Sorry for asking again but I can't find a wire size calculator I like. I have approximately 100' between my array and my classic. I have already buried the 1.25" conduit and ran #4 wire through it. This was fine when I was planning to run 115vdc 10.12a through it to my classic ... in fact it was probably overkill but I don't want stuff to over heat and my voltage loss was like .30%. After reading this forum for months now, it would appear that I want to bring my array voltage down closer to my battery bank voltage which is 24v. I can rewire my array to give me a total VOC of 66.75v, VMP of 53.40v and a rated array amperage of 20.24. Can my #4 wiring safely handle this or should I take the system voltage back up and run the classic less efficiently?
Thanks for any advice. This forum is hugely helpful.
As far as safety- there's no issue all. Even 12 awg wire can handle 20amps.
The issue is voltage drop. Using 53v and 20a, the Southwire voltage drop calculator says 4 awg copper will give you a voltage drop of 2.17% - so you should be ok.
My previous answer on this was its a little hard to say, because full efficiency curves for these things dont exist. In the industry the rule of thumb has been something to the effect of: you lose an extra 2% in the controller per extra "nominal voltage".
However your question made me restudy the data here:
http://www.midnitesolar.com/images/classicFrontPage/graphs.php
Analysed for 24v batt, you get the figures in the attached image below.
If thats right this is interesting, you lose 4.2 % for every 10 extra PV volts. Less for the first and last extar 10v, worst in the mid zone. Midnite can maybe clarify whether ive abused that source graph beyond what it was supposed to mean.
Its a bit more than expected... Makes wire losses look pretty trivial really.
Anyway so you do a rough calculation based on the above. Work out the cable losses for the 2 panels per string, then compare that with cable losses plus expected controller loss for 3 panels per string . Work in watts lost, so you can sum them.
Its a project for someone to do some real efficiency testing on these devices.