Greenhouse lighting is not indoor lighting with a glass roof over it. A greenhouse already receives light, often a great deal of it, so the grow light has a different job: it fills the gap between what the sun delivers and what the crop needs in that week. That gap changes with the season, with the covering material, with the truss height and with the crop, and a lighting plan that ignores any of the four will either overlight in July or run short in December.
This guide answers the question for a small to medium cannabis greenhouse, from roughly 50 to 1,000 square metres. It starts with the daylight the structure already delivers, because that number decides everything downstream. It then covers fixture shape at truss height, the IP rating that condensation actually demands, photoperiod control with a light deprivation curtain, heat load, and the configuration that fits each size band. Where a number is a planning value rather than a rule, the range is stated.
The short version: choose LED greenhouse lighting by the DLI deficit it has to close in your worst month, not by watts per square metre. Then choose the fixture shape by your truss height, and the IP rating by your night-time humidity. A plan built in that order usually costs less than one built from a catalogue page, and it is far more likely to work in February.
A greenhouse grower does not need a fixture that produces a target PPFD. They need a fixture that produces the missing part of a target DLI. Those are different specifications, and confusing them is the most common and most expensive mistake in greenhouse lighting.
The arithmetic is simple enough to do before talking to any supplier. Measure or estimate the daily light integral that reaches the canopy on a clear day in your weakest month. Subtract it from the DLI your crop needs at that stage. The remainder is the daily dose your fixtures have to supply, and it is usually between 8 and 18 mol per square metre per day for a cannabis greenhouse in winter at mid latitude. Everything else, fixture count, wattage, running hours, follows from that number.
Two consequences follow immediately. First, a greenhouse in a high light region may need no supplemental lighting at all for four months of the year, which changes the payback calculation on a large installation. Second, the same room needs a dimmer and a seasonal schedule far more than it needs a high peak output, because the deficit in May is a fraction of the deficit in December. A lighting plan that runs a fixed 12 hours at full output all year is not a plan, it is a switch.
The rest of this guide works through the four variables that set the deficit: covering transmission, month, crop target and fixture placement. Get those four right and the equipment list writes itself.
Greenhouse covering is not transparent in the way a window is. Every layer, every sheet, every year of dust and every condensation film takes a share of the light before it reaches a leaf, and the shares differ enough to change a fixture count by a third.
| Covering | Typical transmission to canopy | What moves the number |
|---|---|---|
| Single layer glass, new | 85 to 90 percent | Cleanliness, frame shading, low iron glass |
| Double glazed glass | 72 to 80 percent | Number of panes, cavity, seal ageing |
| Double layer polyethylene film | 70 to 78 percent | Film age, dust, condensation between layers |
| Twin wall polycarbonate | 70 to 80 percent | Wall thickness, UV layer yellowing after 8 to 10 years |
| Glass with shade or energy screen closed | 35 to 60 percent | Screen type and how often it is deployed |
Two numbers in that table matter more than the rest. The first is age. A ten year old twin wall panel can transmit 10 to 15 percentage points less than its datasheet value, and the loss is uneven across a roof, so the weakest bay sets the plan. The second is the screen. A greenhouse that closes an energy screen at dusk to hold heat has also closed the light path, and a lighting schedule that assumes the screen is open will underdeliver for the hours that matter most in winter.
Measure rather than assume where possible. A quantum sensor held at canopy height under the roof at midday gives a working transmission figure in ten minutes, and repeating it in three or four places shows how much the roof varies. Growers who do this usually find one bay 10 to 20 percent darker than the rest, which is a placement problem before it is an equipment problem.
