Most grow light purchases that go wrong do not go wrong on price. They go wrong in one of three places: the fixture does not fit the room, the paperwork does not survive a utility or inspection review, or the HVAC cannot remove what the fixture adds. None of those failures appears in a price per fixture, which is why the cheapest quote is so often the most expensive installation.
This guide is the checklist we run with licensed cannabis cultivators before a quotation is written. It covers ten decisions in the order they should be made, from room geometry to commercial terms, with the arithmetic behind each one. It is written for a buyer comparing three quotes for grow lights in a flower room, a veg room or a full facility, not for a hobbyist choosing a first lamp.
The last part of this guide covers the decision the spec sheet cannot answer: which supplier. Batch consistency, lead time and after-sales decide whether the fixtures you approved are the fixtures that arrive, and whether a failure in year two is a twenty-minute repair or a lost cycle.
The through line is simple: a grow light is a photon delivery system with an electrical and thermal footprint. Every number below either describes photons, describes what it costs to deliver them, or describes what that delivery does to the rest of the room.
Buyers usually start at the last line, the price, and work backwards. That is why three quotes for the same room arrive in three formats and none of them can be compared line by line. The table below is the sequence that makes quotes comparable, because each line narrows the next one.
| Decision | Question it answers | What you should hold at the end |
|---|---|---|
| 1. Room and canopy geometry | What footprint must be lit, at what height, with which aisles? | A scaled plan with the canopy outline, mounting height and row layout |
| 2. Photon target by stage | What PPFD and DLI does each stage need? | A stage table with PPFD and DLI setpoints for clone, veg and flower |
| 3. Photon output per bed | How many µmol/s must land on each bed? | µmol/s per bed, plus a 10 to 15 percent edge margin |
| 4. Fixture count and layout | How many fixtures, at what spacing, over which beds? | A fixture map with overlap and wall clearance marked |
| 5. Efficacy at the wall | How many watts per µmol/s at full output and at part load? | PPF, input watts, a part-load curve and a DLC listing |
| 6. Spectrum and channels | Which SPD, which channels, which schedule? | An SPD file and a channel plan for veg, flower and finish |
| 7. Certification scope | Which listings cover which models at which voltage? | Listing and certificate numbers that match the quoted model |
| 8. Heat and HVAC | How many BTU/h does the layout add, and where does it go? | A heat load by room, with the radiant share separated |
| 9. Form factor and mounting | Bars, modules or fold-out, and at what height? | An outline drawing, weight and a mounting kit list |
| 10. Commercial terms | Warranty, spares, lead time and price per µmol/s | A warranty document, a spare parts list and delivered cost |
Two lines do most of the work. A fixture that is wrong on line three cannot be rescued by a discount, and a fixture that is right on line three but vague on line five will cost more every month it runs. The remaining eight lines decide whether the installation happens on schedule and whether it can still be serviced in year six.
Those ten lines describe the product. The two sections before the line-up describe the company that ships it, because a fixture that measures correctly on the test bench is still a schedule risk if next month's batch does not match the sample.
The rest of this guide takes each line in turn. Where a number depends on your room, the section says which input to change rather than averaging it away.
The room sets the design before any fixture does. Four inputs pin it: the canopy footprint at the top of the trellis, the finished ceiling height, the clear height you can keep above the canopy, and the aisle width you need for inspection and defoliation. A room with 9 ft ceilings and 8 ft of plants is not the same lighting problem as a room with 14 ft ceilings and the same floor area, even though both might be quoted at 40 W per square foot.
Mounting height is the input buyers leave out most often. A bar fixture hung 8 in above the canopy behaves differently from the same fixture at 24 in: the map gets flatter and dimmer as it rises, wall losses change, and the height you can actually maintain during a grow is set by how far the plants stretch in week three. Ask for the PPFD map at the height you will use, not at the height in the brochure.
