So, on sizing this "Solar Subpanel" that has a line side connection to the service:
1) Suppose for the moment there's no 20A circuit for an Envoy. Then in terms of 125% of the inverter output current, the panel could be 200A as I understand it: The Enphase IQ7+ datasheet lists a "maximum continuous output power" of 290 VA. Each 208V/20A circuit can support 11 of the IQ7+ microinverters (20*208/290/1.25, rounded down). That means the panel would have at most 198 connected microinverters, balanced on the 3 phases, which is the maximum a 200A bus can support (3*120*200/290/1.25, rounded down to a multiple of 3).
However, if the panel is subject to 705.12(D), none of the options in 705.12(D)(2)(3) seem to work for the panel, as 705.12(D)(2)(3)(c) is written in terms of breaker rating, not inverter output current rating. Does the presence or absence of a main breaker in this panel really affect whether or not it is subject to 705.12(D)? Seems like the panel should be subject to 705.12(D) regardless.
But wouldn't a 225A panel be sufficient under 705.12(D)(2)(3)(c)? When adding the current on the A bus, for example, the current from the A-B breakers will not add fully with the current from the A-C breakers, but there will be a factor of sqrt(3)/2. So with (6) 60A 2-pole breakers, the current on each bus could be at most 2*60*sqrt(3) = 208A.
2) Now the 20A circuit for the Envoy would push sum of the breakers on that bus over 225A. A 15A breaker for the Envoy would work, or if one of the 60A breakers could be downsized to 50A because there are 193 or fewer IQ7+ microinverters in the system, and the Envoy is connected to the phase(s) with the 50A breaker, that would work. Is that correct?
[BTW, this cycle I wrote a late proposed change to 705.12(D)(2)(3)(c) that would permit ignoring the smallest breaker in the panel, on the basis of a counting argument, Kirchoff's current law, and that a single breaker can't simultaneously be both a load and a source. I don't think the committee took the time to read and understand it, as the committee response was a non-sequitur. But it would be useful precisely for this 20A Envoy circuit breaker.]
Cheers, Wayne