I Documented 23 Renewable Energy Failures. The Root Cause Is Always the Same.

In September 2022, I stood in a client's driveway watching smoke curl out of a garage that was supposed to hold a "simple" solar battery project. Twelve LiFePO4 cells. A BMS he'd bought from a discount marketplace seller. A charger that arrived in a plain brown box. Total cost: about $1,900.

The fire never fully caught. The smoke came from a cable lug that had been torqued by hand "until it felt right." But the root cause went deeper than a loose connection. It traced back to the questions he asked before he ordered anything:

"How to make lifepo4 battery." "Whats a surge protector." "Level 2 charger for chrysler pacifica."

I recognize those searches because they're the same ones that land in my analytics every week. On the surface, they're innocent. People want to save money, learn a skill, or charge a fleet vehicle without paying dealer markup. I get it. I started the same way—I once "saved" $600 building a battery pack from liquidation parts. It failed in eleven days.

For 9 years, I've handled renewable energy equipment orders and system designs for commercial clients. I've personally made—and documented—23 significant mistakes, totaling roughly $47,000 in wasted budget. I keep the list because every item on it was something I could have caught if I'd asked better questions. Now I maintain our team's pre-commissioning checklist so nobody has to repeat my education.

What these questions have in common

Here's the thing: each of those search queries treats the system as a collection of independent parts. A battery is a box of cells. A surge protector is a plug-in device. An EV charger is a box on a wall. If you know how to wire, you can assemble them, and you're done.

That framing is the original sin of every failed project I've documented. The parts you buy—the battery, the charger, the surge protector—are just the visible 10% of the system. The other 90% is protection coordination, voltage sensing, thermal management, and monitoring. That invisible 90% is where systems go to die.

What I mean is that the physical parts aren't really what you're installing. You're installing an engineered system with four layers: energy storage, power conversion, protection, and monitoring. The parts are just what you touch. The layers are what keep it alive.

Nobody searches for the layers. Nobody searches for "how to coordinate protection devices" or "how to pick a potential transformer." They search for the shiny parts, and then they learn—the expensive way—that the shiny parts were the cheapest part of the system.

The first thing everyone skips: voltage sensing

Let me start with the least glamorous component in the entire installation: the potential transformer. If you know what it does, you're in the top 1% of DIY solar people. If you don't, this is worth reading twice.

A potential transformer steps high voltage down to a safe, measurable level for instruments, relays, and meters. It's how your system knows the actual voltage on the bus. Every inverter, battery management system, and protection relay depends on that voltage signal to make decisions.

The garage battery bank didn't have one. The inverter relied on internal sensing, which worked until the input voltage dipped below spec—and then it shut down at exactly the wrong moment, leaving the BMS and charger arguing with each other.

If you're sourcing an ABB potential transformer for a commercial project, you already know the physics. What you might not know is why it belongs in a system that many DIYers still call "just a battery." The answer: without accurate voltage sensing, your controls are making decisions on fiction.

Dodged a bullet myself last year when I double-checked the PT secondary wiring before closing up a cabinet. One wrong terminal block away from feeding 240V back into a control board—a $14,000 panel replacement and a three-week delay. That's when I added "PT secondary check" to our pre-commissioning list.

Why "whats a surge protector" is the wrong question

I get why people ask whats a surge protector before buying one. But the question is incomplete. The conversation you actually need to have is about surge protection as a coordinated architecture—not a single box.

Look, I'm not saying budget options are always bad. I'm saying they're riskier. There's a category of device called a surge protective device, or SPD, which is tested under UL 1449 to survive repeated surge events at its rated capacity. There's also a category of "surge protector" that is essentially a power strip with a metal-oxide varistor inside—and it can fail silently after one big transient, leaving you with zero protection and no indicator.

Under UL 1449, a surge protective device must withstand repeated surge events at its rated capacity. Consumer power strips with integrated suppression are not held to the same test criteria (Source: UL 1449, 4th Edition; ul.com).

Here's something vendors won't tell you: an SPD is only as effective as the grounding and coordination around it. You need a coordinated set—Type 1 at the service entrance, Type 2 at the distribution panel, and sometimes Type 3 at the equipment—with conductor lengths kept as short as possible. If you just slap one device on the equipment you care about, the surge can still enter through another path and find ground through your expensive inverter.

