Not All Switch Are Created Equal!

[andrewwynn]

i musta pulled 30% outtamabutt.. sometimes best to use a calculator.. i got 20% when i used a calculator which is 'on par' with the 19% you said.. 1000 lumen vs 1200 lumen... i think i must have done the math on one of the 6V bulbs i was using that also used 3.5A.. or even the one that had 5.7A.. in any event.. some were huuged differences.

Wow. that was a swing of numbers. I did my test and measured a 0.16ohm but i was measuring the entire path from the + spring to the - screw tab and shorting out the pr with a piece of metal... that was at 3A (max one side of my power supply will pump out).. the only way to 'really know' is to wire up a tiny test lead to the pins on the bulb and the battery to measure Vbat and Vbulb ... i'm going to be cutting a maglight in half the long way for testing purposes which will make this kind of testing oh so much easier.

-awr

 

[AuroraLite]

Modamag,

Thank you for a very useful set of data, this is a very important piece of information as we are constructing more and more powerful Mag builds. :thumbsup:

 

[andrewwynn]

Ok.. been dying to do an update to my tests...

All tests performed using a bench power supply hooked up directly to the magswitch... using the leads for voltage drop.. typical voltage applied about 1V.. and set to constant current of 3.0A.

Phase 1: stock switch 'reference'.. dropping about 500mV across entire switch.. closer to 250 mv from bat. terminal to spring inside PR base.

that means 167mohm on the entire switch and 83 just to the spring.

my suspicion was that a great amount of the resistance was between the spring contacts and the switch and also between the spring contacts and the spring..

phase 2.. check the switch alone..

3A across the switch dropped almost exactly 33mV = 11mohm... I put pro gold on the contact to see if that would help.. contact looks like it is a silver plated copper ring. (very well made contact actually).. but at best the pro gold dropped off 1mohm and i measured about 30mV drop afterwards... no improvment possible on the switch... it's safe to say 10mohm switch.

phase 3: progold the spring contacts:

oh here is where it gets a little interesting:

progold *everything in sight*. (switch to bottom contact, contact to spring spring to upper contact and on top of the upper contact).

Ok, now it measured about 320 mV drop on the whole switch mechanism.. 106mohm.. a drop of about 60-70 miliohm.. but i still measured a pretty good drop on the spring. so..

phase 4.. fix the damn spring.

I took the spring and two contacts.. i soldered some 18ga wire (about 3 inches curled up like a spring).. between the top and bottom contacts.. 18 ga wire is 1/2miliohm per inch so that is negligible in this test.. 1 1/2 miliohm...

re-assembled and measure:

150 milivolts! holy cow.. down from initially over 500. now my magswitch with PR base is measuring a TOTAL of 50 miliohms! that's bordering on an FET switch.

The majority of the losses were in the spring-to-contact joint.. but of course i still had progold on the spring-contact to switch connection.

In any event.. example.. 1185 solution... 3.35A.. that means a .559V drop that means 10.241V to the bulb at 10.8v battery... or 1024 lumen... by soldering the spring out of the equation... the Vbulb will go up to 10.633v with everything else constant.. that means you get 1168 lumen at the bulb! 14% greater output with no more draw from the battery! just putting the light out the bulb vs heating your switch!

Now.. if you already have the KIU bi-pin bulb solution you are circumventing the majority of the problem.. i didn't do this acid test on that switch yet but i will soon..

In case you are wondering.. the sliding contact as long as it's clean is pheominally low resistance.. i only measured 10-20 milivolts drop on that contact depending on how far in the PR bulb was pushed.. that's 3.3 to 6.6 mohm.

So.. 10 miliohm for the switch.. that's only a 30mV drop (10.8V ->10.77V)..

Summary:

stock switch.. 167mohm (.559V drop to an 1185 bulb)
cleaned up.. 106 mohm (.355V drop)
soldered spring.. 50 mohm (.168V drop).

so.. by just soldering the spring and putting a bit of progold on the remaining contacts you can boost the output 14% on a pr-based mag85..

my estimates are that with the bi-bin bulb the entire switch will measure less than 20mohm.. so.. at 10.8V battery you get an add'l 1/10th of a volt to the bulb and that puts you over 1,200 bulb lumen! no wonder we like our bi-pin bulb solutions!

next step.. FET... the FET i picked out has 4 miliohm resistance.. the wiring ... 18ga wire is only 1/2 a miliohm per inch.. figure 4 total inches.. so 2 more.. and the total switch will be 6 miliohm... that makes only a small improvement.. up to 1225.. but the bigger improvement will be the regulation holding the voltage constant at about 10.7V on the bulb so there is no concern about insta-flashing.

so.. 1024 up to 1207 = 18% increase to take a magswitch and upgrade to a bi-pin.. and it puts it within 2% of the output of going 'maximum drive' with the best FET!