Seasonality is the part of greenhouse lighting that indoor guides skip, and it is the part that decides how much dimming range and how many control channels a project needs. At 40 degrees north, outdoor DLI runs from roughly 8 mol in January to 50 mol in July. Multiply by transmission and the canopy sees a range of about six to one before any lighting is added.
| Month | Outdoor DLI | At canopy, 78 percent transmission | Target DLI | Deficit | Hours at 600 PPFD |
|---|---|---|---|---|---|
| January | 8 mol | 6.2 mol | 20 mol | 13.8 mol | 6.4 h |
| March | 22 mol | 17.2 mol | 30 mol | 12.8 mol | 5.9 h |
| May | 40 mol | 31.2 mol | 35 mol | 3.8 mol | 1.8 h |
| July | 50 mol | 39 mol | 38 mol | 0 mol | 0 h |
| September | 32 mol | 25 mol | 35 mol | 10 mol | 4.6 h |
| November | 12 mol | 9.4 mol | 25 mol | 15.6 mol | 7.2 h |
The conversion behind the last column is worth keeping: 600 PPFD for one hour delivers 2.16 mol per square metre, because 600 micromoles per second times 3,600 seconds divided by one million is 2.16. Run the same figure for any target and any fixture output before committing to a count.
What this table changes in practice is the control specification. A January deficit of 13.8 mol and a May deficit of 3.8 mol cannot be served by one fixed timer setting. The room needs either a seasonal schedule that an operator updates monthly, or a controller with a daylight sensor that tracks the deficit and dims accordingly. The second option costs more at installation and less over five years, and it is the difference between a lighting system that is used and one that is switched on in December and left alone.
It also changes the honest answer about summer. In July this greenhouse has no deficit at a 38 mol target, and running lights would add heat, add cost and add nothing the plant can use. Any greenhouse lighting proposal that does not include a summer off period is a proposal written for a catalogue rather than for a crop.
A greenhouse is tall in a way an indoor room is not. A gutter connected house commonly runs 3 to 5 m to the truss, and a small tunnel or polytunnel can be 2.5 to 4 m at the ridge. That height is useful for air movement and for a person on a ladder, and it is hostile to light, because intensity falls with distance and the fall is not gentle.
For a compact source, doubling the distance to the canopy divides the PPFD by about four. A bar array behaves better over the first metre because it approximates a line source rather than a point, but past roughly 2 m the decay approaches the same curve. The practical consequence is that mounting a fixture at the truss and hoping it reaches the bed is a plan that fails quietly: the numbers on the datasheet are measured at 0.5 or 0.6 m, and the greenhouse has just moved the fixture three times further away.
There are two workable answers. The first is a low hung bar array: many low to mid power bars, each 300 to 800 W, dropped to 0.6 to 1.2 m above the canopy on hangers or a wire trellis, with the truss used only as a support. The second is a retractable system where the whole bar plane is winched up during work and harvest and lowered to the working height for the photoperiod. Both keep the light close to the leaves, which is where the efficiency is.
The number to design against is the working height, not the truss height. Ask what PPFD the fixture delivers at the height it will actually hang at, and ask for the measurement grid. A fixture rated 2.8 micromoles per joule at 0.3 m can be a poor choice at 1.5 m if its beam is narrow, and a bar with a wide distribution can beat it at the same wattage. This is why greenhouse projects should be specified with a layout drawing rather than a fixture list.
The default industrial answer to a tall space is a high bay fixture: a compact, high output lamp hung from the ridge. It is the wrong default for a greenhouse, and the reasons are geometric rather than electrical.
| Property | Bar array at 0.6 to 1.2 m | High bay fixture at 3 to 4 m |
|---|---|---|
| Uniformity across a bed | Even when bars are spaced to overlap, typically within 10 to 15 percent | Hot centre under each fixture, 40 percent or more falloff at the edges |
| Height following the crop | Hangers move weekly, bars stay close to the tops | Fixed at the truss, canopy grows toward it |
| Shadow from structure and gutters | Low, bars sit below the frame | High, the frame crosses the beam |
| Heat at the canopy | Spread over a wide area, easier to ventilate | Concentrated under each lamp |
| Service and cleaning | Reachable from a low platform | Ladder or lift work at height |
| Screen and curtain conflict | Bars mount below the screen line | Often above it, so closing the screen cuts the light |
The screen row deserves emphasis. Many greenhouses close an energy or shade screen at dusk, and a fixture mounted above that screen delivers almost nothing once it closes. Bars hung below the screen line keep working through the night, which is exactly when the winter deficit is being closed. If the structure has a screen, the lighting layout has to be drawn around it, and that usually means bars.
