I'll leave others to comment on the regs side. If the regs can be seen as requiring an isolator, it seems silly in this case, probably because of a focus on central-inverter systems with many panels in a string. Two panels on a microinverter are probably not even series connected, but connected to separate inputs. Perhaps 40 V open circuit.  So it should be fine just to pull out the wires for 'maintenance'. (What maintenance that would be for a microinverter?)

Interesting. Microinverters tend to be connected directory to each panel - and the DC connectors (e.g. MC4) tend not to be rated for breaking d.c. currents (even at extra low voltages, d.c. arcs can be hard to extinguish). My PV (string) inverter came with instructions always to operate the a.c. side isolation first - I understand that the theory was that with the inverter shut down no current was drawn through the d.c. side even though the d.c. voltage was still present - making it then safer to operate the d.c. isolator.

So short of covering the panel, I guess the only approach would be to isolate the a.c. side first, then unplug with a d.c. (extra low) voltage present but no d.c. current flowing. Akin to isolation by disconnecting wires rather than operating a device in table 537.4.

   - Andy.

Probably no maintenance for a micro-inverter other than replacing them when they go faulty - I'm not a PV installer so don't know that much about them.

How would you test the circuit, say during an ECIR? To undo terminations would you rely on the PV system disconnecting when the distribution system is not energised? Are the inverters designed to fail safe? On this system there were no easily accessible plugs to disconnect (you would need to get onto the roof and under the panels).

I have seen the the DTI's (BRE) guide "Photovoltaics in Buildings - Guide to the installation of PV systems 2nd Edition" there is a requirement for a second isolator 

"If the a.c. switch-disconnector and the inverter(s) are not in the same room
a local isolator should be installed adjacent to the inverter(s). This is to
facilitate maintenance of the a.c. cable run and inverter(s)."

I think this installation probably meets the requirements of BS7671, but perhaps falls short of industry best practice.

Probably it's not as thorough-looking as the most extreme practice. Not obvious whether that makes a practice with lots of extra contacts a better practice, if isolation wouldn't be needed regularly and there's some way to obtain it when needed. Australia (fond of solar) has realised that required isolators have probably led to more fires than they've done good - that was older designs but the point is often true that more equipment might make more problems as well as more cost.

I replied again because I was looking at a spec for another reason and saw this (in Enphase IQ7 (pdf)).

"Disconnecting means
 The AC and DC connectors have been evaluated and approved by UL for use as the load-break
disconnect required by NEC 690."

That doesn't help your case in the IEC/EN/BS world, but it shows the principle of accepting pluggable connectors as isolation.

Pretty well what the Enphase instructions say for their microinverters for disconnections.

See this page for an example NHS installation done with these microinverters - so its not just domestics.

The picture also show the DC input leads to the micro inverter. These two connect directly into the output leads from the solar panel.

The AC output socket from the microinverter is the one immediately below the DC input for which Enphase supply the plug and a section of AC two core H07RN-F lead for onward connection to your AC side via a weatherproof JB.

www.thesolarpeople.co.uk/.../