Performance of Bright Lamp Configurations

I ran some tests of some of these super bulbs to compare them with SureFire performance. There has been a question in my mind of just how many lumens are generated by the various home-brew configurations that some of us interested in Mag mods have been talking about.

Part One of the tests was the measurement of relative brightness. I used fresh batteries for each of the configurations. I shone each light against a wall and measured the reflected light with a Wavetek LM631 light meter. Ambient levels were between 0.06 and 0.11.

This kind of test tends to measure the overall brightness of a light and approximates the total lumens put out. Lights with narrow beams (with Mag reflectors) have an advantage and will register somewhat higher with respect to those that have more spill (like the Tigerlight textured reflector).

Part Two was measurement of performance outside of the host. I moved batteries and lamps to my breadboard where I could measure voltage and current at the same time.

Shown in the chart below are:
<ul type="square">[*]Host that the lamp is in
[*]Lamp assembly or bulb
[*]Battery configuration
[*]Lux measurement -- a relative indication of brightness
[*]x30 is an arbitrary multiplier to suggest Surrogate Lumens
[*]Advertised lumens by SureFire
[*]Volts measured on breadboard across lamp under load
[*]Amps measured on breadboard through the lamp
[*]Watts calculated (V * A)
[*]Amps measured while lamp in host
[/list]
Host____Lamp_______Batts_____lux__x30_Adv___V____A___W__AinHost
Mag2D___TL texturd__2x3x123__4.80_144______8.2__2.08__17
SF_M3___MN11________3x123__8.76_263_225__7.2__2.60__19__2.52
Mag2D___WA01185___2x4x123_12.59_378______9.9__3.18__32__2.65
Mag2D___Carley809___2x4x123__5.92_178_____10.9__1.22__13
Mag3C___WA01318__4x123+AA_11.59_348_____10.6__2.00__21__1.93
SF E2e___MN03_________2x123__1.80__54__60__5.3__1.12___6
SF E2+___P61__________2x123__3.72_112_110__4.4__2.44__11
SF L4_________________2x123__2.28__68__65
SF A2_________________2x123__1.84__55__53

The WA 011185 did not live up to the expectations I had. I discovered that the reason is unwanted resistance in my host at high current levels. The current in host is much less than on the breadboard, where I have taken pains to make low-resistance connections. Note that the degradation of current for the SF M3 and in the Mag3C is less severe. I believe the problem is due to the hurried job I did of constructing an adapter to allow two parallel stacks of 4x123 in the Mag 2D.

This is what I have measured. I leave the fun of analyzing it to you. I have found that an inspection of the value of Surrogate Lumens per Watt give rise to some interesting ideas. Any analysis is helped by knowing that:
<ul type="square">[*]The values for load voltage are not precise. On fresh batteries, voltage plummets fast.
[*]The values for current are much more stable.
[*]Surrogate Lumens is an arbitrary number derived from comparative measurements. Anything derived from it will also be relative.
[/list]
I must mention that Ginseng was of immense help in doing this. He loaned me the WA bulbs. But far more than that, he has been a welcome source of information, ideas, and advice. Thanks a lot Wilkey.

Paul

Paul,

Very interesting results. It looks like for the most part, SF's lumen claims are pretty accurate. I am also disappointed in your findings on the 01185 but as you say, host resistance is probably the key factor. It is certainly pulling a massive amount of wattage on your breadboard so once the path is cleaned up, I would expect it to perform more acceptably.

It also looks like the poor Carley 809 took a hit in the 2D host. It's putting out a mere 56% of rated output. If this ratio holds for the 01185, it would be making about 670Lu if the resistance was tamed. Still, a far cry from the 817Lu spec.

The 01318 is the bulb that I thought would do better. The spec is for over 500 lumens but you're running it significantly overdriven and still only getting 65% of spec. I would've expected better. How much current drop did you see in the 3C? I can almost bet the bottleneck would be those two wing-spring contacts inside the core of the mag switch. Very small ribbons of metal and at least on the units I've opened up, pretty well hazed up with a surface film of oxidation. This is why I do a full take-down of the switch assembly and clean and treat every metal piece with contact conditioner.

Come to think of it, the aluminum piece you cut for the battery + contact might be the problem. If it's just a piece of flashing, it's probably coated with machine oil and/or oxidation. Did you grind out the anodizing on the inside surface of the tailcap? You must have to some degree otherwise you wouldn't have been able to complete the circuit.

