Batteries / Off-Grid

Say for example you plan to use a 3600 watt inverter to provide 120V a/c power. The current from your batteries to the inverter with a 12V system would be 3600/12 or 300A. With a 24V system it would be 150A, and with a 48V system it would be 75A etc.

With a 12V system you will need larger more expensive battery cables with higher ampacity and larger more expensive breakers. If on the other hand, you plan to run DC powered appliances direct from the batteries then you will find a far greater selection of 12V appliances than you will at higher voltages.

I tend to go with higher voltages which means lower currents, less losses, and the ability to use smaller and cheaper (also easier to work with) cables and breakers.

12v is generally for short runs, small setups, low total wattage, and/or the direct running of 12v devices.  Once you start looking at running a household on PV, it becomes more prudent to step up to 24v or 48v for the panels and battery bank, and then bump the output to regular 120v/240v through the inverter and use normal (albeit 'energy star' or energy miserly) devices and wiring throughout the house.

As for 96v.... why?  Even if you have 72v panels it is no problem to simply bump them down to charge a 48v battery bank via the controller.  Not much PV equipment is made for 96v setups....as 48v is about the upper limit for most off-the-shelf PV equipment.

Higher DC voltages (typically 80 or more but not limited to that) have a tendency of grabbing a hold of a victim and not letting go of them if they get caught in the flow and, not just pos. to neg. where the victim is the "load." But also if the victim gets in series with neg. to neg. or pos. to pos. in a closed circuit or another way of saying that, while a load is present. Actually, it locks up the victims muscles so they can't let go. At the risk of sounding morbid, have you ever seen what happens to a hotdog overcooked? Just ask Samuel W. Smith. Uuh, have you got a Ouija board? I just want you to know what your working with. The worst thing that can happen is that we get "cavalier" about working with electricity.� 
http://organizedwisdom.com/Electrical_Shock_Deaths
This is only one reason for NEC 690.71b
http://ebsco.smartimagebase.com/generateexhibit.php?ID=719
It happens to somebody everday.
Don't be "somebody."
Be sure all of the tools you use to work on live circuits such as batteries and exposed PV modules and arrays are completely insulated from conducting electricity. Know what all of the electrical potentials are around you at all times. Don't assume a circuit is off, check with an appropriate test meter. Also be aware that flooded cell lead acid batteries can "blow" acid under certain circumstances. Wear the appropriate protective gear when working with batteries.

you bring up a good point.  perhaps i was unclear.  nec 690.7 provides multiplication factors in addition to VOC for poly and monocrystalline modules.  so by taking this factor, the voltage as it pertains to the code and other equipment would be higher.

on the subject of language, you mention 72v panel.  for clarity do you mean "panel" as
a. 1 photovoltaic module
or
b. 2 or more photovoltaic modules wired in series; commonly referred to as a series string, but described in the code as a panel (figure 690.1a)

operating under the assumption of "a", i would assume that the module in your example is a thin film silicon type, or perhaps other type of thin film.  if i had to guess i would say bp millenia.  that being said, if it had a vmp at either 70 or 90v, you are right, even in cold temps, it would likely not make it over the 150v damage threshold of many modern mppt controllers. also, 690.7 would not apply, and you would use the manufacturer's given temp coefficient of voltage to calculate max cold temp VOC.  If  i guessed right and it is the millenia or other thin film module, if you have more than one of those modules, they would all be wired in parallel to prevent the over-voltage (150) that you dont want going to your controller. 

i misused the term derate for 690.7, as it implies lower.  the voltage is not lower, but it corrects voltage higher, thus limiting number of modules per series string in mppt sizing.  that is why i refer to it as derating, lowering the available high end string possibilities.

thanks for participating in the forum, and helping me keep my terminology clear.

james
altE staff