LED vs HPS vs CMH: choosing a grow light technology

Two fixtures can both be sold as "600 W" and put very different amounts of usable light on your canopy. The number that explains the gap is photon efficacy, not wattage: current top-bin LED fixtures convert 2.5–3.0 µmol of light for every joule of electricity they draw, against 1.7–2.1 for double-ended (DE) HPS, 1.7–2.0 for ceramic metal halide (CMH, also sold as LEC), and 1.5–1.7 for single-ended HPS, the cheapest and still the most common mogul-base lamp. These are current manufacturer datasheet ranges, aggregated across listed fixtures and not independently verified; the next section ties the LED and DE HPS numbers to a peer-reviewed test ^[1].

That gap drives almost everything else in this comparison: the electricity bill, the heat you have to remove, how close you can hang the fixture, and how often you replace it. None of the three technologies is simply "better". They trade differently against fixture cost, running cost, spectrum and control, and the right choice depends on which of those your situation actually constrains.

Top-bin LED photon efficacy

2.5–3.0µmol/J

current manufacturer figures [1]

Double-ended HPS photon efficacy

1.7–2.1µmol/J

manufacturer datasheets, aggregated

Electricity for matching light, one 63-day flower cycle

≈185kWh less with LED

worked example below

LED diode array life to L90

40,000–50,000+hours

manufacturer LM-80/TM-21 projections; drivers rated separately. HPS/CMH bulbs: 10,000–12,000 h

The six things that actually separate these three technologies

Read this table before the numbers below it: everything in the rest of the article explains one row.

Photon efficacy: the number that drives your electricity bill

Photon efficacy (PPE, µmol/J) is the fixture equivalent of miles per gallon: it says how many micromoles of photosynthetically useful light come out for every joule of electricity that goes in. Watts alone tell you nothing about light output, because a large share of the electricity feeding an HPS or CMH lamp leaves as heat rather than as usable photons. Reviewing published LED and HID performance, Kusuma, Pattison & Bugbee (2020) reported current blue-plus-red LED fixtures reaching close to 3 µmol/J and white-plus-red fixtures 2.5–2.8 µmol/J, against 1.72 µmol/J for a 1,000 W double-ended HPS fixture reported in the same review ^[1].

Photon efficacy (µmol/J) · higher is better

• Single-ended HPSmanufacturer datasheets, aggregated1.60 µmol/J
• Double-ended HPSKusuma, Pattison & Bugbee (2020) [1]1.72 µmol/J
• CMH / LECmanufacturer datasheets, aggregated1.85 µmol/J
• Top-bin LEDKusuma, Pattison & Bugbee (2020) [1]2.80 µmol/J

That roughly 1.5- to 1.7-fold gap between a white-plus-red LED fixture and a double-ended HPS is a physical one, not a marketing claim: LED chips convert electricity to light directly across a semiconductor junction, while HID lamps first strike an electric arc and then rely on that arc heating a gas fill to produce light, an inherently lossier route. For the physics behind why PPE, not wattage, is the number to compare, see the photon efficacy explainer; the DLI calculator turns a fixture's PPF output into the daily light total your canopy actually receives.

Spectrum is a separate decision from efficacy

Efficacy tells you how much light you get; spectrum tells you what colour it is, and the three technologies differ sharply here too. HPS runs a sodium arc that puts out a narrow, yellow-orange-heavy spectrum with very little blue, a quirk of sodium's emission lines rather than a design choice for plants. CMH's ceramic arc tube tolerates higher operating temperatures and a wider mix of metal-halide salts than an older quartz tube, and that mix of emitting metals gives a broader spectrum closer to daylight, including some UV-A. LED spectrum is the one genuinely flexible variable: two fixtures sold as "LED" can differ enormously, from white-dominant to heavily blue-and-red, and some add dedicated far-red or UV channels. Spectrum is a fixture-by-fixture decision, not a property of the technology.

Heat at the canopy decides how close you can hang the light

Watt for watt, HPS and CMH put far more of their output onto the plant as long-wave infrared. Their arc tubes run hot enough to radiate heat directly onto the leaves below; an LED's heat is mostly carried away by a heatsink and moved by convection into the room air, so much less of it lands on the plant itself. Under typical indoor conditions this puts leaves about 1.3 °C warmer under HPS than under LED at a matched PPFD ^[3]; push it towards a hot, still, water-stressed corner of the room and the gap widens: modelled worst-case leaf temperatures reach about 12 °C above air temperature under HPS against about 8 °C under LED ^[3].

That difference sets how close you can run the fixture. A hotter light has to hang further from the canopy to keep leaf temperature and VPD in range, which spreads its output over a wider area and caps the peak PPFD you can put on the plants in a short tent. A cooler-running LED can typically hang closer, so more of its higher photon output actually lands on the canopy rather than being traded away for headroom.

Fixture lifetime and what replacement really costs

HPS and CMH bulbs lose output and drift in spectrum well before they fail outright, which is why they are treated as consumables rather than fixtures. Nelson & Bugbee's 2014 economic review found double-ended HPS lamps reaching 90% survival at around 10,000 hours, with single-ended (mogul-base) lamps surviving somewhat longer, 10,000–17,000 hours, in the same analysis ^[4]. Growers rarely run a lamp to failure, though: replacing HPS and CMH bulbs every 10,000–12,000 hours, before output and colour drift too far to trust, is grower and industry practice rather than a tested threshold. At 12 h a day that is 4,380 h a year, so a single 12-hour flowering room sees a lamp change roughly every 2.3–2.7 years on that practice; a facility running several rooms on staggered schedules, or one that simply replaces on a fixed annual date to avoid tracking individual bulbs, effectively changes lamps close to once a year.