The stage plan decides zoning. A perpetual harvest with one veg room and two flower rooms needs independent zones and separate schedules. A single flip room can run on one schedule, one controller and one set of setpoints. Buying zoned fixtures for a single-zone room wastes money; buying single-zone fixtures for a perpetual harvest means the dimmer becomes the schedule, which is a job it does badly.
Existing PPFD readings are worth more than any model. Nine readings per bed at canopy height, taken with a calibrated quantum sensor, show what the plants actually receive and where the map falls off. That baseline is also the only way to prove afterwards that a retrofit delivered what the quote promised.
Two numbers get confused in every lighting conversation. PPF, in µmol/s, is what a fixture emits. PPFD, in µmol/m²/s, is what lands on a square metre of canopy. The conversion is one line of arithmetic: µmol/s required equals the PPFD setpoint multiplied by the canopy area in square metres. A 4 × 6 ft bed is 24 sq ft, or 2.23 m². At a 1,000 µmol/m²/s setpoint that bed needs 2,230 µmol/s, and with a 12 percent edge margin it needs about 2,500 µmol/s.
Now the fixture. A 1,300W wide-bar fixture at 3.0 to 3.5 µmol/J emits 3,900 to 4,550 µmol/s, which is more than the bed needs at full output, so it runs at roughly 55 to 65 percent for most of flower. That headroom is deliberate: the same fixture covers the peak weeks at 1,100 to 1,200 µmol/m²/s without adding hardware, and a fixture run below full output usually lasts longer than one run flat out. A 1000W bar at 2.8 µmol/J emits about 2,800 µmol/s, which covers the same bed at 80 to 90 percent.
Rules of thumb such as 30 to 50 W per square foot survive because they are easy to repeat. They also hide the only number that matters. Fifty watts per square foot is about 538 W/m², which at 2.8 µmol/J delivers 1,507 µmol/m²/s and at 2.3 µmol/J delivers 1,237. Two fixtures sold at the same watts per square foot can differ by 40 percent in photons, and one of them will cook the canopy while the other underfills it.
| Canopy bed | Area | Setpoint at canopy | µmol/s needed with 12 percent edge margin | Fixture class that covers it |
|---|---|---|---|---|
| 4 × 4 ft | 1.49 m² | 1,000 µmol/m²/s | 1,670 µmol/s | 600W to 700W bar at high output |
| 4 × 6 ft | 2.23 m² | 1,000 µmol/m²/s | 2,500 µmol/s | 1000W bar at 90 percent, or 1,300W at 60 percent |
| 4 × 8 ft | 2.97 m² | 1,000 µmol/m²/s | 3,330 µmol/s | 1200W to 1400W bar |
| 5 × 10 ft | 4.65 m² | 1,000 µmol/m²/s | 5,210 µmol/s | Two 1000W bars, or one 1,600W fixture plus a bar |
| 4 × 6 ft, veg stage | 2.23 m² | 500 µmol/m²/s | 1,250 µmol/s | One 1000W bar dimmed to 45 percent |
The margin is not marketing. It covers the difference between a centre reading and the corner of the bed, the falloff at the wall, and the week when the canopy grows 6 in closer to the fixture. A layout sized exactly to the setpoint at the centre will be short at the edges from the first day.
Efficacy is photon output divided by electrical input at the wall, written µmol/J. It is the only fair way to compare two fixtures, because it already includes driver losses. Ask for input watts at a stated voltage, not an equivalence claim. The DLC horticultural specification V4.0 sets its threshold at 2.5 µmol/J, up from 2.3 in V3.0 and 1.9 in the first horticultural specification in 2018. A fixture below 2.5 is not automatically unsuitable, but it is no longer competitive for a commercial room built this year.
Efficacy is not constant across the dimming range, and this matters more for LED than for the lamps it replaces. Many fixtures lose two to four points at the bottom of their range, and some shift spectrum as current drops. A fixture rated at 2.9 µmol/J at full output may deliver 2.6 at 30 percent. If the room will run below half output for most of the year, which is what a 1,300W fixture over a 4 × 6 bed usually does, ask for the part-load curve in writing and price the electricity at the point where the fixture actually operates.
