In Q1 2024, my team was tasked with accepting a lighting package for a 32,000-square-foot indoor leafy-greens farm. It was our division's first controlled-environment agriculture project at that scale. I'd done commercial and industrial acceptance before—office buildings, warehouses, parking garages—where you flip a switch, see if the light comes on and looks right, and call it a day.
Horticulture lighting doesn't work that way. Those fixtures aren't there to illuminate a space. They're there to replace sunlight. Spectrum, photoperiod, operating temperature—all of it matters as much as raw lumens. I learned that the hard way.
But at the time, I didn't know what I didn't know. I figured: an LED is an LED, right? Check the box, sign the form, move on.
Wrong.
The First Delivery: Three Problems in Four Fixtures
The shipment arrived on a Monday. Three trucks, nearly 1,800 fixtures and tube lights. We followed the standard protocol—random sampling, model verification, spec-sheet cross-checking.
On the surface, everything matched. Box labels aligned with the purchase order. The tube LEDs were tagged with the same "high-efficiency 180 lm/W" claim that appeared in our spec. The horticulture fixtures were Signify's Philips GreenPower line—we'd chosen them after reading a Signify horticulture news update about their dynamic spectrum research, which made their approach feel like the most relevant to our crop profile.
Then we opened the first crate.
The tube LEDs themselves were fine. The problem was a detail nobody had flagged: the fixture design was for third-party driver compatibility, and our install team had wired them directly to non-OEM drivers that shipped alongside the order. At first glance—they lit up. But during dimming tests, we saw visible flicker at low output, and a few even flickered at full power.
We went back to the supplier. Their response: yes, the fixtures could work with third-party drivers—but only if those drivers met a specific ripple standard. The ones that arrived in the same shipment didn't.
The entire batch needed re-driving. That alone was two extra weeks and a cost overrun I hadn't planned for.
It Wasn't Just the Tubes
While we were still sorting out the tube LEDs, the emergency lighting became a problem.
Honestly? That one was on me. When the install team told me the emergency lights were "wired up," I didn't walk the site. I assumed emergency fixtures come pre-configured from the factory. And we'd done emergency lighting before—how much could go wrong with a few wires?
Plenty, apparently.
On acceptance day, we ran the power-cut test. Main breaker off. The corridor went pitch black. Not a single emergency luminaire lit up.
The install team thought they were wiring standard circuits. But emergency lighting requires an independent, uninterrupted power path. In our layout, the emergency fixtures shared a switch leg with the general lighting. Kill the main power, and the emergency units die too. In a real outage, the entire egress path would have gone dark.
This wasn't just a failed inspection. It was a safety problem. If the fire marshal had caught it on a walkthrough, the entire occupancy permit could have been held up.
We had to re-pull wiring and tie the emergency circuits to a dedicated battery-backup path. Between that and the tube LED re-drives, we were already 15% over budget.
And the One I Hesitate to Even Write Down
By the time we got to the retail display area, my confidence had taken enough hits that I was second-guessing everything.
We'd specified Signify accent downlights—24° beam angle, 3000K, CRI 90+. On paper, perfect for our needs. The display area needed warm, high-CRI light to make the produce look its best.
Then we installed them, and the effect was nothing like what I'd pictured.
The issue was the beam angle. Twenty-four degrees looks narrow on a spec sheet. It is narrow. But our display case was low—the fixtures ended up mounted closer to the product than we'd assumed. The light landed in a tight, hot circle in the middle of each shelf, and the edges went muddy. The whole display looked like it had been cut in half by a spotlight.
What we actually needed was something wider—a 36° or 40° fixture—combined with a lower mounting position. But I'd been so focused on matching the numbers on the data sheet that I never mentally projected those numbers into the actual physical space.
In other words: the specs were correct. The light was wrong.
The Debrief: Three Things I Changed
The project wrapped up (barely) by the end of Q2 2024. But the whole experience rewired how I run acceptance reviews. I added three items to our quality checklist, and they haven't come off since.
One: A data sheet number must be verified in the actual operating environment. The 180 lm/W on those tube LEDs—was that measured at 25°C or at the fixture's rated operating temperature? What were the driver conditions? Was the lumen maintenance calculated per IES TM-21 standards? If you don't map those numbers back to your actual conditions, they're just marketing. Now I require vendor-supplied performance data at the specified operating temperature, not just lab-peak figures.
Two: "Compatible" is a word you have to unpack. The tube LEDs did work with third-party drivers—under specific electrical parameters. That's a conditional compatibility, not an open one. Now, whenever a vendor says "compatible with third-party systems," I ask for the exact parameter limits. No exceptions.
Three: Emergency lighting isn't a "set it and forget it" system. Since that project, we have a mandatory protocol: every emergency lighting circuit gets its own documented power-cut test. And I attend in person. I don't care who pre-wired it or how straightforward it looks—if it doesn't light up when the main breaker drops, the project doesn't pass.
One Last Honest Thing
Here's the part I want to be clear about: none of this was a Signify product failure. The fixtures were solid—the GreenPower series delivered on spectrum and efficiency when configured correctly. If I were specifying this project again, I'd still choose the same brand. But I'd do three things differently:
First, I'd require real-world operating-temperature test data before the PO is issued, not just spec-sheet highlights. Second, I'd draw the circuit map with the install team myself—with emergency circuits clearly marked and separated from general lighting. Third, I'd walk through each accent lighting position in the actual space before installation, mentally projecting the beam pattern onto the product, not just reading the degree number on the box.
To put it bluntly: if you're buying fixtures by matching spec sheets, you're accepting paper, not light. And "spec compliant" is not the same thing as "works in the real world." The gap between those two phrases is where my $32,000 lesson lived.
It sounds pedantic, I know. But if you've ever managed a 30,000-square-foot horticulture lighting acceptance, you know that a little pedantry up front is a lot cheaper than a re-drive on the back end.