On the first Tuesday of September 2024, I stood at the end of a 150-foot greenhouse with two pallets of brand-new lighting gear behind me and a client staring at the same controller screen I was staring at. The screen said No Devices Found. The client asked whether the lights would be running by Friday. I said, “It should work.” That phrase has cost me more than any bolt, bulb, or broken driver ever has.
The Controller vs. a Simple Timer
I run a small lighting-installation company. Most of our revenue comes from commercial horticulture, but we fill gaps with residential and specialty lighting work. I’ve been handling these orders for six years now, and I’ve made enough mistakes to fill a binder. This is one of the pages I’d rather burn.
The greenhouse client had approved a plan around 24 of the Gavita HGC906410 RS 1900e grow light fixtures. I also recommended adding a Gavita Pro 1700e LED controller. The controller would let them dim and schedule lights remotely, and they planned to expand in 2025. A cheap interval timer would have done the basic job, but the controller was the better long-term tool.
At least, that’s what I told the client. I believed it. To be fair, I still believe it. The problem was never the decision to go digital. The problem was the missing process around that decision.
Before ordering, I went back and forth for almost two weeks. The controller offered efficiency; the timer offered simplicity. On paper, the controller made sense. My gut said the client was paying for features I hadn’t tested in that exact combination. I went with the controller anyway, because I do think the industry is moving in that direction. What I didn’t do was read the full compatibility list in the current manual.
The Mistake Was Boring, Not Technical
We didn’t have a formal compatibility-check process. That sounds like an excuse, but it is the actual cause. Everyone on the team had a different piece of the job, and no one was assigned to confirm that the controller could talk to that particular fixture batch.
The purchase order looked right. The fixture model was right. The controller was right. But model numbers in the same ecosystem are not a promise. Somewhere in the documentation there was a supported-devices chart. I knew the chart existed. I didn’t open it before install.
The first sign came during commissioning. The controller powered on, the fixtures powered on, and the controller found zero devices. I tried the obvious checks: cables, connections, power. Then I called technical support and read serial numbers. The person on the phone asked for something I had not checked—the interface requirement for the RS line. I didn’t know what they were talking about. That is the moment I wanted the floor to open up.
The fix was not dramatic. It required an accessory that wasn’t included and a second service visit after it arrived. The exact part may not be relevant to everyone reading this; by the time you install a newer batch, it could be different. The relevant part is that I had assumed compatibility because the products shared a brand.
While I Waited, Two Smaller Lights Taught Me the Same Lesson
While the replacement part was being shipped, I had two smaller jobs on my bench. One was a Piaf chandelier for a restaurant lobby. The other was a selenite chandelier that needed a new cable after shipping damage. Both involved ceiling mounts. Both looked like “hang fixture, connect wires” jobs. But they were not identical.
The Piaf chandelier was lighter and had a smaller canopy. The selenite chandelier looked delicate but was heavy enough that its spec sheet called for a structural support rather than a standard electrical box. If I had treated them the same way, the selenite one would have been a hazard. It took two minutes to read the labels, and that small discipline reminded me what I had skipped on the main project.
A Light Bar, Same Story
A friend also sent me a video from his garage that week. He had installed a Can-Am shock tower light bar on his side-by-side. The mounting holes lined up. The light bar looked great. Then he turned it on and smoke came from the factory harness. The light bar needed a relay and a dedicated wire; the factory circuit couldn’t handle the load.
He didn’t need a better light bar. He needed to check the electrical side before installing. That is the same mistake I made with the greenhouse controller, just in a different package.
Cost and Close Calls
When I added the accessory, the freight, the extra labor for the second visit, and the two hours I spent re-booking the schedule, the mistake added about $1,100 to a job that should have been boring. The client wasn’t angry, which made it worse. I could see in their face that they expected more from us.
After that project, I built a one-page pre-install checklist for every lighting job. It sounds obvious, but the list includes the fixture model, the controller model, the firmware revision, the supported-device chart, the included accessories, the measured input voltage, and a dry-run pairing test before final mounting. Not every item applies to every job, but someone has to initial every line that does.
In the nine months since, that checklist has caught 12 potential failures before we powered anything on. One was a controller firmware mismatch. Another was a missing link component for a different brand. The total savings on those jobs was far more than a few thousand dollars, because the failures were caught while the fixture was still in a box rather than hanging from the ceiling.
What I’d Do Differently
If I could go back, I wouldn’t necessarily choose the cheapest model or the most expensive controller. I’d choose the same controller, but I’d add one rule: no part gets purchased until a human has verified the compatibility chart for that specific SKU.
To be fair, I still don’t think every project needs a smart controller. A timer works for many growers. A Piaf chandelier doesn’t need Wi-Fi. A Can-Am light bar just needs a proper relay. The equipment is not the problem. The problem is when we hurry past the boring paperwork and then wonder why the interesting part fails.
Efficiency is not doing things faster. Efficiency is doing things correctly enough that you don’t have to do them twice.
I am not 100% sure what the next version of the controller will support. Don’t hold me to that. But I am sure about this: the second service visit is always more expensive than the first five minutes with a spec sheet.