The arithmetic is simple enough to check against any quote. To replace the 1,850 µmol/s of a 1,000W double-ended HPS, a fixture at 2.8 µmol/J needs 661 W at the wall. That is 409 W less per fixture, or 38 percent less electricity for the same photon output. Over 4,380 flowering hours a year it is about 1,790 kWh per fixture, near $215 at $0.12 per kWh, before any HVAC saving that follows the same watt reduction.
| Spec line | What a complete answer looks like | Where buyers get fooled |
|---|---|---|
| Power | Input watts at 100V to 277V, measured at the plug | "1,000W HPS equivalent" |
| Output | PPF in µmol/s, with the measurement tier named | Lumens, or "PAR watts" |
| Efficacy | µmol/J at the wall, plus a part-load curve | A diode-only number |
| Footprint | A PPFD map at a stated height, with corner readings | A coverage claim measured at 12 in |
| Service life | L90 hours and driver life at rated case temperature | "50,000 hours" with no L-level |
| Listing | A DLC model listing and an ETL file number | A logo image on a brochure |
The guide to µmol/J and how it is measured walks through the measurement tiers and the questions that separate a laboratory number from a wall number.
"Full spectrum" has no agreed definition, so it cannot be a purchase criterion on its own. What can be checked is the spectral power distribution file, the red to far-red ratio, the blue share and whether the ultraviolet and infrared channels are separate and switchable. A supplier who cannot send an SPD file has not measured the spectrum they are selling.
The evidence is fairly consistent on what spectrum does. Photon quantity drives biomass, and spectrum shapes morphology, secondary metabolism and quality. The 2021 review by Danziger and Bernstein reached that conclusion after comparing spectrum studies in cannabis. The practical consequence is that a spectrum premium cannot fix a photon shortage, and a photon surplus with a poor spectrum still produces a canopy that stretches, crowds and shades its own lower sites.
A workable commercial recipe is a wideband white base with added 660 nm red, a separate far-red channel, and ultraviolet on its own channel for the final weeks. Far-red at the end of the photoperiod nudges the phytochrome balance and can shorten the internode stretch that follows the flip. Ultraviolet is best treated as a finish tool run for a few hours mid-photoperiod in the last two to three weeks, not as a headline number, and it needs its own safety protocol for staff working under it.
The stage numbers that go with that recipe are not exotic: 150 to 300 µmol/m²/s for clones, 400 to 600 for vegetative growth, 700 at the flip and 1,100 to 1,200 at peak flower, with a DLI of 30 to 45 mol/m²/day in ambient CO₂ and 40 to 65 with enrichment. Spectrum changes belong at the flip and in the final two weeks, not on a weekly schedule that nobody can reproduce.
The guide to choosing a spectrum for cannabis flower covers the band-by-band evidence and the trials behind it. If the photon budget is tight, spend on photons first and channels second.
Four documents decide whether a fixture can be installed, rebated and insured: the electrical listing, the DLC horticultural listing, the IP rating and, outside North America, the CE and RoHS declaration. Each one is model-specific. A certificate issued for a 600W model does not cover a 1,300W model with a different driver, and a brochure logo is not a listing number.
The electrical listing, usually ETL or UL to the UL 8800 horticultural standard, is what an authority having jurisdiction checks before a room is energised. Ask for the file number and the exact model string on the label, then confirm the driver carries its own recognition. An inspector who cannot match the model on the fixture to the model on the certificate will fail the installation regardless of how the fixture performs.