========

So.. the regular old mag switch is pretty solid performer actually.. it's the PR base causing the problems in high-drain situations.. 10miliohm is pretty darn good.. a lot of the FETs i looked at were 26 miliohm.. whoops!

the improvement of using FET switch is two fold primarily.. 1) no contact to EVER get dirty.. 2) you can very easily incorporate a 'soft start' by putting a capacitor on the gate of the FET.

we'll i'll be... i had thought the switch was worse than it is... those hard-wired bi-pin solutions are close to FET performance, fantastic.. and a simple kludge on the switch with a spring shorting wire really greatly improved the PR-based design!

Thanks again modamag for bringing this to our attention.. i'm going to put that 'hack' switch i just fixed into one of my PR hotwire lamps right now and measure the actual before and after lux output.

-awr

 

[wquiles]

Andrew,

Can you post some pictures showing exactly how/where you soldered the wires to minimize the resistance?

Will

 

[andrewwynn]

i think i can try to get a pic.. i want to swap that 'tester' into a 'working' light tonight! I can describe it easy enough though.. the spring inside has two 'caps'.. and i soldered the wire to the inside of both caps right in the middle of where the spring is... just heated the metal stinkin' hot from the outside with the cap like a 'cup'.. melted a puddle of solder inside and stuck a pre-tinned stranded 18ga wire into the solder puddle.. (make sure you don't wick solder too far up the wire or it becomes 'solid' not stranded).. Since the wire was 'curled' a bit to be like a spring i could pull it out past the spring (through the middle of the spring) to repeat the procedure on the other side.. with progold and the spring fix I was able to drop the resistance of the entire mag PR switch from 167 to 50 miliohms.. that's pretty stunning... that will turn a 7.2V solution with an 1111 from 584 to 729 lumen (kid you not). (if using a PR base).

These tests were done with a real PR bulb in the prbase just shorted out with a strip of metal from a battery pack going from the bulb bottom to the edge around the flange of the bulb, so very accurate 'real world' testing.

I just realized a possible 'hack' on the switch itself that could half the contact resistance.. the only reason the main switch is 10miliohm is that it's TWO switches in series with each other... if there is enough room to solder the moving contact ring to a moving wire instead of the contact.. maybe even solder that directly to the battery post it would be conceivable to drop the switch resistance to 5miliohm.

again.. stay tuned... one of the designs i have connects a flexible stranded wire to the moving PR base (ground side) vs using the sliding contact.. that would drop that resistance there as well (though it's only 6-10miliohm... a 2" wire is 1miliohm even if 18ga!).. think of how a DC motor brush has a really flexible lead wire.. that's the concept i'm talking about.

-awr

 

[andrewwynn]

update.. swapping the switch i fixed with stock.. I measured 21,300 before and 23,200 lux after.. a full 9 percent gain JUST from soldering the spring and putting pro gold on the remaining contacts.

another update: we tried this 'trick' with just soldering the spring but keeping the spring and not using the copper wire. ended up with 80 miliohms resistance.. about 1/2 but still that's pretty amazing.. looks like 30miliohms resistance in that spring that's not cool what was mag thinking?

We have a really neat update planned.. i'll take pictures... but here is the plan:

1) remove the lower switch contact and solder 'solder wick' directly to the switch 'ring'... this should eliminate 10-15 miliohms resistance...

2) remove and eliminate the bottom spring contact cup completely... solder solder wick directly to the top switch contact feeting it up through the pr assembly... this will get fed through the spring and soldered to the top pr contact... this should eliminate at least 10 more miliohms from my 50mohm solution.

3) connect a ground return wire from the moving PR base where the screw and feed it back to the ground screw.. this will eliminate about 10 more miliohms.. but more importantly.. eliminates the main cause of variable resistance over time... that slidging contact can be hundreds of miliohms easily.

The fixes mentioned should take the 50milohm (60 total).. switch down to about... 5+2+2+10 = 19 miliohms (29 total).. the 10 extra is the contacts to the battery.. and the last '10' is contacts to the PR bulb itself.

at 3.35A... 60mohm = .2V.. makes for 1155 lumen.. with 29mohm.. the voltage drop is 0.1V makes for 1195 lumen..

now.. consider an very good FET solution.. no more switches.. coupled with bipin socket...

the FET i picked out is 4miliohm.. add 3 miliohm for wires, and 8 for the socket.. 15 miliohm.. ten more for the battery contacts =25.. so as you can see.. virtually the same as a good clean-up of the stock switch..

lumen calculates to 1201.. impercevably different.

the main difference of course with putting in an FET.. holding back over-voltage with hot battery packs.. means you can use bulbs that you'd not be able to... or get longer bulb life with the more robust bulbs as well.

