KL4 bezel efficiency tests

I recently obtained a loaner KL4 bezel courtesy of luxlover in order to examine it and run some efficiency tests. I think the results may be of general interest.

I used a variable voltage, current limiting power supply with digital readout which I built myself for the tests. The current limit was set high enough so as not to interfere with the results of the test. I measured the input voltage directly at the battery contacts on the bezel with a multimeter because of ohmic losses in the test leads connecting the bezel to the power supply. I read the current draw directly off my power supply meter. In order to measure the current draw of the Luxeon V emitter I disassembled the bezel, disconnected one of the wires going from the regulator board to the Luxeon, and temporarily soldered an 0.1 ohm, 1% power resistor in series with the Luxeon. This only added about 0.065V, or roughly 1%, to the total output voltage. Although the Luxeon emitter quickly heated up due to the small heat sink it was on, I tried to obtain some consistency by keeping the power on periods to just a few seconds, and letting everything cool for a minute or two between tests. The circuit was in regulation at 3 volts and above. When it was in regulation the Luxeon current was 635 mA and the output voltage (including the drop across the 0.1 ohm resistor) was 6.32V. Therefore, the output power was always 4.01W whenever the circuit was in regulation. The Vf of the Luxeon V emitter itself was about 6.25V, meaning that it is probably S bin with regards to Vf.

Here are my results:

Vin=3.00V
Iin=1.78A
Pin=5.34W
efficiency=75.1%

Vin=4.00V
Iin=1.23A
Pin=4.92W
efficiency=81.5%

Vin=5.00V
Iin=0.96A
Pin=4.80W
efficiency=83.5%

Vin=6.00V
Iin=0.77A
Pin=4.62W
efficiency=86.8%

Vin=7.00V
Iin=0.63A
Pin=4.41W
efficiency=90.9%

I didn't go above 7V input because I didn't have the parts on hand to repair the regualator board if I damaged it. Also, if it is like many step up circuits it will go to direct drive once Vin gets much above the Luxeon's Vf.

In typical step-up regulator fashion the efficiency gets much better as Vin increases due to the smaller input currents that the MOSFET and inductor must deal with. Although it appears that the regulator board will work in both one and two cell configurations, it will likely be out of regulation for a good portion of the time in a one cell unit. The fact that it operates down to about 2.65V means it can literally suck a 2-cell configuration dry. Efficiency at 6V isn't bad. I suspect by maybe using a slightly larger inductor and MOSFET this can be brought up into at least the low 90s.

Output on my lux meter was 1000 lux or a bit under @ 1 meter once the bezel was reassembled. The readings prior to disassembly were the same.

Joe,

Thank you for your offer to test my original and slightly yellow tinted bezel from my second SureFire purchase, the L4. I am surprised that with such an important issue, nobody has had anything to write. Despite all of the models they manufacture, I still think that the L4 is one of SureFire's greatest achievements on all accounts. Your workup may be the only one done on CPF, with this extreme range of testing. The results are very encouraging to me, and should have been to other KL4 bezel owners as well.

I was very curious about it's Vf bin code, and based on your 6.25 volt reading after regulation, it is an S bin. As for output, we had determined that at one meter from the luxmeter the output was 1,050 lux, which is very high for this bezel. Thanks to you, I also found out that the regulation circuit is running at about 88% efficiency. That is a high number as well.

I commend you for sharing your time with all of us. I hope that in the future, you will be able to test other bezels for me, and share the results with the CPF community. It really does matter that guys like you do in depth testing of light sources, especially since SureFire does not offer us any information about the bin codes of the Luxeon emitters they are using. We are left uninformed, hoping that we bought a light from them that will have decent flux, tint, and Vf bin codes. You are a valuable asset to me as a friend. I imagine that you are appreciated by CPF as well, although nobody would ever know that based on the total lack of response to your post. Keep up the good work!

Hey JTR, 1-yr anniversary for you. /ubbthreads/images/graemlins/smile.gif

Good results. Much as expected. That's probably why there aren't many responses. Nothing to discuss.

