Not All Switch Are Created Equal!

modamag

Flashlight Enthusiast
Joined
Aug 16, 2004
Messages
2,101
Location
Bay Area, CA
I've always heard the so call high resistance problem of a Mag switch but no one ever told me how high. Well, I finally got some spare time to conduct the test.

=== 1. Parts List ===
- KRM AEEC-300 DC Power Supply
- Fluke 23 Digital Multimeter
- Bel MERIT DX460L Digital Multimeter
- 4x 0.8 Ohm 1W 1% resistor
- 20x Brand New MagD switch

=== 2. Setup ===
The circuit is connected as follows:

Power Supply (+) -- Fluke 23 (200mA mode) -- Alligator clip wire -- Switch -- 4x Resistor in parallel -- Power Supply (-)

Bel Merit is connect in parallel to the (-- Alligator clip wire -- Switch -- Resistor --) to measure the voltage.

=== 3. Test Scenario ===
Case #1: Resistance from the anode spring to the bulb contact point with the spring extended to the maximum height. Sample set of 20x switch randomly selected.

Case #2: Resistance from the anode spring to the bulb contact point with the spring compressed to the miniimum height. Sample set of 20x switch randomly selected.

Case #3: Resistance of the switch core, the part that you slide out of the switch assembly. Sample set of 10x switch with the lowest resistance from earlier test case.

=== 4. Test Result ===
Base Info: the voltage across the Alligator clip & resistor is 1.49 V and the current load is 2.98 A so the total resistance is 0.500 Ohms.

Case #1: Uncompressed Spring
Mean Switch Resistance = 0.045 Ohms
Median Switch Resistance = 0.044 Ohms
Standard Deviation = 0.008 Ohms
complete data set

Case #2: Compressed Spring
Mean Switch Resistance = 0.033 Ohms
Median Switch Resistance = 0.034 Ohms
Standard Deviation = 0.008 Ohms
complete data set

Case #3: Switch Core
Mean Switch Resistance = 0.005 Ohms
Median Switch Resistance = 0.003 Ohms
Standard Deviation = 0.003 Ohms
complete data set

Note: 15% upper and lower outlier ommitted from statistics. All data have been normallized for thermal affect on resistors and wires.

=== 5. Analysis ===
So what does this all mean? Here's couple quickies.

1. If your using a bi-pin slug (incandescent) or ANY PR based adapter such as Diamond LED, you're basically adding 0.05+ Ohms of resistance. It could also be worse (much worse) depending on the oxidation and condition of your switch.

2. You could be loosing the following % in efficiency.

Mag74 (9V @ 3.0A) ... 0.3% - 2.4%
Mag85 (10.5V @ 3.5A) ... 0.3% - 2.4%
LuxIII (K-Vf @ 1.0A) ... 0.8% - 2.0%

The root of all this resistance is from the contact of the switch core to the spring.

I have also check against the connection between the switch core and the housing plate in the back, there is practically no resistance. It seems to make contact very well because of the rigid structure.

=== 6. Remedie ===
Well many of us may already know, but just in case there are newbie or our memory gets away from us.

1. By-pass the spring if you can. Solder directly to the leaf spring of the switch core with the appropriate wire gauge (18G+ for incan, & 24G+ for LEDs).

2. If you have to use the PR socket then try soldering the spring to the can will help, and make sure you thoroughly clean the contacts with alcohol.

3. Make sure the battery anode make good contact with the switch spring.

=== 7. Closing ===
With proper precaution/modification you could easily use the MagD Switch for 100W application. In such case you'll loose ~1W of heat thru the switch contacts.

Also from preliminary analysis the MagC & MagCharger have a slightly "better" switch mechanism but due to lack of sample size I can not conclusively give a verdict.

Editted section #4 & #5 for higher current test which is more applicable to flashlight usage.
 
Last edited:

270winchester

Flashlight Enthusiast
Joined
Aug 9, 2004
Messages
3,983
Location
down the road from Pleasure Point.
WOw, thanks Johnathan.

I realized that the stock mag has resistance evrywhere, but not that high!!!

I'm glad my 100w light only retains the mag tube and everything else is straight wires and high-current appropriate connectors and switch. My last 100w light, using a 4D, was hopelessly yellow even when driven hard by 12 1650s, and I knew somethign was wrong when I could drive a 62318 wuth 12 cells without blowing it. Now that's why!!!!

Great work, Jonathan.

Nick
 

Ginseng

Flashlight Enthusiast
Joined
Feb 27, 2003
Messages
3,734
Great data!

Thanks for your hard work. So it seems that the main spring and the cap it contacts are the bad guys. Perhaps not coincidentally, those are the parts that seem to be plated with something to make them corrosion resistant. Do you think that might have anything to do with the high resistance at that junction?

Wilkey
 

Timson

Enlightened
Joined
Mar 5, 2004
Messages
526
Location
Leeds - England
Very Enlightening...