LED diode arrays are commonly rated L90 (90% of initial output retained) at 40,000–50,000 hours or more, based on LM-80/TM-21 test-based projections, but that number covers the diodes only. The driver, the electronics that convert mains AC to the diodes' DC current, is a separate component with its own, usually shorter, temperature-dependent life. A spec sheet quoting one lifetime figure for "the fixture" is telling you about the diodes, not the driver; the DesignLights Consortium's Horticultural QPL lists L90 ratings by qualified fixture, and is a reasonable place to check a specific model rather than take a headline number on faith ^[5].

The running-cost worked example: a 600 W HPS against a load-matched LED

A 600 W double-ended HPS lamp on its ballast draws roughly 645 W at the wall once ballast losses are counted (a typical figure; check your ballast's datasheet). At 1.72 µmol/J ^[1], that puts out about 1,110 µmol/s of PPF (645 × 1.72). Matching that output with a top-bin LED fixture at 2.8 µmol/J needs only about 400 W at the wall (1,110 ÷ 2.8 ≈ 396, rounded up): roughly 245 W, or about 38%, less electricity for the same light.

Over a 63-day flowering cycle at 12 h a day (756 hours), the HPS fixture draws 645 W × 756 h ≈ 487.6 kWh; the load-matched LED draws 400 W × 756 h ≈ 302.4 kWh, a difference of about 185 kWh. At €0.15/kWh that is roughly €28 saved per cycle; at €0.25/kWh, about €46. Multiply by however many cycles a year your room runs, and by the fixture count, to get your own annual figure.

That 245 W gap is also 245 W less heat the HVAC has to remove, on top of HPS's higher radiant fraction (previous section). In a single tent that is a modest saving; in a 24-light commercial room it is close to 5.9 kW of avoided electrical heat before you even count the radiant-fraction difference, worth a proper cooling-load recalculation when you retrofit a room, not a rule of thumb. Set that running-cost saving against the capital gap: LED fixtures commonly cost several hundred euros more than an equivalent HPS fixture at this wattage today (the multiple was 5–10× per photon delivered in Nelson & Bugbee's 2014 analysis and has narrowed considerably since as LED prices have fallen ^[4]). Get current quotes for your own wattage class, divide the price difference by your saving per cycle, and you have your payback in cycles.

Dimming, spectrum control and where the driver lives

LED fixtures typically dim smoothly from about 10% to 100%, and many commercial models allow independent dimming of separate diode channels, so the spectrum can shift as intensity changes through a grow cycle. HPS and CMH ballasts step-dim in a handful of fixed increments, commonly 50%, 75% and 100%, and cannot change spectrum at all, because the spectrum comes from the fixed chemistry of the arc, not from the ballast.

The other practical difference is where the electronics sit. HPS and CMH almost always use a remote ballast connected to the reflector by a heavy cable, which keeps that component's own heat out of the tent but adds a cable run to manage. LED drivers are usually mounted on or built into the fixture itself, which simplifies wiring but puts the driver's heat inside the grow space: a small but real addition to the heat load discussed above, and a reason some commercial LED installations now specify remote drivers too.

Who should choose what

A budget-constrained first tent. If fixture cost is what is stopping you starting, an HPS or CMH fixture usually costs less to buy for the same light output, even though it costs more to run; check current quotes rather than older comparisons, because LED fixture prices have fallen a long way since the 2014 cost analysis referenced above ^[4]. A single-ended HPS or a 315 W CMH is often the cheapest way to buy useful PPFD for a first 1.2 × 1.2 m tent (a purchase-price observation, not a tested figure); see the LED fixtures for a 1.2 × 1.2 m tent comparison before you decide the gap isn't worth closing at your scale.

A heat-limited small space. Where headroom, not budget, is the constraint (a short cabinet, a wardrobe conversion, a tent you can't hang a light far from), LED's lower radiant load is usually the deciding factor over its running-cost saving, because it is what lets you reach useful PPFD without cooking the top of the canopy.

A commercial room already running HPS. A lamp-for-lamp LED retrofit is not a plug-in swap: lower total wattage per fixture changes the room's electrical service sizing, and the reduced (and differently distributed) heat load changes the cooling and dehumidification design the room was built around — see sizing extraction before committing capital. Fixture spectrum, dimming behaviour and canopy uniformity, not just efficacy, also change when you retrofit; that is worth a room-specific comparison, not a single-fixture swap decision.

Before you compare specific fixtures on a spec sheet, check how the headline PPE number was measured: at what junction or ambient temperature, whether it covers the whole fixture or just the diodes, and whether "efficacy" and "PPF" both come from the same test as the price you're paying. The DLC Horticultural QPL lists tested figures for qualified fixtures rather than manufacturer claims alone ^[5], and our guide to reading LED spec sheets walks through the rest. Whichever technology you land on, spectrum is still a fixture-by-fixture choice, not a technology-wide one; the bar vs board LED fixtures comparison is the next step once you've settled on LED.