One thing of interest to note, it seems like the WA bulb is quite a bit more efficient compared to the SF MN11. Makes sense as this is one of WA's more efficient bulbs at a spec eff of 28.8 Lu/W.

I think that when I get the WA01274 7.2V, 2.77A, 553Lu bulb, I might ask you to take a look at it. It should be quite conducive to a 2x3x123 setup in a 2D. Still...

My nagging concern is now the Mag internal resistance. Ever since I got a look at the inside of the switch assembly, I've suspected it would be a bottleneck for intensive applications. There are just too many tiny metal pieces to oxidize and tiny metal strips do not provide the best free-flowing current path, right? The whole point to dropping in these high and ultra-high output bulbs is diluted somewhat by the issue of internal resistance. How do we get past this to unlock what these little gems can really do?

Still, all in all, I'm very pleased with the WA01318 bulb. It packs a decent wallop in the 4.5x123 config and in a Mag with a cleaned up switch assy, might do significantly better.

As for my mentors, gotta give the thumbs up to lemlux and illuminated. I said it before and I'll say it again...isn't there anything that can be done to a Mag body that these guys haven't already done? Come on guys, throw the greenhorns a bone!

Wilkey

I wish I could get a look inside a MagCharger switch assembly. I'd love to see if it's set up differently for the higher wattage bulbs it uses. I'm sure the regular Mag bodies were never designed for the type of push we apply.

This work makes my recent investigations of switch swapping all the more relevant. I'm presently looking at an industrial mini-rocker with 20 milliohm maximum internal resistance for LED applications but if I could find some way to embed it in the stock switch assembly, I would try it for these incandescent applications. Holy crap, an idea just came to me. Gotta go think about this a bit and then write it down.

I'll be back in the morning.

Wilkey

Wilkey,

Right. It probably requires some sleeping on.

I have some comments on your comments. First, I don't agree that SF's lumen claims are pretty accurate. Remember, that 30x factor is arbitrary. The results are, for the most part, proportional to SF's Lumens -- not equal to. All we have learned is that if they have overrated one light by X percent, all of them are overrated by about that amount. I strongly suspect, however, that their claims are valid.

I too am disappointed on the 1185 results. Part is due to losses in my particular host configuration. Part is due to the fact that I was impatient in getting the light to focus tightly. With more light off-axis, the readings fall. It's a big part, I think, of why the 1318 performed so much better than the 1185.

I don't think that the Carley 809 performance was reduced significantly by a host-resistance problem. There's relatively little current flowing in this circuit. The problem may be appearing because of comparing the output with that rated by the manufacturer. I believe that this is the major source of all disappointment with these bulbs -- that Carley and WelchAllyn have standards for specifying Lumens that are different from those of SureFire. SureFire rates the whole light, while Carley and WA rate only the bulb. I suspect that a 500 Lumen SureFire flashlight will give perhaps as much as 50% more output than a WA or Carley bulb in a Mag lite -- and not just because of internal electrical resistance, but also because of photons getting lost in cracks and o-rings and lenses and other unaccounted sinks. Looking at it the other way around, I suspect that the MN11 lamp assembly yields more light than the M3 flashlight in which it is used.

I do not mean to say that WA and Carley lamps are no good, but that we may be expecting too much based on their ratings. The 1318, for example, is touted at 500 Lumens, but I am beginning to think a realistic rating in a light is more like 300 to 350 Lumens.

Getting back to resistive losses for the 1318 in the 3C, there seemed to be no significant loss in the host, as shown in the chart. In fact I was quite pleased with the overall performance. I lucked out in that when I first turned it on it had a magnificent beam -- a tiny circular spot. I didn't have to mess with bulb placement. I can see why you list this as your favorite light. I like it a whole lot.

Because there was no degree of degradation of current in the 1318 inside the Mag 3C, yet there was some for the 1185 in the Mag 2D, I conclude that the problem is not the Mag, but my series of plates and such that made up the adapter. I used clean pieces of aluminum. However, the pressure between contacting surfaces (and there were quite a few) was uncertain. I had no room for springs, so I just tightened down the cap as much as I could. This is a very poor way to house high current devices. I'm quite sure this is the source of the problem for the 1185. The Mag 3C had no plates or disks.