Personally, I'll have to pull my thumb out and do some testing on mine. That way, we can have more than a sample of one. But not yet. I'm still very frustrated as I can't open the darned thing to access the innards. Everytime I look at it, I feel the urge to try to unscrew it again. And I've already scarred it a couple of times. I've tried heat and cold combined with leather straps on vices and something called a Baby-Boa (essentially a wrench that uses a rubber strap to grip the item). Still nothing.

Oh well, doesn't really matter, I guess. Still haven't found a good bin to replace the stock LED anyway.

jtr,

assuming that your KL4 is biased at the stock 660mA, it will be drawing a lot of juice, especially from a single 123. Did you do a test simulating the voltage drop of the batteries under load? If so, how did this affect efficiency? Since you had it on the bench supply, did you check to see what the actual Vf of your luxeon was?

[ QUOTE ]
Chop said:
jtr,

assuming that your KL4 is biased at the stock 660mA, it will be drawing a lot of juice, especially from a single 123. Did you do a test simulating the voltage drop of the batteries under load? If so, how did this affect efficiency? Since you had it on the bench supply, did you check to see what the actual Vf of your luxeon was?

[/ QUOTE ]
Yes, all that info is buried in the long second paragraph of my post but I'll repeat it here for clarity. It seems that the bezel I tested was biased at a bit less than the stock 660 mA. I measured it at 638 mA. This is fine-with anything mass produced there will be some variation from part to part. The regulation board has no fine adjustment pots. A variation of a few percent either way from nominal is to be expected. Vf was difficult to measure at best because the emitter only had a small piece of metal for a heatsink after the bezel was disassembled. As a result, the Luxeon heated up rapidly, and Vf quickly fell. I measured Vf at about 6.25V, but under actual operating conditions in the bezel it is likely a little more (perhaps 6.30V) due to the better heatsinking.

As far as the single cell testing, I didn't do that because the bezel was out of regulation under 3.0 volts. Even so, you'll notice that as input voltage decreases, efficiency drops. Therefore, under a loaded condition with a single 123 cell you might have Vin of 2.7V or so. Assuming that the bezel was still in regulation at that point, which is isn't, I would interpolate the efficiency at perhaps 72 or 73%, which would translate to an input current of a little over 2 amps. This is a heavy load to place on a 123 cell, so I would imagine the bezel was purposely designed to go out of regulation in a single 123 situation for safety reasons. Lithium cells have been known to explode when they get too hot. A regulated single cell situation could well result in thermal runaway. As the battery gets depleted the voltage drops, but at the same time current draw increases to maintain the power level, dropping voltage still further, until you have the electrical equivalent of almost a short on the battery (i.e. perhaps 1.5V with a current of 4 amps). The KL4 is really optimized to work with 2 123 cells (or 3 to 5 NiMH). Anything more, and you get into a direct drive situation which may destroy the emitter. Any less, and you go out of regulation. Indeed, the bezel won't even turn on until about 2.65V, and it draws a fairly small current until it gets near the point at which it starts regulating. At 2.65V it only draws 0.05A. At 2.85V, which is about where it would be in a safe single cell situation, it draws ~0.80A.

BTW, thanks luxlover for bumping this thread. /ubbthreads/images/graemlins/wink.gif I knew there was a lot of interest in it based on the page views. I was just surprised that nobody had anything to say until now, especially since I've seen this very question on efficiency asked in a few threads. As long as it can be opened, I can measure efficiency.... /ubbthreads/images/graemlins/smirk.gif

If anyone else can do some measurements of their own, it would be nice to have a few data points for comparison purposes. I may have measured an especially good (or bad) bezel. There's really no way of knowing.

First anniversary? /ubbthreads/images/graemlins/clap.gif I didn't realize it's been a year already. How time flies. Thanks for the sentiments, Steelwolf. It's been a great year. And I have my 42nd birthday coming up in 8 days as well... /ubbthreads/images/graemlins/party.gif

/ubbthreads/images/graemlins/cool.gif
thanx for doing it!
not that I have anything to add here ... it's all techie chinese for me /ubbthreads/images/graemlins/icon15.gif
bernhard

Thanks for the work and report! /ubbthreads/images/graemlins/thumbsup.gif Thanks for writting in a fashion that us non techies could follow as well! /ubbthreads/images/graemlins/grin.gif