The DLC listing is a model-level performance claim and, in most US utility territories, the key to a rebate. Standard and Premium are different tiers with different thresholds, and a Premium claim is only usable if the exact model appears in the Premium qualified products list on the date of purchase. Treat the listing date as part of the specification, because a model can drop off the list when a driver or diode changes.
| Document | What it proves | What to verify | Who asks for it |
|---|---|---|---|
| ETL or UL listing | Electrical safety to the horticultural lighting standard | File number, exact model string, driver recognition | Inspector, insurer, facilities manager |
| DLC listing | Model-level performance and quality; rebate eligibility | Model in the qualified list, tier and listing date | Utility rebate program, procurement |
| IP rating | Dust and water ingress protection | IP65 or better for washdown rooms; gasket and cable entry | QA, insurer, sanitation lead |
| CE, RoHS, UKCA | Market access and hazardous substance limits outside North America | Declaration of conformity with the model and standards listed | EU or UK buyer, customs |
The IP rating deserves a sentence of its own. A room that is washed down between cycles needs IP65 or better at the fixture and at the cable entry, not just at the lens. Condensation forms on the driver side of a bar as readily as on the diode side, and a fixture rated IP54 for a dry room will fail in a wet one within a year.
The voltage decision changes the electrical cost more than the fixture choice does. A 1,300W fixture draws about 5.4 A at 240V and about 10.8 A at 120V. On a 20 A circuit with the 80 percent continuous load rule, that is two fixtures per circuit at 240V and one per circuit at 120V. The same room therefore needs roughly twice as many circuits, twice as much home-run wire and twice the breaker spaces if it runs at 120V, which is a five-figure difference in a room of any size.
Driver location is the next line. An external driver moves the heat source and the service point out of the canopy, which helps in low rooms and makes driver replacement a five-minute job instead of a fixture teardown. It also adds cable, mounting brackets and a second enclosure to the bill. Neither choice is wrong; the error is comparing a fixture price that includes an internal driver with one that does not.
Mounting hardware, controllers, sensors, labour and spares belong in the same sheet. Suspension kits, ratchets, ceiling anchors and leveling wire are small lines that add up across forty fixtures, and a room that skips the spare driver is a room that loses a week when one fails in week six of flower.
| Cost line | Planning share of installed cost | What to ask the supplier for |
|---|---|---|
| Fixtures | 55 to 70 percent | Delivered price per fixture and price per µmol/s |
| Mounting hardware | 3 to 6 percent | Kit list per fixture, anchor type, ceiling rating |
| Electrical circuits and panel work | 10 to 20 percent | Load schedule, breaker count, wire gauge, voltage |
| Controls and sensors | 4 to 8 percent | Protocol, zone count, controller compatibility |
| Labour and commissioning | 8 to 15 percent | Hours per fixture and a PPFD verification pass |
| Spares, freight and duty | 5 to 10 percent | Spare count, lead time, landed cost, driver price |
A quote that lists fixtures only is not an installed cost. The difference between the two is usually 30 to 45 percent, and it lands in the project budget whether or not it was in the quote. Ask each supplier for the same six lines and the comparison becomes honest.
Every watt that enters a room leaves as heat. The conversion is fixed at 3.41214 BTU per hour per watt. Twenty 1,300W fixtures add 26 kW, which is about 88,700 BTU/h, or 7.4 tons of sensible cooling, before envelope gains, people, dehumidification and any other equipment in the room. A facility that buys the fixtures and then discovers the cooling gap is buying a second project.
The path the heat takes differs by technology and changes how the room should be controlled. A high-pressure sodium hood sends a large share of its energy down as radiant heat, so the canopy absorbs it directly and leaf temperature can run 2 to 4 °C above room air. An LED bar sends little radiant heat to the plants; most of its waste heat leaves the fixture body and driver by convection, which can be captured near the ceiling if the layout gives it a path. That is why a room converted from HPS to LED often needs the air setpoint raised 1 to 3 °C to hold the same leaf temperature and VPD, rather than the same setpoint with less cooling.
Layout matters as much as technology. A bar mounted 6 in above the canopy with no extraction above it can overheat the top layer even at modest wattage, because the convective plume has nowhere to go. The fix is airflow and clearance, not a dimmer. The VPD guide covers the leaf-to-air relationship and the setpoints that follow a lighting change.
