Ok.. too busy with a bazillion things to get to this sooner but here goes..
The step-by-step pictures are here
new ground wire.. bypassing the sliding contact.. Important note.. the little circular ring.. now i actually make a whole loop back to itself and solder so there are doubled-up wires coming off.. 1/2 the resistance and double the strength.. it is absolutely necessary for that to be soldered back to itself so it does not separate when tightening the screw.. or the wheel will no longer roll.
I recently measured an old magswitch to have 18 to 100 ohms.. not miliohms.. OHMS!.. and as a testimony to Pro-Gold.. i put in 2 drops and worked it in and got it down to 6 or 8 miliohms in 20 seconds! nonetheless.. since it is the main part that fails.. i shorted it out with a wire!.. i actually cut out the slider so it can't short out with the new insides.
(resistance is now 1.43 miliohm)... typically it is 7 to 15.. not terrible actually... just when it gets dirty it's the biggest problem.
hardwire the switch.. drops about 10-20miliohms.
solder wick through the spring.
This is by far the biggest improvement.. typically that will measure on the order of 100-120 miliohms.. in-fact.. the 160+miliohm resistance will drop to 80 (half) by ONLY soldering the ends of the spring.. and still using the spring vs the solderwick.
this solution shown here drops the resistance of the spring to about 3-4miliohm and the whole ckt from the pos. terminal on the PR adapter in this case to about 12 miliohm.
I made a small change since this picture was taken.. i no longer bother to attach the wire to the bottom.. it comes out a hole on the side at the bottom and is attached to a wire that goes directly to the top contact on the switch.
Unfortunately i don't have any pictures yet of the battery spring.. but the wire goes through a hole drilled straight through the plastic to the bottom.. a notch is cut in the inner circle and the last 1/2 inch of wire is stripped and the end soldered to the edge of the top of the spring (so that it doesn't interfere with the spring's ability to be pressed all the way down).
The end result.. the switch i started with measured 500mV drop at 3A.. or 167mohm... after i was done with the modification.. it measured 27.3 mohm.
That puts it in league with the KIU socket kit. most of that resistance is where the spring contact touches the bottom of the PR bulb or adapter.. tough to get around that one.. but even the KIU socket i measured 7mohm between both contacts... this pr adapter to bi-pin i measured 3mohm per contact.. on-par with the KIU.. so since there is probably about 1-2mohm of wire, and 12mohm in the wire and contact.. that means 12-2-3 = 7mohm contact on the bottom of the bulb adapter.. that's as good as the best i've gotten a magswitch with progold (just the contact of the switch).. each contact outside the switch is another full 'contact' which will be about another 7-10mohm typically... and not trustworthy.
So.. the final results..
167mohm turns to 28.. 1/6th the resistance!
so a couple examples of 'the difference' using the fixed switch or stock:
1160 on 6V bat.. 5.39 to bulb.. 424/275 L > 5.89V 577/375
1111 on 7.2V bat.. 6.6 to bulb.. 649/722 L > 7.09V 834/542
1331 on 10.8V bat.. 10.46 to bulb.. 720/468 L > 10.74V 790/514
1185 on 10.8V bat.. 10.25 to bulb.. 1026/667 L > 10.7V 1194/776
and my favorite comparison.. a lesson in power transmission.. why does the power company use super high voltage to transmit power.. here is why:
1323 on 14.4V bat.. 14.21V to bulb.. 414/269 L > 14.37V 430/280
(that is 15.6 vs 15.9W)
1183 on 4.8V bat.. 4.19V to bulb.. 274/178 L > 4.68V 406/264
(that is 15.1 vs 18.0).
so.. barely a difference in output with the high voltage low current solution... but in the low voltage solution... there is a 40% drop in output with the stock switch!
Well there you have it.. look at the bulb options and see if you are in there.. think about it.. it is CURRENT that causes the voltage drop.. so if you have a bulb with low current you'll be less affected.. this means that the 1060 and the 1185 and the 1183 will be most affected..
You may have noticed the omission of the 1166.. it's because it's a sucky solution for direct-drive.. that bulb is underdriven a lot from a 10.8V solution.. enough that there is only a 30 L swing in the output... now that is absolutely a fantastic lamp when pushed properly to 12.5V or so.. 900/600L.. My little 1D FM light is outputting 900/600L for 20 minutes on a charge .. and it takes a cent to charge.. probably not even.. so 3 cents/hr compared to the $8 to $100/hr with the M6.
I hope this helps people.. it is not a terribly difficult mod to undertake.. you just need solderwick, some regular wire (extra flexible type is good.. i used 14ga speaker wire.. and some 18ga wire (which is fine.. 18ga is 1 miliohm per two inches!).. you need a drill, some cutters and of course a 5/64th screw.
Oh.. very helpful tip.. to solder to the cups.. it's pretty tricky usually.. it can be done with a higher watt solder iron or gun.. but by far the easiest is to hold the cup concave side up... in the likes of the holder shown.. and heat the bottom with a micro blowtorch.. puddle some solder into the cup (careful.. you'll likely start some rosin on fire).. and then touch the solder wick into the puddle of solder.. then put down the flame and use a heat-sink to cool the cup.. else it takes like 10-15 seconds for the solder to solidify.
if you only solder across the spring and do no other mod.. well .. put some contact cleaner on the slide while it's apart.. you will drop the resistance to about 60mohm.. and that will have this level of result:
1111 on 7.2V.. 6.6V to bulb 649/422 L > 6.98V 790/514.. comparing to: 7.09V 834/542 if you 'go all the way'
So.. just fixing the spring will gain you 141L and shorting out all the contacts also will get you 185L gain.. however..that means a 22% vs a 29% gain, so balance how much it's worth in deciding what to fix.
order of importance:
spring spring spring.. over half the resistance
sliding contact... good initial resistance.. TERRIBLE long-term.
outer switch contacts.
There is a weird thing about how the magswitch is designed.. it is specifically designed to be cheap to assemble and reliable (i'm talking about the electrical switch not the whole assembly).. but at a serious penalty to the electrical resistance.. here is why...
as anybody that knows ohms law knows.. series resistances add up...
ANY and EVERY time.. you have a 'contact' you are going to have SOME resistance.. and typically even the very best contacts are between 5 and 10 miliohms.. The problem with the design of the switch is this..
Rather than having a hard-wired single SPST switch which could be 5 or 8 miliohms... There are FOUR series switches in a mag switch..
(1) contact from + spring pad to switch lower contact
(2) switch lower contact to connecting donut
(3) connecting donut to upper contact
(4) upper contact to PR holder spring contact.
2 and 3 are exceptionally low resistance.. 4mohm if you treat them.. more like 6 if you don't.. but 1 and 4 are iffy.. 10 to 15mohm is not unlikely..
that means.. rather than 5 or 8 mohm.. the mag switch stock is more likely 30mohm to 40 mohm.. easy enough to fix with hard-wiring the troublesome contacts and get it down to 10 or so.
Hope people can follow and it helps!