Here are the technical notes from my breakfast meetup with Mark Baldassari on 2025-03-08. (Mark recently retired from Enphase.)

Early on, some microinverters (micros) failed because of water intrusion. The trick was to pot (i.e. flood) the enclosure with a protective compound. That way, there'd be no place for the water to go.

There was a lot of troubleshooting (because all bets are off in the field). In one case, one Enphase gateway (Envoy) collected data from a neighbor's Enphase micros. This issue was resoloved with power-line communication (PLC), which isolated the systems from each other.

With PLC in general, one has to deal with noise on the power lines. ...Noise from loads will find its way to the power lines between the micros and gateway (because the entire site's electrical system is interconnected). Some power lines are noisier than others. Excessive noise at one house was traced back to a night light in the kitchen (it turned out this was a prototype night light, with a bypass-capacitor issue).

Enphase has a testing lab in Petaluma. They cycle micros and other devices thru extreme conditions. The micros that work... boring, throw them away. The ones that fail... are coveted. Take them apart, and figure out the specific point of failure. (This is the sort of testing that has gotten Enphase devices rock solid over time.)

Enphase's S-series micro was the first of their micros to meet California's Rule 21 requirements. With Rule 21, the distributed network of behind-the-meter inverters help stabilize the grid (e.g. inject reactive power to bring grid voltage up, and absorb reactive power to bring it down). ...One inverter isn't significant... but the inverters in aggregate significantly help out.

If inverters trip off because of grid instability...

IEEE 1547-2003 required the inverters to stay off for five minutes. Grid managers: "Don't muck this up while we investigate and resolve the issue".

Rule 21 _requires_ that inverters restart after fifteen seconds. Grid managers: "Come back fast to help us resolve this issue!"

Mark confirmed that the required grid-support functionality reduces solar production. Maybe a percent or two on average.

Larger electric grids have more inertia (i.e. stored kinetic energy in spinning generators i.e. turbines). Islands (with less grid inertia) have a harder time maintaining grid voltage. High-precision oscillators help maintain a 60-hertz frequency.

Rule 21 is one "grid profile". Different grid profiles are required in different parts of the country (based on local situations and priorities). Enphase installers choose the appropriate grid profile in the commissioning process (typically using the default). I asked if there would ideally be one grid-profile standard across the country. Mark noted that there's benefits diversity.

Mark gave props to Enphase's John Berdner. "John is Enphase's grid guy". John got SMA into the U.S. market. John played a core role in getting "power control systems" into the NEC, starting with NEC-2023 705.13.

Mark also gave props to Enphase's Martin Fornage. "Martin is the brain of the microinverter."

UL 3141 is currently an "outline of investigation", that's in process of becoming a standard. It has UL 1998 as a dependency on the software side, and UL 991 as a dependency on the hardware side.

A critical aspect of UL 3141 is the "Failure Mode Effects Analysis" (FMEA). ...What are ways that the device could fail, and what would the consequences be?

To get UL 1998 listed, safety-related computer code (e.g. anti-islanding code, rapid-shutdown code) needs to be reviewed by a NRTL. But computer code changes over time, and applicable code updates need NRTL review). ...It's best to segment out safety-related computer code, and put checksums on it to verify that it hasn't been edited. The NRTL also expects well-written safety-code documentation.

Enphase uses ASIC chips in its devices. (These are custom-designed chips, not general microcontrollers.)

Enphase will offer a meter collar. The process of regulatory acceptance is going very well in California. And very poorly in Florida. (This is a state-by-state process.)

Enphase's forthcoming bi-directional direct-current EV plug needs to meet IEEE 1547 requirements.

Hot dang, thanks so much Mark for openly-sharing the delightfully-geeky knowledge. And thanks again for the awesome industry contributions.