The Kyocera instruction manual suggests wire gauge from 14 to 10. I believe they limit it to AWG 10 because you just can't get a larger size to wrap around the #10 terminal screws in the connection box. In fact, its a real chore to fit AWG 10 wire on those screws.
The total length of DC wiring in my 5.4 kW system is 797 ft. The string loops range from 40 ft. to 51 ft. in round-trip length. During much of each sunny day the current in this wire will be about 6 amperes. The I2R copper loss on a cold day would be at least 73 watts for #14, 29 watts for #10, and only 13 watts for #6. I chose #6 in spite of the Kyocera recommendation.
My inverters are located very close to the PV panels that they serve. I suspect the DC copper losses for roof-mounted arrays might become pretty significant.
Important wire parameters for PV applications are:
And, in my specific case, flexibility because of moveable PV panels. My research revealed the following possibilities:
resistant Wet location Nylon
Heat resistant Wet
But its 2,000-volt insulation made it a bit tight through the sealing bushings. I found a 1,000-volt version but the chief county inspector told me he would not approve it because the UL ratings were not printed on the jacket. This seemed to be a big deal for inspectors.
Zero Halogen Cable
PostScript: This is the wire I used. As it turned out, when my system was inspected (by both county and SDG&E inspectors) no one even looked at the wire.
Of course #6 wire must be fitted with terminals for connection to the #10 screws in the PV junction box. It is important to know that Class 1 stranded wire requires different terminals than ordinary stranded wire. Here is the UL certified combination of crimper and terminals I used - not your average Home Depot variety.