Bars also handle the crop height problem that a greenhouse has and an indoor room does not. A cannabis canopy rises 1 m or more between the flip and harvest, and a tomato crop can run 3 m up a wire. A bar plane that can be raised in steps keeps the distance roughly constant. A fixed high bay does not, so the top of the crop gets more light than the bottom and the lower third of the canopy thins out.
Greenhouses are wet in a way that is easy to underestimate. Night-time relative humidity commonly runs 85 to 95 percent, and when the roof cools below the dew point, water condenses on every cold surface in the house, including the lens and the driver housing of a light that was switched off an hour earlier. A fixture designed for a dry indoor room will fail in that environment, and it usually fails at the driver rather than at the LEDs.
| Rating | What it actually resists | Fit for a greenhouse |
|---|---|---|
| IP20 to IP44 | Dust limited, splashes from one direction | No. Indoor rooms only |
| IP65 | Dust tight, water jets from a nozzle | Workable in a well ventilated house with no washing |
| IP66 | Dust tight, powerful water jets | The sensible minimum where roofs are washed or sprayed |
| IP67 | Temporary immersion | Over-specified for most houses, useful in flood-prone tunnels |
The rating that matters is on the lowest point of the assembly, not on the LED board. Water collects at the cable gland, the breather valve and the driver connector, so a fixture with an IP66 lens and an IP44 connector is an IP44 fixture in practice. Ask for the rating of the complete luminaire, and ask how the driver cavity is sealed and whether the breather is a membrane type that lets pressure equalise without letting water in.
Two more details belong in the same conversation. First, cleaning: a greenhouse roof and a fixture lens both get washed, often with a disinfectant or an acidic cleaner, and the seal and the lens material have to tolerate it. Second, sulphur: many greenhouse growers run sulphur vaporisers for powdery mildew, and sulphur vapour attacks seals, reflectors and some lens polymers. A fixture rated for a clean room is not automatically rated for a house that gets a sulphur burn.
A cannabis greenhouse cannot rely on the calendar the way an indoor room can. At 40 degrees north, natural day length runs from about 9 hours in December to 15.5 in June, and the plant needs 12 hours of uninterrupted darkness to flower. From April to September the sun is up longer than that, so the house has to be darkened, not just lit. Photoperiod control in a greenhouse is a curtain problem first and a lighting problem second.
The lighting plan has to fit inside the dark window. If the curtain closes at 19:00 and opens at 07:00, the supplemental lighting can only run in the hours between, and it has to be off before the curtain closes. A controller that runs the lights on a fixed schedule and the curtain on another is a recipe for a light leak at the seam, which shows up three weeks later as a crop that will not finish. Single controller, single schedule, with the curtain and the lights interlocked, is the specification to ask for.
Light pollution from inside the house matters as much as light getting in. A seam, a torn curtain, a door left open or a status LED on a fixture can deliver a few micromoles at canopy level, and cannabis is sensitive at low levels. Walk the house at night during the dark period with the lights off and look for glow at the canopy. The fix is usually a curtain repair and a piece of tape over an indicator, not a new fixture.
Outdoor lighting is the other direction of the same problem. A greenhouse lit from outside by a yard light or a security flood can miss its flip, and a neighbouring grow can be affected by a house that leaks. Shield exterior fixtures, aim them down, and check the dark period from outside the house as well as inside.
Every kilowatt of lighting becomes heat inside the envelope, at roughly 3,412 BTU per hour. A 20 kW installation adds about 68,000 BTU per hour to a house that is already trying to shed solar gain, and in a small greenhouse with limited vent area that load shows up as a temperature spike in the first hour after the lights come on.