For commercial and industrial installations, I spec ABB OVR surge arresters because they publish test data and replacement criteria—which matters when a warranty inspector asks why your "protected" system failed. A $15 power strip doesn't come with a coordination chart. Fine. But if you're charging a fleet of plug-in hybrids, you need the architecture, not the box.

The Pacifica charging story (and the cost of the parts mindset)

Take the search for a level 2 charger for chrysler pacifica. A client of mine had six Pacifica Hybrids in a corporate fleet. They bought six consumer-grade Level 2 charging stations, shared one circuit, and skipped surge protection because "the chargers said it was built in."

Eight weeks later, a construction crew down the street hit a 12kV cable. The resulting transient took out the control boards on two charging stations. The chargers were still under warranty, but the manufacturer's warranty specifically excluded surge damage.

The client paid $680 each for replacements, plus $400 for the electrician, plus a week of vehicles running on gas alone. Total: around $1,760—or rather, $1,760 plus the replacement units taking three weeks to arrive because the vendor was back-ordered. We'd already installed ABB charging equipment with integrated RCD protection on a second site by then.

That's the cost column of my ledger. Item by item, it doesn't look catastrophic. But it adds up: $890 in redo costs here, $3,600 in mis-specified instrument transformers there, a $2,100 insurance deductible after a home office surge event. The $47,000 total is spread across 9 years, 23 documented failures, and a lot of uncomfortable phone calls.

The systems in my ledger don't fail because the parts are cheap. They fail because the decisions between the parts were missing. And the consequences are rarely just money—they're delays, denied warranties, and occasionally, a garage that almost burns down.

The deeper cost: you don't know what you don't measure

The irony is that most failures I've documented would have been caught by monitoring tools the client already had. One client had an ABB login to the inverter portal and never used it. Another had a dashboard but never set up alerts, so when one phase sagged for six weeks, nobody noticed until the compressor motor fried.

What most people don't realize is that monitoring is only as good as the sensors behind it. A portal login is the last step, not the first. If you don't have potential transformers and current transformers sending accurate data to that portal, your dashboard is just a pretty graph of estimates. Put another way: the ABB login gives you access to the truth, but only if the measurement chain feeding it is telling the truth first.

I didn't learn this until Q1 2024. We installed six potential transformers with the wrong secondary wiring at an industrial site. The system ran—or rather, appeared to run—for three days. The inverter was reading 10% low and compensating. By the time we caught it, the inverter had put thousands of hours of wear on a contactor that should have lasted years. $450 in replacement parts, and the only clue had been a pattern in the monitoring data that nobody was watching.

Where I landed: what I actually recommend now

After 23 documented mistakes, I've stopped recommending "the best products" and started recommending the right system boundary for each customer. Here's my honest take.

DIY is fine if: you're building an isolated, low-voltage system (12V or 24V) that doesn't interconnect with the grid, you buy matched LiFePO4 cells from a reputable source within ±3% internal resistance, you use a UL-listed BMS, and you follow torque specs exactly. Camper vans and trolling motors qualify. I've built several myself.

Bring in an engineer if: the system is grid-tied, over 48V, or charging vehicles you depend on. That includes home EV charging. A level 2 charger for a Chrysler Pacifica is not a weekend DIY project—it's a continuous 240V load that needs a dedicated circuit sized for the vehicle's 6.6 kW onboard charger, proper grounding, and coordinated surge protection (Source: SAE J1772; FCA documentation).

And if you're in the other 20%—unreliable utility voltage, old wiring, or a history of lightning damage—don't rely on consumer-grade gear. Get a coordinated SPD system, proper instrument transformers, and actually watch the monitoring data. The value of engineered components isn't just the hardware; it's the certainty that the spec sheet is true.

I've also stopped assuming the lowest quote wins. The lowest quoted price is rarely the lowest total cost. When you add rework, downtime, warranty disputes, and the hours you lose supervising a failed installation, the consumer-grade box has usually cost double what an engineered component would have.

Our pre-commissioning checklist now starts with three questions that would have saved me $47,000: What is the protection architecture? What is the measurement architecture? What is the monitoring architecture? If you can't answer all three, you're not ready to buy parts. Protection, measurement, monitoring—in that order.

I'm still documenting failures, and I still make mistakes. The difference is that they're smaller now, and they come with fewer phone calls. If you're about to build something, start with those three questions. Your garage—and your wallet—will thank you.