L90 is the point at which a fixture retains 90 percent of its initial photon output. Q90 is the DLC version of the same idea, with a 36,000 hour requirement in the current specification. Ask for the TM-21 projection from LM-80 data and the driver case temperature it assumes, because a lifetime number without a temperature is a number about a different fixture. A driver rated at 60 °C case temperature in a room that runs at 45 °C ambient will not deliver the same hours as one rated at 80 °C.
Warranty terms should be read as a service document rather than a marketing line. What is covered matters more than how many years: driver, bars, coating, labour, freight and the diagnostic time. A five-year warranty that excludes the driver covers the part that fails first. A warranty that requires the fixture to be shipped back to the factory costs more in freight and downtime than the replacement is worth.
Spares and obsolescence are the year-six questions. Can a bar or a driver be replaced in the room, with a part number published and stock held locally? Will the model still be sold in three years, or has the series been renamed twice since the quote? A room with forty fixtures from a discontinued series becomes a room with forty different light outputs after the first failures, which is a data problem before it is a lighting problem.
| Question | An answer that holds up | An answer that should worry you |
|---|---|---|
| What is the L90? | L90 at 36,000 hours or more, with TM-21 data and a stated case temperature | "50,000 hours" with no L-level or temperature |
| Is the driver covered? | Driver included for the full warranty term | "The driver is a consumable" |
| How fast is a replacement? | Local stock with a 48 to 72 hour dispatch commitment | Factory lead time only, measured in weeks |
| Can a bar be replaced? | Field-replaceable bars and drivers with published part numbers | Sealed unit with no service parts |
| Will the model still exist? | A stated availability window for the series | A new model number every season |
Form factor is a room decision disguised as a product choice. Wide bars spread photons evenly across a rectangular bed and fit under low ceilings, which is why a 4 × 6 ft bed is usually served by one wide-bar fixture rather than two smaller ones. Modules are shaped to replace a hood one for one in an existing layout, so a room can change technology without re-planning every suspension point. Fold-out fixtures exist for transport and storage, not for performance.
Mounting height follows from the map and the trellis, not from a rule on the box. Most commercial bar fixtures are hung 6 to 12 in above the canopy in flower and higher in veg, and the number that matters is the PPFD at the canopy top and at the lower flower sites, not the height on the spec sheet. Leave enough clear height for the week-three stretch, and keep the aisle clear so a person can inspect the crop without moving a fixture.
Under-canopy lighting is the second layer, and it belongs in the same purchase decision rather than a later upgrade. Bars in the 120W to 300W class mounted at lower flower sites add photons where the top light no longer reaches, typically at 20 to 30 percent of the top-light PPFD, and they need their own zone and their own heat and airflow plan. A combined top and under-canopy set, such as the 1,240W kit built around a 1000W top light and two 120W bars, is easier to commission than two separate purchases because the control and the mounting are designed together.
Most commercial rooms speak 0-10V. Some speak PWM, some use a proprietary bus, and many run a daisy chain over RJ45 between fixtures in a zone. Before buying a controller, confirm which of those the fixtures accept and how many fixtures one channel can drive. A fixture that dims beautifully on its own but cannot join the room's existing control is a manual fixture with extra steps.
Dimming depth is worth a specific question. Below about 10 percent, some drivers flicker, some lose several points of efficacy and some shift spectrum, which matters in a propagation room running at low output for weeks. Ask for the dimming curve and the minimum stable output, not just the dimming range.
Ramps and channel control are the operational features that pay back in consistency. A 15 to 30 minute sunrise and sunset ramp avoids the light shock that shows up as leaf curl in propagation, and separate ultraviolet and infrared channels let the finish schedule change without swapping the main spectrum. Both are configuration decisions, so they need to be in the purchase order rather than discovered after delivery.
Three cost metrics get mixed together in lighting purchases. Cost per fixture is a procurement number. Cost per µmol/s is an engineering number. Cost per gram is the only one that describes the business. Fixtures are compared on the second, decisions are made on the third.
