I read the manual for the Fronious TL 3.6. Key points are that it is transformer-less so there is no galvanic isolation between the ac and dc sides; it incorporates RCMU - residual current monitoring unit designed to disconnected from the mains in the event of ac or dc earth leakage; if an rcd is used then it is to be a Type A: this is taken from the manual page 13:
The inverter is equipped with an RCMU according to DIN VDE 0126-1-1. It monitors residual currents from the solar module to the inverter grid connection and disconnects the inverter from the grid when an improper residual current is detected. Additional residual current protection may be needed depending on the installation's protection system or the requirements of the utility company. In this case, use a type A residual current circuit breaker with a release current of at least 100 mA.
There is no specific guidance on earthing and bonding other than to connect the cpc to the inverter.
My opinion is that with such earth leakage detection equipment and no galvanic isolation (ie: transformerless) then the dc array + and - should be left floating and the pv array metalwork should be bonded to the MET. The dc cables should be insulated and sheathed or insulated and then run in conduit to provide the equivalent of double insulation protection. And this would apply for TN-S, TNC-S and TT installations. But this is only my opinion and not something authorative to refer to.
A bit more delving and I turned up this DTI sponsored study in 2002 ( so later material could be available) on the subject of transformerless PV inverter system, earthing, bonding and earthing systems which is worth a slow read as I have done:
https://webarchive.nationalarchives.gov.uk/20100104194450/http://www.berr.gov.uk/files/file16525.pdf
It chimes with my opinion but includes a proviso to check first with the DNO about the connection of the PV array metalwork to the MET for PME systems so presumably TNC-S too. There is a very clear and helpful earthing decision tree on page 42. This study informed some DTI guidance to installers which so far I have not managed to find.
Once can though end up in a debate about 'exporting PME' depending on whether the roof array is considered to be within or without the equipotential zone. This is a question well covered in the EF and others can provide up-to-date advice to you; I seem to remember one cannot export PME unless an conductor equal to or greater than the incoming line conductor is employed - but don't rely on me for the correct details. I hope others will chip in on the rules and detail.
If I turn up anything else I will report it but this early reply hopefully is of some use to you. I am not familiar with what the IET regulations might say but presumably you have a copy and have looked/could look? Anyway some pointers which may lead to a definitive, authoritative reference.
[automerge]1588668524[/automerge]
Just found this later document (2012) by ECA and MCS:
See 2.2
Design Part 2 – Earthing, Protective Equipotential Bonding and Lightning Protection on page 38 and a similar earthing flow chart Fig 10 on page 40. Para 2.2.4 covers the question of whether the array is an extraneous conductive part and the matter of PME etcetera.
It seems to me that an informed judgement has to be made about whether a roof top array can be considered an extraneous conductive part and this is guided by some resistance testing between the array frame and MET; less than 22kOhms it is an ECP and greater than 22kOhms it is not. I wonder what you do if the R is say 25K?
I reckon I would bond the array back to MET after asking the DNO first for advice in the PME/TNCS setting.
[automerge]1588669233[/automerge]
( Press post too quickly)
I reckon I would bond the array back to MET after asking the DNO first for advice in the PME/TNCS setting. That way if someone came along to wash the array with water from the premise's cold/hot water system I'd be sure the array and water were at equipotential.