So the best bet - is to do away with the cammed focusing system and PR based bulbs and hard wire a bi-pin adapter into the switch core for minimum loss.
Focus via mag head threads.

Tim.
 

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
Best bet would be to rewire the POS switch so that the current is carried in a P channel FET and the switch just switches the gate of the FET... The Source/Drain of the FET would of course be soldered with appropriate gauge wires directly from the battery +ve contact to the bulb +ve side.

george.
 
Last edited:

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
evan9162 said:
George,
Any reason why an N-channel FET wouldn't work as well?

It would be easier with a P-chan - wiring wise, since what you want to do is cut the positive feed from the battery (that normally goes through the switch) and put that into the FET. Output from FET then goes to the bulb/holder etc. The ground connection to the bulb/holder remains as before. You rewire the switch so that it now switches the ground path (similar to the nFlex/uFlex etc Mag mods - but you leave the switch mechanism alone, i.e. DON'T convert to momentary). So, now when the switch is 'on' it grounds the Gate of the P-FET and turns it 'on'.

You need a 100K pullup resistor as well to keep the Gate high (off) when the switch is not 'on'. The Gate is extremely high impedance so it's not as if it will draw any significant current when 'off'.

george.
 

js

Flashlight Enthusiast
Joined
Aug 2, 2003
Messages
5,793
Location
Upstate New York
FINALLY!

Excellent work! I'm so glad to see some scientific testing done on this. We have all wondered for years what the real amount of resistance was, and now we know. I never would have guessed it would be this high. Yikes! Another good reason to NOT use PR base lamps.
 

wquiles

Flashaholic
Joined
Jan 10, 2005
Messages
8,459
Location
Texas, USA, Earth
modamag,
Thanks for the hardwork - great info !!!




georges80 said:
It would be easier with a P-chan - wiring wise, since what you want to do is cut the positive feed from the battery (that normally goes through the switch) and put that into the FET.

George - could you please suggest 1-2 specific part numbers for a good P-chan FET for us to buy and try out?


Will
 

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
wquiles said:
modamag,
Thanks for the hardwork - great info !!!

George - could you please suggest 1-2 specific part numbers for a good P-chan FET for us to buy and try out?


Will

There's so many out there. It really depends on what voltage you plan to run and what current you need to handle.

For low voltage, like 3V, you need a MOSFET that has logic level at the Gate so that it fully turns on a 3V (or even 1.8V).

For higher voltage, like >10V, then most MOSFETs will fully turn on and can easily handle 10+ amps with resistance well under 0.1 ohms.

International Rectifier makes a large range of MOSFETs and with some of the above questions answered an appropriate choice could be made.

george.
 
Last edited:

wquiles

Flashaholic
Joined
Jan 10, 2005
Messages
8,459
Location
Texas, USA, Earth
I do as well for the LED's. Using the FET should be great since my 18xAA Fivemega Mag85 is currently using the potted bulb :sick2:

Will
 

Crux

Newly Enlightened
Joined
Nov 4, 2003
Messages
160
Location
Mentor, Ohio
I did a quick count of the series of contact points that carry the current in a 2D mag...

Nineteen!! :wow:

Correct my count if needed, but there are a lot of links in that chain. Maintaining, improving or eliminating them will increase efficiency. Just be careful not to make things worse, you have to figure that Mag wouldn't have put in a part that didn't need to be there.

Crux
 

andrewwynn

Flashlight Enthusiast
Joined
Apr 28, 2004
Messages
3,763
Location
Racine, WI USA
guess i'm glad my Mag85 has the bipin mod (direct wired to the switch).. should be only 0.12 ohm.. x3a = 1W loss on the switch which is about 3%.. i can live with that... wowza on the 25-45% loss with the springy switch.. ouchorama.. I thought that has to be wrong 'til i did the math myself and got 25.7% of the power going to heating the switch.. i'll stick with my 3%... anybody with PR 1185s out there.. upgrade.. whoa!

glad to hear there is a big improvement when the spring is compressed... my BAM! device goes right into the PR base so it's a zero-mod design.. fortunately it does use a driver for constant output and considerably lower current.. it's probably safe for me to use the 1 ohm values for calculating the circuit in my case.. wow that's still up to 3.25W or 20%.. guess time for me to do some testing on that design as well..

I am wondering how much the low current in this test may have affected your results.. i would be very surprised if i didn't notice a 3W drop on the PR base in all my testing.. i think we need to see the results of running 2A or 3A through the switch and what is the voltage drop in such a case. I have noticed for example.. that i have MUCH worst problems with the kroll switch when it's trying to control 20mA than if it's switching 700mA.

I guess time to do some more measuring thanks for bringing this important issue to the table.