Cost per µmol/s is the fastest way to compare quotes with different wattages. Divide the delivered price by the PPF. If fixture A costs 1.4 times fixture B but delivers 1.6 times the photons, A is cheaper per photon, and the room needs fewer of them. Run the same arithmetic on the installed cost, including circuits and mounting, because a cheap fixture that needs its own circuit can lose the advantage in the panel.
Cost per gram needs your own numbers. Take grams per square foot from the last two cycles in the same room, then change one variable at a time. A lighting change that also changes temperature, CO₂ or irrigation is not a lighting result. The payback arithmetic for a like-for-like replacement is straightforward: 409 W saved per fixture, 4,380 flowering hours, $0.12 per kWh, about $215 a year per fixture. Against a capex difference of $600 per fixture, that is a simple payback near 2.8 years before HVAC and maintenance savings, and the number moves with your tariff and hours.
Grams per watt comparisons across rooms are close to meaningless unless the veg time, CO₂, strain and canopy management match. A room that vegs for 28 days will beat a room that vegs for 14 on grams per watt while producing less per square foot per year. Compare grams per square foot per cycle, then multiply by cycles.
Most of the failures in this guide are visible before the purchase, in documents the supplier either has or does not have. The ten below are the ones that recur in real quotes.
Any one of these can be answered with a document. A supplier who answers eight of the ten is usually a supplier who can answer the other two, and a supplier who answers two is selling a box with a logo on it.
Two quotations can carry the same 2.9 µmol/J, the same 4 × 6 ft footprint and the same five-year warranty and still describe two different products. The difference sits in what a spec sheet does not print: the diode bin, the driver, the thermal interface, the burn-in hours and whether next month's batch matches this month's sample. The spec sheet describes the design intent; the batch describes what you hang over the canopy.
The full process, including batch change control, AQL acceptance, contract clauses and a weighted scorecard, is in how to choose a grow light supplier.
You do not have to arrive with a model number. Give five inputs and a supplier can return a configuration instead of a price list: crop and stages, room dimensions with bed layout and mounting height, CO2 level and cooling capacity, current fixtures and electrical capacity, and a target PPFD or yield in g/ft² with the date the room must be lit.
Give the budget in a unit, not a lump sum: a figure per bed or per µmol/s produces proposals that compare. Then read the reply for three signals: does the supplier ask about the parameters you left out, does the proposal lead with µmol/s and a PPFD map, and does it say which models are wrong for this room.
For buyers who want to see how the checklist maps onto hardware, our own line is organised around the same decisions. The Golden 1300W wide-bar fixture covers a 4 × 6 ft bed with 3,900 to 4,550 µmol/s at up to 3.5 µmol/J, six bars and 3,984 LEDs, 0-10V dimming with RJ45 daisy chain, ETL and DLC listing on the listed configurations, and a five-year warranty confirmed per order. The 4 × 6 ft series runs from 1000W to 1400W for rooms that want the same footprint at a different output, and the 1:1 HPS replacement module covers 800W to 1200W for retrofits that keep the existing hood layout.
For the second layer, under-canopy bars run from 120W to 300W across four cover variants in 4 ft and 8 ft lengths, and the 1,240W Grow Light Set combines a 1000W top light with two 120W bars so the control and mounting arrive together. External-driver and fold-out versions exist for rooms where heat separation or transport is the constraint, and OEM and ODM work covers spectrum, PPFD and housing changes for buyers who need their own specification.
What we need to quote a room is the same list the guide has just walked through: a plan with the canopy footprint, the mounting height, the current fixtures and any PPFD readings, the stage plan and the electrical voltage available. With those five inputs we return a fixture count, a PPFD map at your mounting height, wall watts and a quotation. That is also the point at which the checklist stops being a checklist and becomes a purchase order.
Efficacy, heat path, spectrum control and service life. LED fixtures run 2.5 to 3.5 µmol/J at the wall, against 1.3 to 1.9 for double-ended HPS, 1.3 to 1.7 for CMH and below 1.1 for fluorescent tubes. LED sends most waste heat up by convection instead of down as radiant heat, which changes how the room is cooled. LED also dims and shifts spectrum by channel, while HPS and CMH are fixed-output lamps with a warm-up and restrike period.
