures in the outer rows².

The reliable check is to measure. Nelson and Bugbee mapped each fixture with 121 quantum-sensor readings on a flat plane². At home, a grid of readings at canopy height shows the average and the weak corners (our suggestion, after²); keep the sensor level and clean⁷. One reading under the centre overstates the average.

Spectrum, in one paragraph

Spectrum is covered in Light: PPFD, DLI, spectrum and photoperiod; here only its cost matters. Blue photons (400 to 500 nm) reduce stem length and leaf expansion in nearly all species, and fixtures typically contain 5% to 30% blue to keep plants compact¹. Blue costs efficacy: a blue LED makes about 3.5 µmol J⁻¹ against 4.5 for red, and blue photons are about 20% less efficient in photosynthesis than 660 nm red¹. Far-red photons (700 to 800 nm) increase leaf expansion in lettuce but stem elongation in many other species¹. White LEDs cost about a fifth as much as red ones and let people see pests and deficiency symptoms, which is why white plus red has become common¹.

The economics: cost per mole of photons

Nelson and Bugbee compared fixtures by the five-year cost of a mole of photons, fixture plus electricity; in 2014 LEDs cost five to ten times more per photon of capacity, so at equal efficacy their cost per mole was 2.3 times that of HPS². By 2020 LEDs led on efficacy, but HPS fixtures still sold for 5 to 9 µmol s⁻¹ of output per US dollar and LEDs for 0.5 to 1.5¹.

To repeat it, divide the fixture price per µmol s⁻¹ by the moles that capacity delivers in five years, and add the tariff ÷ (efficacy × 3.6). The table uses the 2020 prices of¹, efficacies of 2.5 (LED) and 1.7 (HPS), and two cases: a sole-source room at 16 hours a day with an assumed tariff of USD 0.25 per kWh, and the 3,000 hours a year and USD 0.11 per kWh that Nelson and Bugbee assumed for greenhouses². Results are our calculation, in US cents per mole, ignoring discounting, lamp replacement and cooling. The LED price per photon in¹ (USD 1,000 to 3,000 per kW, quoted as 0.5 to 1.5 µmol s⁻¹ per USD) corresponds to about 1.5 µmol J⁻¹; for a 2.5 µmol J⁻¹ fixture at the same price per kW, the LED fixture column would be about 40% lower and the cheapest LED would match HPS even at 3,000 hours (our calculation).

At long hours and a high tariff, electricity is most of the cost of LED light, and a cheaply bought LED beats HPS. At short hours and a low tariff, the fixture is half or more of the cost, and HPS can still be cheaper. Kusuma and co-workers put the LED payback over HPS at three to five years for sole-source and five to eight years for supplemental lighting¹.

For a rough cost per kilogram: a modelling study of a plant factory lit by LEDs at a PPFD of 500 µmol m⁻² s⁻¹ for 16 hours, twice the daily light of the shelf above, found that 1 kg of lettuce dry weight needed 247 kWh of purchased electricity, with lighting the largest single load¹¹. The study assumed 7% dry matter, which gives about 17 kWh, or EUR 4.3, per kilogram of fresh lettuce; at the 4% to 6% dry matter usually quoted for lettuce¹² it would be 10 to 15 kWh, or EUR 2.5 to 3.7 (our estimate, scaled from^11,12).

Reading a label and buying a light

In 2017 seven plant-lighting researchers proposed a standard lamp label showing PAR efficacy, photon flux in key wavebands and a map of horizontal light distribution (read here from the abstract only)⁸. The DesignLights Consortium, a North American industry body, set a 1.9 µmol J⁻¹ minimum for its qualified list in 2018, with a 5% tolerance that admits products down to 1.81^1,10; a 2022 draft proposed 2.30, measured over 400 to 700 nm with fans and controllers counted in the input power⁹. From the mechanisms above, ask a seller for:

• Efficacy in µmol J⁻¹ from a certified third-party test¹, for 400 to 700 nm⁹; a wider band inflates the number.
• A PPFD map at a stated hanging height, not one centre reading.
• Input power at the wall. “Equivalent wattage” names a lamp the product claims to replace, not the photons it emits.
• The basis of lifetime claims. A typical LED fixture keeps 70% of its output for 50,000 hours, but projections may not exceed six times the hours actually tested, and drivers often fail first¹.

Dimming helps twice: LEDs are more efficacious at lower drive current¹, and a dimmer or timer delivers exactly the daily light integral you planned.

Key numbers

Common mistakes

1. Buying by watts. Watts measure input, not light: a 2014 fluorescent fixture (0.84 µmol J⁻¹) and a 2020 LED fixture (2.5 to 3.0) differ about threefold in photons per watt, by our arithmetic from^1,2.
2. Comparing lamps by lumens or lux. Those units are weighted to human vision, and a cool white LED with more than twice the lumens per watt of a red LED delivered fewer photons per joule².
3. Sizing from one PPFD reading under the centre of the fixture. Fixtures spread their photons unevenly and the edges get least, so only a grid of readings gives the average².
4. Forgetting the heat. Unconverted electricity heats the fixture and absorbed light mostly heats the leaf, so a closed space must remove nearly all of a lamp’s power (our inference from^1,4,5).
5. Expecting a small HPS lamp to match the efficacy of a 1,000 W double-ended one. A 400 W magnetic-ballast HPS measured 0.94 µmol J⁻¹, against 1.70 for the best large lamps².