LED helps here, but not as much as the marketing suggests. The gain over HPS is in the spectrum and in the fact that less energy leaves the fixture as radiant heat toward the canopy, but the electrical energy still ends up as heat in the house. The genuine advantages are that the heat is spread over a large bar area instead of concentrated under a lamp, and that the fixture can be dimmed, which is a heat control as much as a light control.
The scheduling answer is to light when the house is coolest. Pre-dawn and post-dusk hours need less ventilation than midday, and in winter the same kilowatt hours do double duty: they close the DLI gap and they hold the house temperature. In summer, the same schedule is wrong, which is another reason the summer off period matters. If a house must light in summer, run the photoperiod across the coolest six hours and keep the dimmer low.
Ventilation and dehumidification capacity should be sized for the lighting load, not for the empty house. Ask what the temperature rise will be with the lights at full output and the vents at their setpoint, and check that the answer leaves margin for a still, humid night. A house that cannot shed the lamp heat will run the lights at 60 percent for most of the year, which quietly turns a well specified installation into an expensive one.
With the deficit, the truss height, the IP rating and the heat load settled, the equipment list becomes a sizing exercise. The table below gives planning values for cannabis at a 600 PPFD supplemental target, which is a common choice for a house that is closing a winter deficit rather than running a sealed room.
| Canopy area | Fixture type | Unit power | Rough count | Control |
|---|---|---|---|---|
| 50 to 100 m² | Bars on hangers, movable | 300 to 600 W | 8 to 16 bars | One or two zones, timer plus dimmer |
| 100 to 300 m² | Bars on fixed wire or rail | 600 to 800 W | 20 to 40 bars | Two to four zones, 0 to 10 V dimming |
| 300 to 600 m² | Bars on winchable frames | 600 to 800 W | 40 to 80 bars | Four to eight zones, central controller |
| 600 to 1,000 m² | Bar arrays plus a few high output units over tall crops | 800 W and up | Calculated per bay | Zoned distribution, daylight sensor, DALI or 0 to 10 V |
The count column assumes a bar spacing that gives roughly 600 PPFD at the working height and a 10 to 15 percent overlap between adjacent bars. It is a starting point for a layout drawing, not a purchase order. Two houses of the same area can differ by 30 percent in fixture count because one has a clean glass roof and the other has a ten year old polycarbonate roof with a closed energy screen.
For a house under 100 m², the movable bar approach has a second advantage beyond cost: the bars can be taken to the crop rather than the crop grown under fixed bars, which suits a propagation bay that is used for eight weeks and then emptied. For a house over 300 m², the fixed frame with a winch pays for itself in labour alone during the first harvest, because raising the whole plane for work is a one person job.
A single zone forces the whole house to run one setting, and a greenhouse rarely has one condition. The propagation bench wants 150 PPFD for 18 hours, the flower bay wants 700 for 12, and the mother area wants something in between. If they share a circuit and a timer, two of the three are wrong most of the time.
Zoning costs very little at installation: a second contactor, a second dimming channel, a run of control cable. It costs a great deal to retrofit, because the cable has to be pulled through a house in production. Specify zones on the drawing, even if only one is wired in the first season. The usual split is propagation, vegetative, flower early and flower late, with the last two often sharing a schedule but separated on dimming.
The controller matters more than the zone count. A greenhouse needs a controller that can hold a seasonal schedule, accept a daylight sensor input, and interlock the lights with the photoperiod curtain. A basic astronomical timer does the first job badly and the third not at all. Ask for the schedule resolution, the sensor input type and the curtain interlock before comparing prices, because those three features separate a greenhouse controller from a room timer.
Most small greenhouses that add LED lighting are retrofits, and the surprise is rarely the fixtures. It is the electrical capacity, the structure and the cable route, in that order.
An older house often has a lighting circuit sized for a few incandescent lamps and a fan, perhaps 10 or 16 A at 230 V. A 20 kW LED installation needs about 87 A single phase or 29 A at 400 V three phase, plus diversity for the rest of the house. If the supply is single phase, the upgrade may be the largest line item in the project, and it is worth pricing before the fixture choice is made. A three phase supply also lets the load be balanced across phases, which matters when the house already runs pumps and fans.