Oh.. some more intersting switch fixes.. if you have PR bulbs.. and the 1D mag, or a 2D mag and run 1111 bulb.. check these out:

1183 bulb stock switch: 271 lumen
1183 bulb.. braid both hot and ground, stock switch innards.. 401 lumen!

1111 bulb with 7.2V solution:

stock... 584 lumen.. 'patched' as above.. 749 lumen..KIU bipin.. 763..

as you can see.. you can fix a PR base to be on-par with the bipin.. if you know how to dremel and solder fix it up.. you'll be mind-blown by the difference fixing the switch makes.

oh.. i'll toss in a fun 'gimme'.. if you have the 1D FiveMega light (4AA).. putting 4 LION 14500 in series and using the 1323 bulb (12V bulb).. will get you 414 lumen with the STOCK switch... 431 with KIU socket.

(the reason for very little change.. with 3x the voltage.. needs 1/3 the current.. voltage drop is current x resistance.. 1/3 the current = 1/3 the voltage drop.. exactly why the power company uses 10s or 100s of thousands of volts to transfer power).

We've been trying to find very high voltage bulbs to solve this resistance issue, but not much luck, so really optimize the lowest resistance.

Oh.. another consideration.. LIon vs NiMH... 1/3 the cells.. 1/3 the connections..

if you count up 5 miliohms per contact.. 9AAs = 50miliohms! (.168V)... now.. switch to 3xLiON and the contact resistance drops to 20miliohm.. (.07V)... 3 cheers for LiON... fortunately.. the apx. .1V lost from LiON batteries is already 'factored in' in most of my calculations... i always use 3.6V for my LiON calculations per cell.. and apparently they are closer to 3.7 at the cell because i measure about 3.6V/cell even with heavy load on the battery pack as a whole.. and i've measured 1.20V/cell on 6 packs of NiMH even at like 5+A.. so they must be a little higher than 1.2 at the cell as well.. this works very well for doing calculations... however my battery packs are very carfully made ... if you double the reistance to 10miliohms per battery contact than you have 100mohm in a 9-pack.. which is ouch.. .335mV! get the progold on those contacts!

ok bedtime.. hope this helps somebody out there.. I'll try to get pictures of the PR low-resistance mod.

-awr

 

[AtomicX]

Hi Guys... was researching threads to try and find my M@g85 (dim) problem and found this...I think my switch is stopping the Lumens dead. anybody know if a replacement switch for a MagCarger (bi-pin) will fit proper as a replacement in a standard 3D host? is it a direct swap-in? I'd be willing to fork over 23 bucks for one if it is give away the PR potted bulb I bought gladly for that matter. Anybody use this solution in a 3d Mag85 setup? Any help much appreciated :wave:

 

[andrewwynn]

from what i've heard there is a bi-pin replacment from the MC but i don't know that it fixes the problem.. if it uses the same spring and slide system as the standard you won't gain benefit of lower resistance.

I have a DIY solution to the magswitch that drops the resistance to 1/6th stock:

http://prfix.rouse.com check out that thread. ONLY doing the spring fix will 1/2 the resistance..

-awr

 

[AtomicX]

Thanks Andrewwynn, I had found that thread and already have done the spring fix, found it tough to solder to the plated material so on the large cap end I wound it between the spring wires right at the end, still in contact with the caps on both ends, just that one end is soldered and one end not. I may take another shot at it with wider solder wick than the thin stuff I had on hand. I will dremel the plating slightly this time and maybe the solder will adhere better. I also want to "finish" the whole switch mod while I'm there. I think I'll grab a replacement while I'm out because my switch is old and .....well, better yet I'll grab a different host. Everything will be new, the tailcap threads on mine are beat, and have been cleaned up with a rotary tool after a batteryu leaked years ago. This light is old. about 1988 if I remember correctly. Maybe I should spend 15 dollars and just get a new one. Do I risk "instaflash" more after the resistance is gone (cut down) ??? and one more thing while I'm here, is there anything like Progold perhaps I can use to substitute...even if it's not as good, but helps and is less costly ?? thanks in advance Andrew or anyone that knows...