Now I am the one who needs some sleep. I appreciate your comments and questions a lot. I'll take a look at this response in the morning and see if it makes any sense. Good luck with that new idea. /ubbthreads/images/graemlins/smile.gif

Oh, yes. Lemlux and illuminated. Where would we be without them? I hope they will have some sage advice or thoughts to add.

Paul

Paul,

I'll concede the point on the SF output ratings. You've got much more experience with those torches so I accept your elaboration. I agree that the lumen output measurement is a bugaboo that can tend to mislead one in the quest for the ultimate output. Except for SF, who must measure their lamps in-light, the bulb manufacturers simply stick the bulb in a jig and then insert the unit into an integrating sphere spectrophotomer. In fact, this is how WA states they measure their bulb output. I assumed you came up with the arbitrary modifier of 30X by comparing the SF listed to measured output. Was this how you arrived at this factor?

However, I'm a bit uncomfortable with the fact that up to 200 of 500 lumens may be lost to flashlight-specific absorption losses. I don't know what proportion of the applied electrical energy is converted to photons, but a 40% loss seems intuitively high to me. I'm not saying it can't be happening. Just that I don't see what in the head of a Mag could suck up 200 lumens of light. That's the equivalent to the measured output of a W bin Luxeon in a SN II.

I looked at your chart again and I see the huge current drop of the 01185 in the 2D setup. The M3 and 3C seem to both be running at pretty high electrical efficiency. I can't wait for the DSpeck butt cap to come in so I can make my 3C robust again. I am torn about the 01185 though.

Wilkey

Good morning Wilkey,

Yes, that's how I arrived at the 30x factor. I calculated the factor for each individual SF light. It ranged from 26 to 33. So I chose 30. No more scientific than that. /ubbthreads/images/graemlins/blush.gif

Well, I'm uncomfortable also with that much loss in a light. I think I may have been blowing some of smoke last night. There could be 10% or perhaps 15% loss here (a stock Mag lens absorbs or reflects about 8 to 9 percent) -- but you're right, 40 to 50 percent photon loss is not reasonable.

Now I'm thinking that perhaps there is a component of loss which is due to a mismatch between rated voltage and actual voltage in the level that the bulb is run. I'm sure that SureFire optimizes each bulb to the specific battery configuration that is used. This is not the case with bulb makers. We try to fit the bulb to the battery configuration. Sometimes we get too much overdrive (Carley 808 with 4x123) -- very bright, but with low bulb life. Other times we get not enough overdrive (Carley 809 with 4x123) -- no risk of flashing, but disappointingly dim. And then, there are the rare times when the overdrive is perfect (WA 01318 with 4x123 + AA) bright and stable.

In this continuum of degree of drive level, there could be an almost magic place where we would get the lumens advertised by the bulb makers. That is, with a completely adjustable power supply, we could make the bulb perform at spec. It would mean there's a sort-of bulb lottery. You gotta test it to know. And even then, there's a small amount of variance between individual bulbs with the same number designation.

When using lithiums and driving a bulb in the region of 10 volts, we can get voltage to within +/- 0.75 volts if we allow the use of AAs. That's a range in drive level of about +/- 7.5 percent. If we're confined to just 123s, it's a range of +/- 15 percent. With rechargeables, the range is in between those two. And, of course, for configurations with lower total voltage, the range is even larger (in percentage terms).

So, perhaps the combination of loss in flashlight and mismatch of driving voltage could result in the loss of performance we're seeing, remembering that the "loss" is not really a loss, but a difference in rated performance between a flashlight and a bulb. Does this seem reasonable to you?

Like you, I await that tailcap from DSpeck. When it arrives, the 1318 in 3C will be formidable. With regard to the 1185, I'll do some more experimenting. As I recall, the light was a little yellow, meaning that the bulb could be driven more. It will take more thought, experimentation, and willingness to go to a 3D to make it work well, I think. Perhaps a 2x4x123 + AA would be appropriate? I would give in and use rechargeables, except that my usage profile is such that rechargeables are just too inconvenient. (In any event, I still plan to include some potted 1185s in my WA order.)