-awr
 

andrewwynn

Flashlight Enthusiast
Joined
Apr 28, 2004
Messages
3,763
Location
Racine, WI USA
update: i actually measured a D-sized switch... i measured through the entire switch mechanism.. from the postive spring to the negative strap... i shorted out the PR with a piece of metal (back side of a steel tweezer)..

I had enough length in my leads i could run direct from my bench supply and set the amps to 2.0 and to 3.0A with no added resistance... (of course at like 1V or something)

Here's what i measured.. a repeatable 0.31V drop at 2.0A and a repeatable 0.48V drop at 3.0A.. ohms law makes for .31V/2A = 0.155 ohm at 2A.. 2Ax2Ax0.155ohm = 0.62W.. at 2A i'm usually putting more than 16W into the emitters.. so 0.62W/16W= 4%loss. (high speed with the RT4).. and at 3A.. it's 0.48V/3A = .16ohm.. basically the same... but at 3A it's 3x3x0.16ohm = 1.44W.. however i'm pulling about 35W in my mag85.. so 1.44W/35W = 4.1%... so.. just about the same percentage loss... my mag85 has a direct-wired bipin.. so it should be a little less..

So.. promising news... i hate to throw such a monkey wrench in the works...

Take the middle example.. compressed spring.. 1ohm switch resistance and 37mA.. only 1.37 mW lost on the switch... i think that the low current threw a curveball into the test.. based on a few things like that i've never noticed a HOT switch which you most certainly would... putting 15W (worst case example of the mag 85) into the switch would turn it blue and melt it (the steel) in pretty short time.. we melted a spring in the modamag 8xAA adapter making the 'mag100' when there wasn't enough spring tension (we fixed that and now it works fine).

Thanks for the thought provoking test... i don't have the parts to do a more detailed test but it would seem that apples and oranges come into play here.. i.e. what happens at 37mA is not what happens at 3000 mA.. there is some difference in the resistance based on the actual measurments i did on my switch.. of course i only did it one switch.. hopefully you can repeat the test..

I have a question about how you measured.. why did you measure across the switch and the resistor and subtract the resistor? and did you measure just out the postitive of the PR base or did you measure back through the switch as i did? You should be able to hook your mV meter from the pos. spring on the mag switch to the neg strap next to the hex nut and short out the pr.. putting some aluminum foil around an actual bulb and putting it in should do the trick.. if you have a curent limiting power supply like mine where you can dial in the current it'll pump out 2A or 3A directly and mine is designed to handle direct shorts (that's how you set the limit.. short it out at like 1V and dial in the current max).. so it is a cake-walk to get the current set..

Keep us posted about if you can re-do any parts and what your findings are... thanks much.

-awr
 

IsaacHayes

Flashlight Enthusiast
Joined
Jan 30, 2003
Messages
5,876
Location
Missouri
Modamag - I just took apart a D switch for the first time. It is different than a C switch. And it has tiny contact points. A C switch has much greater surface area. Hmm C mags are superior, switch, and space in the head for thicker Perfect Heatsink :) Night all!!! no more modding for me! bed time!
 

andrewwynn

Flashlight Enthusiast
Joined
Apr 28, 2004
Messages
3,763
Location
Racine, WI USA
after doing some 'real world' testing.. manoman.. my high current lights are all getting an FET switch... with the mag85 we'll get about 0.45 more volts to the lamp! With BAM! the driver just sucks more amps so it doesn't care about the voltage dip and it just means there is some efficiency loss on the switch.

example.. 1.8A (typical bam draw).. .288V drop meaning .51W loss.. not terrible since i'm pushing like 18W through the system..

however on the mag85 and other hotwires.. holy cow.. when you change the voltage by 1/2V.. it will change the lumen output by 30+%..

Luckly somebody posted step-by-step destrutions how to make a mag switch into a 'momentary' to run the likes of an nFlex driver.. but could easily be converted into an FET controller.. using the 'switch' just to control the gate of an FET.

the intl. rectifier 3030 FET is 0.026 ohm and they have a mega premium ($7) FET that is 0.006 ohm with 10V on the gate.. and 0.009 ohm with 4.5V on the gate.

Soo.. guess i'm going FET... hope to have some really simple directions with photos that people can copy.. thanks for the 'seed' that got this going MM.

-awr
 

modamag

Flashlight Enthusiast
Joined
Aug 16, 2004
Messages
2,101
Location
Bay Area, CA
I reperformed the test with a more "realistic" current load of 3A. The results significantly changed as you could imagine from the increased voltage potential.

The first post section #4 & #5 have been editted for the new results. Keep in mind that these are brand new never used, thoroughly cleansed switch.

BTW: Andrew, 0.5V gained for a Mag85 is ~ 19.4%

Vdesign = 9.6V
Vapplied = 10.1V
ReRated lumen = (Va/Vd)^3.5 * Lumen_rated
ReRated lumen = (10.1/9.6)^3.5 * 816.81
ReRated lumen = 975.66 lumens
 
Top