Solar and Inverter Systems in Lake Forest, CA

The wrong way to size a solar system is by counting how much roof you have. The right way is to measure daily consumption first, because the array only has to replace what leaves the bank between sunrise and sunset. A shunt monitor installed for a week produces that number, and it is nearly always different from what the owner estimated.

Once consumption is known, the array is sized against realistic production rather than nameplate watts. A flat mounted panel in Orange County produces well in summer and considerably less in December, and it produces nothing at all under the part of the roof shaded by an air conditioner shroud or a satellite dome. Shading matters more than most owners expect because on a series string one shaded panel drags the string down.

The layout that comes out of this is a compromise between panel count, series and parallel arrangement, and the fixed obstacles already on the roof. We lay it out on the actual roof with the vents, fans, antennas and air conditioners in place rather than on a rectangle drawn on paper.

A pulse width modulated controller pulls the panel down to battery voltage, which throws away a meaningful share of what the panel can produce. A maximum power point tracking controller converts the panel's higher voltage down to charge voltage and keeps the extra as current. On a cold clear morning that difference can be twenty to thirty percent, and it is the difference between the bank finishing and not finishing.

Controller selection also has to match the array wiring. A tracking controller has an input voltage window, and a string wired for a higher voltage than the controller accepts will either shut it down or damage it. Wiring for higher voltage and lower current is usually the better choice because it allows smaller conductors on a long run from the roof to the bay.

Programming is the step that gets skipped. The controller needs the correct chemistry, absorption voltage, absorption time and float setting for the bank it is charging, plus a temperature sensor where the bank location runs hot. Units from the mainstream manufacturers all support this and none of them do it by themselves.

A modified sine inverter is inexpensive and it will run a resistive load. It will also make an induction motor run warm, upset a residential refrigerator's control board, and put an audible buzz through anything with a transformer. Pure sine output from a Magnum, Xantrex or Victron class unit is what a modern coach with electronics and a residential refrigerator needs.

Sizing follows the largest load you intend to run and the surge that load demands on startup. A residential refrigerator compressor, a microwave and a rooftop air conditioner all present a surge several times their running draw. An inverter that is comfortable on running load and marginal on surge is the one that clicks off when the microwave starts.

Integration is the part that separates a good install from a frustrating one. An inverter charger with an internal transfer relay picks up the load when shore power drops, and the sub panel behind it decides which circuits stay live. Deciding which outlets and which appliances live on the inverter side, and labeling them, is a design conversation we have before the wiring goes in rather than after.

The high current side of an inverter installation is where corners get cut and where the consequences are worst. A three thousand watt inverter can pull upward of two hundred and fifty amps from a twelve volt bank, and that current has to travel on cable sized for the distance, not just for the amperage.

We size conductors for an acceptable voltage drop over the actual measured run length, then fuse at the battery end of every run because the purpose of that fuse is to protect the cable, not the appliance. Class T fusing is used where the bank can deliver the fault current that demands it, particularly on lithium installs.

Routing gets the same attention as sizing. Positive cable crossing a frame member gets a grommet and a strain relief. Runs are separated from data and antenna cable to keep switching noise out of the audio and television systems. Terminals are hydraulic crimped, sealed and labeled so the next person to open the bay can read the system instead of tracing it.

Every panel bracket is a hole in the roof. On a coach that hole goes through a membrane, a layer of decking and often a foam core, and how it is sealed decides whether the solar install becomes a water damage job in three years.

Mounts are set on a bed of the correct sealant for the membrane material, fastened into structure rather than into thin decking where the layout allows it, and then lap sealed over the fastener heads. On a coach where the roof cannot take fasteners in the required location, adhesive mounting systems are used instead of forcing a mechanical fixing into decking that cannot hold it.

The roof is also where wire enters the coach, and the entry gland is the most reliably leaky part of a poorly done install. We use a proper cable gland set in sealant with the wire looped below the entry so water follows the drip loop instead of the conductor. This is the same standard applied to every other roof penetration on the vehicle.

A large share of the solar work here is not new installation. It is correcting a system that was installed by a previous owner or a low bidder and never worked properly. The failures repeat: undersized conductors from roof to controller, a pulse width controller feeding a lithium bank, no fusing at the array, and mounts that were sealed with the wrong product.

Diagnosis is straightforward with a clamp meter and an irradiance sensible day. We measure open circuit voltage at the roof, again at the controller input, then output current at the battery. Losses show up between those readings and point directly at the run, the connections or the controller.

Panel failures themselves do occur, usually as a failed bypass diode in a junction box or as delamination on an older module. Individual panel testing under load identifies which module in a string is dragging the rest, which is far cheaper than replacing an entire array on suspicion.