November 7, 2023, 9:47 AM. I was squatting on the concrete floor of a client's garage in Simi Valley, staring at the shutdown screen on a freshly installed inverter. A refrigerator had just kicked on. The inverter had cut out. Third time in an hour.
The client was standing right behind me, phone out, ticking through the appliances that were supposed to stay on during an outage. His medical device. His lights. And now a fridge that wouldn't stay running.
Six weeks earlier, this had looked like a straightforward home solar battery bank installation. Panels on the garage roof, a 48V lithium battery, an EPEver 50A MPPT charge controller, and an EPEver 5kW inverter. The client had done his homework — he'd built his own DIY solar system model with daily and peak loads. The numbers looked clean. Then a refrigerator turned a clean design into an expensive lesson.
What the Design Looked Like on Paper
In September 2023, a homeowner contacted my team. He wanted backup for a handful of loads: a refrigerator, a medical device, a few LED lights, and a small space heater for winter. He'd calculated a daily consumption of roughly 2.7 kWh and a peak of about 3.1 kW. Not a big system. Not expensive.
We spec'd an EPEver 5kW inverter, a 48V 100Ah LiFePO4 battery, and an EPEver 50A MPPT charge controller matched to a modest 1.2kW array. In principle, fine.
Here's the part I got wrong: my battery selection assumed the inverter's full-load current would never approach the battery's continuous discharge limit. But on a 48V system, a 5kW inverter pulling full load needs roughly 110A at the DC side once inverter efficiency (typically around 90–93%) is accounted for. Most 48V 100Ah LiFePO4 batteries have a BMS rated at 100A continuous discharge. That's an uncomfortable number even before surge loads show up.
If I'd done that math properly the first time, I would have either upsized the battery or downsized the inverter to 3.6kW and matched the load accordingly. I didn't. The client's DIY model and my spec sheet both said "peak load 3.1 kW," and I trusted that number.
Why I Sized It the Way I Did
I'll be honest about what was going through my head. We had two options on the board — a 3.6kW inverter with a single 100Ah battery, and a 5kW inverter with a double battery bank. I went back and forth between them for about a week. The smaller one fit the client's stated loads and the budget. The bigger one gave us breathing room for future loads and surge. The client wanted to keep it affordable; so did I.
I chose the 5kW inverter with a single 100Ah battery. My reasoning at the time: 5kW gives headroom for future appliances; the single battery keeps the price down. It was the kind of compromise that feels responsible because it splits the difference between two options.
It wasn't responsible. It was a spec where neither the inverter nor the battery was matched to the other.
How Battery Storage Works — and What I Missed
If you strip it down, a home battery bank works in a loop: solar panels feed an MPPT charge controller, which converts panel voltage into the required charging profile for the battery. When the house calls for power, the battery sends DC current through the inverter, which converts it to AC for the appliances.
Every part of that loop has a limit, and two of them are the ones that matter for surge:
— The battery's BMS has a continuous and peak discharge limit.
— The inverter has a continuous and surge output rating.
In most residential sizing, people check the inverter surge against motor startup. That's correct as far as it goes. But they don't always check the BMS peak against the inverter's full-load draw, and they don't always account for efficiency loss at full load. I was one of those people.
The Failure and the Fix
On installation day, we powered up slowly. Lights worked. Medical device worked. Then the client opened the fridge, the compressor kicked, and the inverter went dark. The inverter hadn't failed — it was working exactly as rated. The battery's BMS had tripped its overcurrent protection and shut down. With no DC input, the inverter dropped the load. To a homeowner, all of that looks the same: the lights go out.
I had about four hours to figure out a fix because the client's backup window overlapped with a medical need that evening. I could either cap the inverter output and accept the lost headroom, or add a second battery in parallel and lift the BMS limit. I chose the second battery. I also reprogrammed the EPEver charge controller to the correct LiFePO4 profile, since the default lithium preset wasn't the same as the battery manufacturer's recommended curve.
I'm glad I caught it early. An hour later and I would have been packing up with a limited inverter, explaining a problem I couldn't name.
The system has been stable since. But I won't pretend it wasn't expensive. The second battery meant two extra days on site, an unplanned cable run, and about $1,900 in additional hardware — call it $4,200 in total impact, once you count the reworked paperwork and the apology discount I gave to keep the account.
Why I Keep Talking About Pakistan
Around the same time, a colleague of mine who runs a distribution business in Lahore asked me what a reasonable landed price for an EPEver 5kW inverter would be for his market. I gave him a rough range — those numbers shift with freight and channel, so verify locally — and then he asked what battery we'd pair with a system like that.
What he said next stuck with me: most of his DIY customers calculate battery capacity from daily energy use alone, and almost none of them ask about inverter efficiency loss or motor startup surge. I said that's true everywhere. Because it is. A DIY solar system model is really good at telling you how much energy you use on an average day. It's not set up to tell you what happens the moment a compressor kicks on.
So this isn't a geographic problem. It's a physics problem that every simplified model hides.
The Four-Point Check I Use Now
Every backup battery project we run now goes through the same four checks before I sign off on the spec:
- List every motor load — refrigerator, well pump, HVAC, washing machine — and record the locked-rotor amps (LRA), not the running amps.
- Match BMS peak to inverter full-load draw — including inverter efficiency (assume 90% if you don't have the datasheet). If the inverter can pull more DC current than the battery can supply, the system will find that edge eventually.
- Confirm the charge profile matches the battery chemistry — EPEver controllers allow custom LiFePO4 settings, and the default lithium preset isn't the same as every battery's recommended curve.
- Do a live startup test — turn on the largest surge load while other loads are running. Watch the inverter and the BMS, not just the lights.
The efficiency lesson here is the same one I learned the expensive way: a written checklist catches things memory doesn't. When I was relying on my own judgment, I missed one number. When the checklist came in, it caught things I would have skipped.
What I Do Differently
I still size systems for real-world loads, not worst-case ones. I don't over-engineer every installation, and I don't tell customers they need battery banks three times the size of their actual usage. But I do run the surge math, I do verify the BMS rating, and I do insist on a live startup test with the client present.
That Simi Valley customer ended up referring his neighbor. He didn't refer me because the first install went smoothly. He referred me because I came back the next day, fixed it, and explained exactly what went wrong in plain language. The system has been running for over a year now, and it hasn't tripped once.
If your system is dropping loads and the inverter says it's fine, don't assume the inverter is lying. Check the battery's continuous and peak ratings. Then check whether your inverter can actually pull more current than the battery can supply. It's a boring calculation that most people skip. Skipping it cost me a weekend, a difficult phone call, and about $4,200.