I agree something is wrong with the design and certainly the relay kits do fix the problem (with other benefits). I am planning on servicing my starter switch anyway and would rather insert a $6 snubber than a $70 relay setup.
The labor cost (or value) swamps the parts cost. The 2.4 Ohm resistor
and 1 uF capacitor
cost 65 cents per snubber.
I don't know if it makes a difference but when you turn on the ignition the headlights turn on so they aren't cold when the headlight cut out turns them off then back on in about ~1sec or so. I believe that the cold filament inrush current is what typically damages the headlight contacts in the ignition switch. My plan to protect the ignition switch is to turn on the ignition with the starter button pushed in and the clutch cutout switch preventing it from turning over (certain models only, credit to GW).
I think you overestimate the degree to which the filaments stay hot during a start. A good measure of how quickly they cool is how quickly they become non-light-emitting. That looks to be below a hundred mS to me.
The clutch cutout switch does not carry headlight current in any schematic I have seen. It is in series with the coil of the starter relay.
I don't think it is the inductive energy in the wiring that is damaging the contacts. I think the extra unaccounted for (in the design) energy is the generator start up transient that the R/R cannot dampen completely. This is my only extra assumption but I think it is real.
The only energy worthy of concern for switch contacts is that which exists because current is flowing and which must vanish when the current ceases. The battery buffers anything like that which the generator exhibits. The battery is highly capacitive, and there are other, more explicit capacitors on the 12V bus, at the computer at least. So the only inductance I see needing some snubbing is in the wiring harness.
Cutting off inductive loads means the energy of the inductor is dumped across the contacts causing damage. The snubber just gives that energy a place to go besides the contacts. This is exactly analogous to the transient energy of a starting generator dumping across closing contacts. In both cases a snubber would give that energy a place to go instead of across the contacts, I just need to know what values of R & C to use.
I agree with that first sentence, provided that "loads" includes all inductance in series with the switch. Technically, the inductive energy is first going into the arc between the contacts. It only gets into the contacts as heat or metal ion motion. Nevertheless, the snubber provides an alternative destination for most of that energy.
When the contacts close and stay closed, there is no mechanism for damaging them other than excessively high current flowing through too little contact area. This is because the voltage across the contacts is low when they are closed, and so the power dissipated there (which is the product of voltage and current) is necessarily low.
It is contact bounce, where there is not simply a simple closing but rather a closing followed a few mS later by a brief opening, which can damage contacts. It is a rare mechanical switch that does not have 1 or 2 bounces upon closing. And it is at that event that the 10-15 X inrush current is flowing, which translates to 100-225 X as much inductively stored energy which will go into an arc if not a snubber.
Trepidator, I really appreciate your comments and expertise. Please feel free to shoot a hole in my theory so I can move on to other ideas. Thanks.
Perhaps this will count as partial penance for my contribution to the mercifully killed thread on climatology.