In the case where PV negative is already connected to chassis by some path (e.g. negative through charge controller goes to negative of battery and that is grounded),
If PV positive shorts to frame, the frame would become high voltage relative to earth. If PV frame had separate ground rod, the dirt right at the rod would be same voltage but some distance away would be lower. The resistance of grounding to earth is required to be under 25 ohms for U.S. residential installations, which is not a dead short. If you touch somewhere on the array frame while standing on the dirt, you could get a shock.
This is the reason for a wire back from array frame to chassis of inverter/charge controller.
Many systems have PV floating, but there is internal coupling to AC. A couple people here have reported getting a shock and measuring voltage from PV frame or metal roof on a second building relative to a metal ladder. Wire back to all equipment including inverter and utility entrance panel keeps those at same voltage.
With all metal bonded together, there is low enough resistance to short out the voltage, keep them at the same voltage.
Maybe the would be higher than earth, so a ground rod (that 25 ohms or less) holds the voltage down to same as the dirt you're standing on. No current to speak of, so good enough. One ground rod at service entrance takes care of that.
With AC wires run to two buildings, like my house and garage 100' away, code requires a rod for each. It is also bonded to pipes, foundation steel, etc. I think that is so if I touch chassis at other building, which provides a wire back to ground rod at first building, there could be some voltage drop (current through wires x resistance) so hold dirt at same potential.
Mike Holt doesn't like an additional ground rod because lightning strike nearby would use rod, wire, rod as lower resistance path to spread out across the ground, through your system. It is just the ground wire, but running parallel to PV+/-, it would induce common-mode current in them (inductively). Voltage bounce will also induce (capacitively). But by having not electrocuted yourself you will be able to buy new equipment to replace whatever is damaged.
Probably no lightning arrestor at array. If lightning hits face of panel with PV cells, there will be a hole burned in them. Assume it hits metal frame or structure, follows ground wire and/or footings into earth. For a lightning prone area, consider lightning rods above array, wire draped between them, insulated from array frame and grounded separately.
Back at the inverter or where ever PV wires enter, a lighting arrestor between PV+/PV-, PV+/ground, PV-/ground would clamp the voltages low enough that hopefully electronics doesn't see too much. Some of my PV inverters from SMA have MOV between those wires.
Surge arrestors are used on AC and DC side of PV inverters for protection, to clamp high voltage transients. One inverter I have (actually several) are in the SMA family SUNNY BOY 5000-US / 6000-US / 7000-US / 8000-US. These come with a kit of MOV to protect the DC side: The MOVs are about...
diysolarforum.com
Midnight has a good unit with MOV between the wires, also LED showing their protection is present, or they have blow and need replacement. Various models for different DC and AC voltages (Surges on utility grid are a normal occurrence and you want protection there as well.)
MidNite Solar Category Products.
www.midnitesolar.com
My inverters are spec'd to operate up to 600VDC. The MOV supplied by SMA clamp to about 800 ~ 1000V (higher for high current surge.) That is OK because equipment can take brief transients; Midnight mentions about 2500V spikes as part of compliance testing. I ended up buying a surplus industrial surge suppressor which clamps at 200V, detailed in same thread. That will protect better but trigger more often and wear out sooner.
Lightning is more of a problem in some parts of the country. Utility line surges are normal, from switching of loads and utility company transformers & capacitors. There can also be lightning strikes to power lines.
Don't bother with Delta lightning arrestors. They may be useful for extreme events at higher voltages, but neither Midnight nor I (nor the lab Midnight went to) could get them to fire in the low kV range.
I've been using Delta Lightning Arrestors on my system since they were spec'd by Real Goods as part of the package I bought from them. I now have a surge capacitor on the AC input as well, which is supposed to slow rising edge of voltage so it doesn't get as high before the Silicon Oxide...
diysolarforum.com