 

[andrewwynn]

well the soldering to the end cap is fairly important.. that is apparently a really lousy connection, but the spring is about 1/3 of the resistance of the whole switch so having the copper just pressed in is good.

The trick to soldering to the cups is holding them in the likes of a '3rd hand' with the cup able to be heated from the outside (i use a tiny torch).. and make a puddle of solder inside the cup.. and dip the solder wick into that puddle.

just about anything that is designed to help contacts will work.. including most likely WD-40, etc.. i don't know if the likes of RS have any contact enhancer that is sold in small quantities.

re: instaflash.. depends on the batteries and bulb.. of course with lower resistance.. if there was a chance before it's greater after the fix... certain combos.. especially with bigger cells i'd be worried..

but if it's a 9AA 3D with the 1185 for example.. rested cells are rare to blow that lamp.. but i wouldn't do it with 9x17500 LiON. The 1111 from 2x or 2x3 LiON i have very good luck with.

also. with my LiON17500x9 solution.. that works very well with the pr fix and the 1166.. (which is way under-powered normally from 3xLiON).

-awr

 

[AtomicX]

I will swap hosts sometime tomorrow, was out holiday shopping and didnt get to do anything today. Time to sleep now, job interview tomorrow. I'll post after the other host's switch is done late tomorrow, I can get my dad's mini torch since I dont have one and I do have a third hand, will stop at RS while I'm out for contact enhancer...hopefully. Thanks for info, I'll do up the other host instead of this old light and give the RS XPR103 / 5 C Nimh setup to this "older" light for my house user and steal the newer unit to build the 85. Should have done that to begin with. Time to sleep.

 

[andrewwynn]

i think i have a picture in the prfix page http://prfix.rouse.com that shows how i'm holding the metal cup in the third hand.. it coincidentally has the spring in it but shows how i have the metal cup in.. it's easies to do the bigger cup first so hopefully the one you have left to do is the small one, i may have mentioned that in the descriptions.

in my first pr fix test i put 18ga copper wire vs solder wick through the spring and other than that only used pro gold on all contacts including inside the switch itself and brought the resistance down to 50 mohm (from about 170).. with the full-boat solution it only goes down about another 20 mohm to 30, which is a significant percent improvement but not that signficant an improvement when comparing to the original starting point... ie. just doing the spring and contact enhancer does a fine job.

-awr

 

[enLIGHTenment]

For anyone looking at the FET route, Vishay will shortly have in production an N-Channel MOSFET with an Rds(on) of 0.0018 ohms. It is, however, an SMT part.

http://www.vishay.com/docs/73384/73384.pdf

 

[bfg9000]

I'd like to modify the switch as well, but have the older (more difficult to remove) style switch. Can anyone confirm that the slider and spring fix is the same in these older switches? AtomicX, if your host is indeed from 1988 then it should also have the old switch.

I like these prehistoric Mags because they are already bored over 35mm straight from the factory. After burning in a Fivemega axial bulb in one, it now turns on and off while focusing, so it looks like I'd have to go in there to clean the slider anyway

 

[andrewwynn]

actually.. i measured a drop in resistance from 18 ohm down to 0.020 ohm from 2 drops of pro-gold.. i had the exact problem you described with my old Mag3D. I can't confirem or deny exactly what the fix would be but if you can get it apart i'm sure that the braided copper across the spring will halve the resistance.. you most certainly have more than that in the slider if youc an see a difference in brightness. (much less on/off).. which is almost what mine did.. went from about MiniMag to solitaire in brightness before the pro-gold. now it's as bright as new with fresh cells.. sadly about 1/30th of my M85 and 1/100th the M100.

-awr

 

[wquiles]

For anyone looking at the FET route, Vishay will shortly have in production an N-Channel MOSFET with an Rds(on) of 0.0018 ohms. It is, however, an SMT part.

http://www.vishay.com/docs/73384/73384.pdf

You don't have to wait. The N-channel MOSFET that I used in the Soft-start circuit had an Rds(on) of 0.0045 ohms (at a very low Vgs of 4.5Volts), so basically 4.5 miliohms vs. 1.8 miliohms - nothing significant, even at the 10 Amps of the super hotwires we get to play with

Will

 

[bfg9000]

Wow, I followed awr's suggestion for ProGold (+ used DeOxit first), and problem solved! Perhaps I'll hold off on the spring mod for now since I'm pushing the axial bulb a bit on 8 cells. I'll try it when the drop-in regulated battery carriers become available. Thanks andrewwynn!

 

[andrewwynn]

PG is just amazing stuff.. it's used on every contact of every light i build or modify.

-awr