Structure is the second item. A glass house truss is designed for glass, snow and wind, not for a bar plane plus cable loads, and the fixing method matters as much as the weight. Bars at 2 to 4 kg per metre are light, but 40 of them plus cable tray and a winch frame add up, and drilling a truss chord in the wrong place is a structural question rather than a lighting one. Ask the greenhouse manufacturer or an engineer before drilling, and prefer clamps or rail systems that spread the load.
Cable route and condensation are the third. Cables run through a greenhouse need UV stable jacket, a drip loop at every fixture, and a route that does not sit in the gutter. A run that pools water at a connector will fail in the first winter, and the failure will look like a driver fault. Specify the cable type and the drip detail on the drawing, and check that the installer follows it.
There are greenhouses where the honest answer is to install less lighting, or none. Saying so is more useful than a proposal that assumes every house needs a full bar plane.
The first case is a spring propagation house that is empty by June. If the natural light from March onward is already above the target for the crop, supplemental lighting only serves the first six weeks, and the payback on a fixed installation can run past the life of the bars. A few movable bars over the bench, used for six weeks and stored for the rest of the year, is the right specification.
The second case is a house with a very low electricity price and a winter heating requirement, where the waste heat from an existing HPS installation is doing real work. Replacing it with LED reduces the lighting bill and increases the heating bill, and the two can cancel. The comparison has to be run on the whole energy bill, not on lumens per watt. This is a narrower case than it was ten years ago, because LED efficacy has roughly doubled, but it still exists in cold climates with cheap power.
The third case is a house that cannot be darkened and is growing a short day crop in summer. Lighting cannot fix a photoperiod problem, and installing bars to compensate for a light leak is spending money on the wrong variable. Fix the curtain first. The lighting plan comes after the dark period is reliable.
| Mistake | What it looks like | What it costs |
|---|---|---|
| Sizing by watts per square metre | A supplier quotes 200 W per m² without asking about the roof or the month | Overlighting in summer, shortfall in December |
| Installing indoor fixtures | IP20 or IP44 bars bought on price | Driver failures in the first wet season |
| Hanging at the truss | Fixtures 3 to 4 m above the canopy because that is where the steel is | Half the rated PPFD, twice the fixtures needed |
| One zone for the whole house | Propagation and flower on the same timer | One stage always runs the wrong setting |
| Mounting above the screen | Bars fixed to the truss, screen closes at dusk | The lighting stops working at the hour it is needed |
| Ignoring the electrical supply | Fixtures ordered before the panel is checked | A supply upgrade that arrives after the lights |
Each of the six is cheap to avoid at design time and expensive to correct later. The pattern behind them is the same: a greenhouse lighting plan built from a fixture catalogue treats the house as a room with a glass roof. A plan built from the four variables, transmission, month, truss height and humidity, treats the house as a structure with a light budget, which is what it is.
A greenhouse quote is only as good as the brief behind it. The list below is the minimum for a layout drawing that can be checked rather than guessed at.
| Item | Why the supplier needs it |
|---|---|
| Canopy area and bay dimensions | Sets the bar count and the spacing pattern |
| Truss or ridge height | Sets the hanging system and the working height |
| Covering material and age | Sets the transmission and therefore the deficit |
| Latitude and weakest month | Sets the seasonal schedule and the peak output |
| Crop, stage and target DLI | Sets the PPFD target and the photoperiod |
| Screen or curtain type and travel | Sets the mounting line and the interlock |
| Supply voltage and available current | Sets the circuit layout and the dimming protocol |
| Night-time humidity and cleaning regime | Sets the IP rating and the seal specification |
With those eight items, a supplier can return a layout with a fixture count, a PPFD grid at the working height, a seasonal schedule and a kW figure. Without them, the same supplier can only return a price on a fixture, which is not a lighting plan and will not tell you whether the house will finish a crop in January.