Summary

A grow light is a photon source, and its photon efficacy in µmol J⁻¹ compares it with any other lamp. Red LEDs are the most efficacious because red photons carry the least energy; white LEDs lose photons in the phosphor, and every fixture loses more in its driver, optics and heat. In 2014 the best LED and HPS fixtures were equal at about 1.7 µmol J⁻¹; by 2020 the best LED fixtures reached 2.5 to 3.0^1,2. To size a fixture, multiply the target PPFD by the area, divide by the efficacy, and multiply by the hours. Electricity dominates the cost of light at long hours and high tariffs; the fixture price matters most at short hours. Buy on a tested efficacy and a PPFD map, not on watts, lumens or equivalent wattage.

Next in the series: Temperature, humidity and VPD.

References

1. Kusuma, P., Pattison, P. M. and Bugbee, B. (2020). From physics to fixtures to food: current and potential LED efficacy. Horticulture Research, 7, 56. https://doi.org/10.1038/s41438-020-0283-7
2. Nelson, J. A. and Bugbee, B. (2014). Economic analysis of greenhouse lighting: light emitting diodes vs. high intensity discharge fixtures. PLoS ONE, 9(6), e99010. https://doi.org/10.1371/journal.pone.0099010
3. Nelson, J. A. and Bugbee, B. (2015). Analysis of environmental effects on leaf temperature under sunlight, high pressure sodium and light emitting diodes. PLoS ONE, 10(10), e0138930. https://doi.org/10.1371/journal.pone.0138930
4. Taiz, L. and Zeiger, E. (2002). Plant Physiology, 3rd ed. Sinauer Associates, Sunderland, MA. Chapter 7, Photosynthesis: The Light Reactions.
5. Castilla, N. (2013). Greenhouse Technology and Management, 2nd ed., translated by E. J. Baeza. CABI, Wallingford. ISBN 978-1-78064-103-4.
6. Resh, H. M. (2013). Hydroponic Food Production: A Definitive Guidebook for the Advanced Home Gardener and the Commercial Hydroponic Grower, 7th ed. CRC Press, Boca Raton. ISBN 978-1-4398-7869-9.
7. Torres, A. P. and Lopez, R. G. (2010). Measuring Daily Light Integral in a Greenhouse. Purdue Extension publication HO-238-W, Purdue University, West Lafayette, IN. https://www.extension.purdue.edu/extmedia/ho/ho-238-w.pdf
8. Both, A. J., Bugbee, B., Kubota, C., Lopez, R. G., Mitchell, C., Runkle, E. S. and Wallace, C. (2017). Proposed product label for electric lamps used in the plant sciences. HortTechnology, 27(4), 544–549. https://doi.org/10.21273/HORTTECH03648-16 (abstract only)
9. DesignLights Consortium (2022). Technical Requirements for LED-Based Horticultural Lighting, Version 3.0, Draft 1 (31 March 2022). Industry document. https://lightingdesignandspecification.ca/wp-content/uploads/2022/04/DLC_HORT-Technical-Requirements-V3-0-DRAFT1_03312022.pdf
10. Runkle, E. (2022). A closer look at LED efficacy. GPN (Greenhouse Product News), January 2022, p.42. Trade-magazine column. https://greenhousehort.ca.uky.edu/sites/greenhousehort.ca.uky.edu/files/2022-03/LED%20Efficacy.pdf
11. Graamans, L., Baeza, E., van den Dobbelsteen, A., Tsafaras, I. and Stanghellini, C. (2018). Plant factories versus greenhouses: comparison of resource use efficiency. Agricultural Systems, 160, 31–43. https://doi.org/10.1016/j.agsy.2017.11.003 (full text read in the open-access reprint: Graamans, L. (2021). STACKED. PhD thesis, TU Delft, chapter 3. https://doi.org/10.7480/abe.2021.05
12. Dediu, L., Cristea, V. and Zhang, X. (2012). Waste production and valorization in an integrated aquaponic system with bester and lettuce. African Journal of Biotechnology, 11(9), 2349–2358. https://doi.org/10.5897/AJB11.2829
13. Zhou, J., Wang, J. Z., Hang, T. and Li, P. P. (2020). Photosynthetic characteristics and growth performance of lettuce (Lactuca sativa L.) under different light/dark cycles in mini plant factories. Photosynthetica, 58(3), 740–747. https://doi.org/10.32615/ps.2020.013

This article was drafted with the help of AI tools and reviewed and edited by PonicLabs. Sources are listed above; tell us if you find an error.