Project note

Why an SRNE Inverter Showed 380V PV Voltage at Midnight (And Why the Inverter Wasn't the Problem)

2026-11-13 · Yuna Park

A Friday 16:47 Call

It was November 2024. I was sitting at the parts shelf in our distribution warehouse when my phone rang — Leo calling.

Leo runs a small off-grid solar installation outfit operating along the Arizona–New Mexico line. He's been buying SRNE hybrid inverters and LiFePO4 batteries from us for about three years. We're past the point of small talk.

"Do you still have stock on the SRNE HF series?"

That wasn't a stock question. That was a pre-commitment to a backup plan.

The situation: a residential 48V off-grid setup on an SRNE HF-series hybrid inverter. Battery bank was full, MPPT charging curve was clean during the day. But around 9 PM, the inverter started throwing "PV voltage high" — reading 380V on the PV input port. It was pitch dark outside. A string under load should read 0–5V at that hour.

My first read was that this probably wasn't an inverter problem. We carry the line, we've handled the after-sales, we've seen the factory test logs. The HF series is a mature platform. Reading 380V PV at midnight is not something it wakes up and decides to do on a quiet night.

I guessed it was grounding.

I asked Leo to walk the wiring around the combiner box and send me video.

Two Things in the Video That Didn't Belong

Leo walked the site and sent me a clip. Two details stood out.

One: the battery disconnect was a side post battery disconnect. The part itself isn't a bad one — but the negative leads were stacked on the same post, and the grounding pigtail looked like it had been re-terminated by someone in a hurry.

Two: the array mounting brackets had been moved. Leo said the client had hired a drone solar panel cleaning service a few days earlier, and then a handyman had added one extra panel at the edge while he was up there.

Neither of those things alone is fatal. Together, they made the 380V-at-night reading completely plausible.

What Was Actually Happening

I asked Leo to take two voltage readings that night. First, PV string to ground. Second, the side post switch positive terminal to inverter chassis.

The first reading had a few volts on it. On a dark night, a string should have almost nothing.

The second reading was non-zero as well.

That gave us the answer. The drone cleaning had contacted the PV cable — one conductor had been scraped along the edge of a metal bracket. Not a fatal break, but enough to form a leaky path to ground. The side post disconnect wiring had shifted the reference point on the battery negative path. At night, the inverter's PV sense circuit wasn't reading PV voltage — it was reading a mixed potential fed through the battery bank, the damaged conductor, the chassis, and the grounding point.

The inverter was reporting what it saw. It wasn't broken.

The Overnight Phone Call

We were on the phone from Friday evening until almost two in the morning. The sequence went roughly like this:

  • Leo stepped through each PV string with a handheld meter to isolate which one was leaking to ground. That took about forty minutes because the three strings were dressed identically and the labels had faded.
  • We found it on the east run of the roof — right below where the handyman had added the extra panel after the drone cleaning.
  • He re-terminated the side post disconnect negative leads and moved the grounding pigtail to the chassis bonding point specified in the SRNE manual.
  • He re-insulated the damaged cable, added two extra cable clips, and routed it away from the bracket edge.
  • We re-measured. The PV port reading on the inverter dropped back to zero.

I wasn't on-site. What I was doing on my end was pulling three years of after-sales logs looking for precedent. That took maybe twenty minutes — I found two prior tickets with the same pattern: "nighttime PV voltage high, inverter fine, root cause on the grounding and cable wear."

Around 1:40 AM, Leo said, "The alarm's gone."

I didn't relax immediately. I asked him to sit on it for fifteen minutes. It didn't come back. Around 2:05 AM I finally believed it.

Somewhere in the middle of the meter readings, Leo mentioned that his son had asked him earlier that evening how many planets in the solar system. "Eight," he told the kid. "And not one of them is as complicated as this junction box." I said I wasn't going to argue with either of those facts.

The Cost

That night itself wasn't expensive — Leo burned through maybe $400 in overtime and fuel, and the client got woken up three times, which was the bigger problem.

The part that actually bothers me is the chain behind it. To save about $30, Leo used a side post disconnect he had on the shelf instead of the positive-break spec we recommend. To save about $100, he didn't send the original crew back to inspect the array after the drone cleaning touched the mounts. Add it up: $130 saved, and it cost $400 plus a weekend where a client nearly cancelled the handover.

That math doesn't work. At least, that's been my experience every single time I've run it.

What I Changed

After this one, we updated two things in the handoff checklist we give installer clients.

One: the side post battery disconnect got its own line item — not a ban, a pre-check. Before install, the positive-break path and grounding reference have to be confirmed in writing. (I should note: this isn't a defect in the side post part. It's a sequencing question. Get the order wrong and the sense circuit will read things you don't want it to read.)

Two: any client who has had a drone cleaning, roof work, or any other service that touches the array brackets must have a visual pass done on the PV cables, bracket edges, and combiner box within forty-eight hours of that service.

Both rules sound obvious when you write them down. But in the emergency calls I've handled, the gap between obvious and process is usually a two-in-the-morning phone call.

The bigger takeaway: when an inverter throws "nighttime PV voltage high," suspect the grounding, the disconnect wiring, the string insulation, or a bracket someone touched — not the inverter itself. Industry documentation on PV system grounding (IEC 62548 and NEC Article 690 both address the general principle) treats this class of fault as a leakage-to-earth problem, not an inverter internal fault. The inverter is doing its job by reporting it.

As for the SRNE HF — it did nothing wrong. It just told us what nobody wanted to hear, at the exact time nobody wanted to hear it.

Yuna Park

Yuna Park

Yuna Park is a stationary battery analyst covering lithium-ion cells, LFP and NMC modules, battery racks, battery management systems, DC protection, thermal management, and transport preparation. She applies IEC 62619 industrial-battery safety tests and UN Manual subsection 38.3 transport evidence while comparing usable capacity, C-rate, depth of discharge, round-trip efficiency, temperature limits, cell imbalance, cycle-life conditions, and fault response. Her research helps storage engineers and buyers evaluate cell-to-system boundaries, operating windows, logistics evidence, maintenance needs, and warranted duty cycles.