It also helps to state the constraint you care about most. A house with a weak supply wants the lowest kW that closes the deficit, which may mean fewer bars run longer. A house with a tight dark period wants the highest PPFD per bar so the lighting fits inside the window. A house with a sulphur programme wants seals and lenses specified for it. The same crop and the same area can produce three different layouts, and the brief is what separates them.
Start from the DLI deficit, not the floor area. Estimate the daylight reaching the canopy in your weakest month, subtract it from the target DLI for the crop, and convert the remainder into hours at a chosen PPFD. A 100 m² house closing a 13 mol deficit at 600 PPFD needs about six hours of lighting, which is roughly 12 to 16 bars of 600 W depending on spacing. A supplier who quotes a count from area alone is guessing.
Only if they are rated for the humidity. A greenhouse at night commonly sits at 85 to 95 percent relative humidity and condenses water on cold surfaces, so a fixture rated IP20 or IP44 will take water at the driver connector. Use a complete luminaire rated IP65 at minimum, and IP66 where the roof or the fixtures are washed. Check the rating of the gland and the breather, not just the lens.
Enough hours to close that month's deficit, and no more than the photoperiod allows. In January at mid latitude, six to seven hours at 600 PPFD is typical for a cannabis crop; in May it may be under two. The lights must also sit inside the dark window when a light deprivation curtain is used, so the schedule is bounded by the curtain on both ends.
Usually not for a cannabis crop at mid latitude. Outdoor DLI in July is around 50 mol, and after transmission the canopy sees close to 40, which is above a typical flower target. Running lights then adds heat and cost with no yield response, so a summer off period belongs in the plan. Higher latitudes and heavily shaded houses are the exceptions.
Close enough to the canopy to keep the rated PPFD, which in practice means 0.6 to 1.2 m for bars and no more than 1.5 m for a wide distribution fixture. Truss height is a structural fact, not a mounting instruction. Design against the working height, and ask the supplier for a PPFD measurement at that height rather than at the 0.3 m used on many datasheets.
They are worth it where the house has a real winter deficit, a photoperiod requirement, or a propagation programme that starts before natural light is adequate. They are harder to justify in a spring-only house that is empty by June, or where cheap electricity and existing HPS waste heat already cover winter heating. Run the comparison on the whole energy bill, including heating and ventilation.
Lighting alone cannot, because flowering is triggered by the length of the dark period. In a greenhouse that means a light deprivation curtain that holds 12 hours of uninterrupted darkness, with the lights scheduled inside that window. Supplemental lighting supports yield and quality once the photoperiod is correct; it cannot replace the curtain.
IP65 is the working minimum for a ventilated house, and IP66 is the sensible choice where roofs or fixtures are washed or sprayed. IP67 is rarely necessary except in tunnels that flood. The rating has to apply to the complete luminaire, including the cable gland and the pressure equalisation breather, because that is where water enters.
Measure PPFD at canopy height on a clear day, convert to DLI by multiplying by the seconds of daylight and dividing by one million, then subtract from the target. As a shortcut, 600 PPFD for one hour adds 2.16 mol per square metre. Do the calculation for the weakest month, because that is the month that sets the fixture count.
They add heat at about 3,412 BTU per kW per hour, so a 20 kW installation adds roughly 68,000 BTU per hour. Whether that overheats the house depends on vent area, roof height and the hour the lights run. Lighting pre-dawn and after dusk reduces the ventilation load, and in winter the same heat offsets heating. Size the vents for the lighting load, not for an empty house.
Below. A fixture mounted above a closed energy or shade screen delivers very little light to the crop, and screens are typically closed at dusk, which is when winter supplemental lighting is most useful. Hang bars below the screen line and check that the screen travel does not strike them.
Send the canopy area, bay dimensions, truss height, covering material and age, latitude, crop and target DLI, plus your supply voltage and available current. We answer with a bar layout at the working height, a monthly schedule, the kW figure and the IP specification for your night-time humidity.