On the same photon output, a 2.8 µmol/J LED needs about 38 percent fewer wall watts than a 1,000W double-ended HPS system: 661 W against 1,070 W for 1,850 µmol/s. The saving per fixture is about 1,790 kWh over 4,380 flowering hours, before the HVAC load that follows the watt reduction. Actual savings depend on the efficacy of the exact models, the tariff and the hours run.
No. The number on an LED fixture is usually input watts, while a 1,000W HPS rating describes the lamp, with another 70 to 120 W lost in the ballast. A 1000W LED at 2.8 µmol/J emits about 2,800 µmol/s, which is 40 to 65 percent more photons than a 1,000W double-ended HPS system. Compare PPF in µmol/s, not the wattage on the label.
Prices vary too widely by wattage, listing and configuration for a single figure to be useful. Compare delivered price per µmol/s instead: divide the price by the PPF, then add the electrical, mounting and control cost for that fixture. A fixture that looks 30 percent cheaper per unit often loses that advantage once the cost per photon and the circuit count are included.
Not in every jurisdiction, but a DLC horticultural listing is the model-level proof most US utility rebate programs require, and it is the easiest way to verify that the efficacy on the spec sheet was measured to a published standard. If you are claiming a rebate, confirm that the exact model is in the qualified products list on the purchase date.
At a 1,000 µmol/m²/s setpoint with a 12 percent edge margin, the bed needs about 2,500 µmol/s. One 1000W bar at 2.8 µmol/J covers that at 80 to 90 percent output, and one 1,300W fixture at 3.0 to 3.5 µmol/J covers it at 55 to 65 percent. Check the PPFD map at your mounting height before fixing the count.
Plan for 700 µmol/m²/s at the flip, 850 to 1,000 through early and mid flower, and 1,100 to 1,200 at peak, with a DLI of 30 to 45 mol/m²/day in ambient CO₂ and 40 to 65 with enrichment. Buy a fixture that can reach the peak setpoint at part load rather than one that runs flat out to reach the average.
Ask for L90 at 36,000 hours or more, which is the DLC Q90 requirement, together with the driver life at rated case temperature. A five-year warranty is common on commercial fixtures, but the terms matter more than the length: the driver, labour and freight should be covered, and replacement parts should be available locally.
If the canopy is dense and the lower flower sites are shaded, yes. Under-canopy bars in the 120W to 300W class add photons at the lower sites, usually at 20 to 30 percent of the top-light PPFD, and they need their own zone and airflow plan. Buying them with the top light keeps the control, mounting and commissioning in one design.
Ask for evidence about the shipment rather than the brochure: a batch test report with the batch number and date, the burn-in hours and the field failure rate over the last twelve months, a listing file number that matches the quoted model, a production schedule with weekly capacity, and a spare parts list with part numbers. A supplier who answers five of these in writing is verifiable before the order. A supplier who answers only the price is not.
Tell them what you grow and which stages the room runs, the room dimensions and bed layout with the mounting height, the environment (CO2 level, cooling capacity, humidity), the fixtures and electrical capacity you have now, and your target PPFD or yield with a budget per bed or per µmol/s. With those five inputs a supplier can return a fixture count, a PPFD map and the arithmetic, instead of a unit price.
Buying grow lights for cannabis comes down to the same ten decisions every time: room, target, photons, fixture count, efficacy, spectrum, listing, heat, form factor and terms. The eleventh question is who answers them, and whether that company can prove the batch, hold the schedule and supply a driver in year four. A supplier who answers the ten in writing is selling a lighting system. A supplier who answers the price and nothing else is selling a box.
Send the room plan, mounting height, current fixtures and any PPFD readings. We return a fixture count, a PPFD map at your mounting height, wall watts and a quotation in the same units, so the comparison is arithmetic rather than a brochure.