BTW, I took your advice. I bought a Dremel to nibble out the cavity in the bulb adapter. It took me 6 hours to read the material, learn how to use the thing, and exchange it for another with accessories that better matched my needs. And then it took only 10 minutes to do the job. That was a big investment in time and frustration for me. But I believe it was a good one, because now I'm set up to modify Carley reflectors and do some of the things that you've been doing. Thanks for your counsel on this. /ubbthreads/images/graemlins/grin.gif

Paul

I agree wholeheartedly with your referring to this performance difference as a deviation and not a loss. That makes much more sense. Still, I get the feeling we're missing something in this equation. Shouldn't we be able to consider these bulbs as simple resistors? If so, wouldn't the power dissipated obey straighforward electrical laws? And if the photon output is directly related to power dissipated (which the WA rerating formulas seem to imply) shouldn't then there be a direct relationship to brightness? Perhaps only if the photon efficiency is unchanging across the V/I range. If the photon efficiency were not stable (ie. fluctuates either high or low outside of the V/I sweetspot) then that could explain your measurements and the deviation from spec. I'm sure when WA measure their bulbs, they hook it up to a robust, highly regulated supply able to lock to 9.600V and 3.150A. Alas, even SF123s can't do that.

I am also in agreement about the flexibility that AA liths and rechargeables provide in fine tuning the initial voltage. This makes the powerpack reasonably tunable given the single series arrangement in a C tube. It's somewhat more complicated in a parallel serial arrangement in a D tube but with the added benefit of reducing the current draw per stack. I just wish it were more convenient.

Since the 01318 and 01185 are both 9.6V bulbs, I'd think the 2x4.5 in a D would be pretty good for the 01185. The correlated color temperatures of both bulbs are pretty darn close so if it looks yellow, it is definitely being underdriven.

Congratulations on the Dremel purchase. I think you'll find it quite useful...and capable of getting you into all kinds of mischief /ubbthreads/images/graemlins/wink.gif

Wilkey

PS. I think illuminated would have chimed in but he is away from computer until the end of the week.

I have no way except for my Mk1 A1 Revision1 eyeballs to measure anything.

That said I can not find one single thing to complain about the level of light that comes out of my M*g 3C with an 809 bulb and 4x123+1X1.5AA. It is awesome!

I also have a 2D that seems to eat up some power somehow.

I just like the C lights better anyway!

Joe,

I like the 809 in 4.5 setup as well. That was my prior fave. The WA01318 is even a step above that. Not twice as bright by any stretch but mighty impressive. As PaulW and I've come to find, of the 9.6V Carley and WA bulbs we've tried, the WA01318 has emerged as the best...able to run well on 4.5xlithium, intermittent usage not requiring an aluminum reflector, only $4.85 potted and ready to rock, and bright!

Wilkey

Also running 4x123 and an AA?

If so, how can I get one (or two)???

I guess I have to watch for the next group buy.

Wilkey,

Yes, I think that the WA 01185 can be driven harder. I think it would be interesting to take the voltage even higher than 2x4x123 + AA -- maybe 2x5x123? Or did I read somewhere that someone tried that and it flashed on him? Well, I won't be doing any destruction testing until I get a bunch of them. And I'm beginning to believe that destruction testing may sometimes be a necessary part of this process.

Regarding the Dremel purchase -- just what are all the kinds of mischief that the Dremel can help me get into? /ubbthreads/images/graemlins/grin.gif

Paul

2x5x123...I'm afraid. Still, blowing an $3.90 bulb is a lot less painful than blowing two $8 bulbs in the space of a few minutes. Ouch!

Wilkey

Well, the spiral cut drill bit is really nice for cutting drywall and even light plywood. The grinding stones sharpen lawn and garden blades fast! But I haven't had much luck with the polishing attachments, probably not using the right compound. I had so much fun I bought a set of diamond burrs. Oh, and for serious grinding, I use a cutting oil. Really makes the cutting go faster and trouble free.

Wilkey,

1. Whoa. $3.90 for the 1185? I thought it was $2.20. Maybe I'd better be a little less rambunctious in my testing.

2. Cutting and grinding wasn't the kind of mischief I was hoping you'd advise me on. But if you're going to get serious on me, I'll have to thank you for the cutting oil advice. It makes sense, but I wouldn't have thought of it.

Paul

$3.90 potted. Carley has raised some prices. They want something like $1.73 to pot...and a few bulbs to trash as they get the technique down. I had no idea they needed sacrificial bulbs.

Wilkey

Paul,

Do you have any feeling as to which is more critical in achieving the higher brightness, voltage or current? If a 9.6V 2A bulb is driven to 10.8V and 2A would that be brighter than 9.6V and 2.5A?

