I think you would need a Franklin lightning rod installed some distance above the light and the normal large down conductors bonded to the tower steel,terminating in a ground grid system around the tower.
You probably also need surge protection on the power circuit up to the light.
I wouldn't worry so much about the lamp being struck directly, but if lightning strikes the tower, the tower top potential will be very high with respect to the potential of the lamp supply conductor. A bit more than 600 volt insulation would withstand.
Maybe surge arresters at the top and bottom of the tower? An arrester at the top only would just shunt some of the surge to the supply conductor and cause a problem at the source.
At a minimum, the conduit for the supply conductor should be bonded to the tower at the top and at the bottom.
When we design high mast lighting systems wether for freeway interchanges, railroad of intermodal yards, sports stadiums, etc..., or just large parking lots, and the isoceraunic index and other risk factors indicate we need a lightning protection system, we generally concentrate on the protection of the lighting towers themselves. The aircraft warning lights, and other ancillary devices are all bonded to the towers and are basically part of the tower structure. The important issue is the downleads from the lightning rods for the tower itself, and assuring that there is path to ground for the high energy levels anticipated from a strike. Keeping the downleads isolated from the tower structure and electrical devices will maximize the amount of energy directed directly to earth and hopefully minimize damage to the facility. A good source of information is Thompson Lightning Protection Inc. at
EEJamie-
NFPA 780 requires downleads to be bonded to vertical metallic structures upon which they are routed at the top, bottom and at 200ft intervals, to reduce possibility of side-flashes. Do you consider your philosophy of insulating the downlead as contrary to NFPA780?
In 15 years of servicing (mostly 500') communications towers,
I have never seen a top light take a hit. But bear in mind
that, almost without exception, these towers had at least one
(and often 3) 10' to 20' solid metallic antenna top-mounted.
I seemed to be more important to make the structure a good
conductor from top to bottom (either by bonding every joint
or by running a solid wire from top to bottom, and not
relying on the cable shield bonding, although that did add to
the total path), so the tower itself attracted the streamer.
I suppose that due to the higher impedance of the relatively
small wire to the lamps, the lighting decided that it was
just too much trouble to mess with them.
Not to say it could never happen, but I would put it
farther down the list of attractors. I don't think
anyone can predict with certainty exactly where the
strikes will occur, unless there is some obviously
overbearing stable, grounded structure. One or two feet
up/down never seemed to make a difference, regardless.
Nor, surprisingly, did corrosion at the bonding straps
(except at the bottom). Apparently the leg joints made
up good enough, or the bolt mating surfaces stayed clean.