From looking at the WA and CA lamp listings, I suspect that current is the bigger factor in how bright a given family of bulbs get but I don't know how to rationalize that.

I was playing with my WA01318 in PR912 tonight and an unexpected benefit emerged. The soft potting allows one to fine tune the filament position relative to the reflector axis. I guess it'll last until it's hardened up and crackly. And damn, makes a nice amount of light. I really like this bulb.

Wilkey

Guys,

If memory serves...the info from the WA site "electrical characteristics" file states that small variations in current can cause signifigant changes in lamp performance, or something similar to that. I can't go back and re-read it right now due to need for sleep...

BTW - Thanks for the comments, but you guys give me waay too much credit. Sounds like you've done quite a bit more testing/evaluating than I have. I've just learned a few tricks (mostly mechanical) for lamp retrofits and focus improvements. Much of my knowledge on batts I learned from Lemlux (Thanks) /ubbthreads/images/graemlins/smile.gif and from tinkering.

As far as the switch resistance issues goes, well, I'm still convinced that a packaged semiconductor power management device designed to drive a MOSFET will be the hot setup, and low battery cut-off for rechargeables can be included as well. No more need for switch cleaning, flickering , etc. Micrel and others have such devices listed, and some look rather promising.

Gotta run (sleep) - John

Wilkey,

Well, brightness depends on color temperature, which depends on power consumed in the filament, which is the product of voltage and current. A graph of voltage vs. current shows that, for a given bulb, the relationship of voltage to current is fixed. So, I think that your question is more philosophical than practical. But, to be philosophical, my view is that voltage plays a bigger part since it varies faster -- as some power (greater than one) of current. The relationship is given, as I recall, in the WelchAllyn re-rating charts, as John mentions. When I hook up a bulb to a fresh battery configuration, the voltage plunges very rapidly, whereas the current drops relatively slowly.

Thanks for the alert on the potting. It sounds like I will want to diddle with the bulb positions in the sockets right after I get them. Yes, the 01318 is great. In practice it has a higher color temperature that the 01185, meaning I'm not driving the 01185 as hard. I'll see what I can do about that. Hmmm, I remember using an optical pyrometer in a materials/shop class I once took. I wonder if someone makes cheap ones for the home experimenter?

John,

Yes, the WA site's "electrical characteristics" file is the place with the info. Getting philosophical and building on your comment, I'll contradict what I said to Wilkey. I think I now believe that current plays a bigger part because a small increment will cause a larger change in power than the same percentage increment in voltage. Ah, but that's playing with words. It's more fun to play with flashlights. /ubbthreads/images/graemlins/grin.gif

I like what you have to say about "a packaged semiconductor power management device designed to drive a MOSFET." I'm going to snoop around a little on that. I'd appreciate hearing about anything you come up with.

Paul

[ QUOTE ]
PaulW said:
I ran some tests of some of these super bulbs to compare them with SureFire performance...I shone each light against a wall and measured the reflected light with a Wavetek LM631 light meter. Ambient levels were between 0.06 and 0.11.

This kind of test tends to measure the overall brightness of a light and approximates the total lumens put out. Lights with narrow beams (with Mag reflectors) have an advantage and will register somewhat higher with respect to those that have more spill (like the Tigerlight textured reflector).
Paul

[/ QUOTE ]

Paul,
I was wondering if you knew why narrow beams have an advantage? The only way I would think that you could determine that would be to measure the focusing lights at various degrees of focus. If there were a solid correlation between defocus, or the size of the main hotspot, and the measured reading, I would be surprised.

My own guess is that the wall surface has its own reflectance value which is independent of intensity. If this is true, then the same bulb at various degrees of defocus would throw the same number of photons over a progressively larger area that has reflectance of <100%. Let's say the paint on your wall reflects 80% of the incident light. If you shined a focused cone with a hotspot area of 20 sq.in., you would lose 20% of the photons in that area. If, however, you shined a diffuse cone over 200 sq.in., you would still lose 20% of those very same photons which are now distributed over a tenfold larger area. This explanation seems to negate your statement about the advantage of tight-focusing beams.

I wonder what blend of luminous intensity and luminous output the lightmeter is actually measuring.

Wilkey

Definitely worth a bump for the new folks on the block.