Review: ArcMania's new SMJLED2 PR2 SS review - Page 6

chimo · #**151** 09-11-2006, 09:18 AM

That's a pretty drastic drop in output.

This type of drop could be achieved after only a few battery changes.

I am curious if these results are being investigated by the device creator and if there is an explaination (such as: the wrong LED was installed by the manufacturing plant).

Are these still being sold?

Christoph · #**152** 09-11-2006, 09:32 AM

What happened to the 70% 50% and 25% lines!

NewBie · #**153** 09-11-2006, 10:37 PM

Quote:

Originally Posted by Christoph

What happened to the 70% 50% and 25% lines!

Oops, fixed in last update.

Enjoy!

NewBie · #**154** 09-12-2006, 07:10 PM

Chris,

Added a 10% line for you.

NewBie · #**155** 09-13-2006, 08:00 PM

Okay, we are below the 10% output for lifetime test now (THIS IS NOT A RUNTIME TEST!). I was *very* surprised that the ArcMania SMJLED2 PR SS degrades so rapidly, and had expected to start flattening out much sooner, and dwindle off much slower:

As I mentioned earlier, a host with better heatsinking looks like it would help a lot, and something like a 2D Maglite with a modified HotLips would help a lot.

chimo · #**156** 09-14-2006, 10:14 AM

Well done on the thorough review!

Wow, that is quite the nose dive. I take it this will not get your recommendation.

Paul

nerdgineer · #**157** 09-14-2006, 11:04 AM

Newbie, I'd heard that the original SMJLED PR bulb had a 50% lumen maintenance point in the hundreds of hours range when running off alkies. Do you have any information on the original SMJLED PR bulbs in that regard?

PlayboyJoeShmoe · #**158** 09-14-2006, 11:13 AM

I'm more glad than I EVER was that I have old SMJLEDs!!!!

Hondo · #**159** 09-14-2006, 11:45 AM

I am curious what the input current was in this last test. From post #111, it looks like the LED is seeing around 200 ma at 2.8 volts in. Since this is a boost circuit, at the lower input voltage, the current should be even higher. I can only measure input current at the batteries on my lights, and I use NiMH's or not-too-fresh alkalines, and my lights range from 80 to only 120 ma on a ten amp scale. I was worried about just this phenomenon, and it had been my guess that I was keeping the LED significantly below the 100 ma rate at which the older MJLED's were known to degrade fairly rapidly. Thanks again!

Hondo

idleprocess · #**160** 09-14-2006, 01:41 PM

Any speculation or real information about what exactly is degrading? Phosphor/die being killed by the heat? Trace UV from the die turning the package opaque? Driver components going bad?

I imagine you'd have to shut down the rig to determine this, so I guess it's just speculation.

It would be interesting to know how many hours it would take to degrade under a more realistic duty cycle of 25 or 10%, but I understand that this sort of (ultimately) destructive testing is expensive and time-consuming.

NewBie · #**161** 09-14-2006, 08:20 PM

Quote:

Originally Posted by Hondo

I am curious what the input current was in this last test. From post #111, it looks like the LED is seeing around 200 ma at 2.8 volts in. Since this is a boost circuit, at the lower input voltage, the current should be even higher. I can only measure input current at the batteries on my lights, and I use NiMH's or not-too-fresh alkalines, and my lights range from 80 to only 120 ma on a ten amp scale. I was worried about just this phenomenon, and it had been my guess that I was keeping the LED significantly below the 100 ma rate at which the older MJLED's were known to degrade fairly rapidly. Thanks again!

Hondo

Keep in mind, these are completely un-regulated. As the voltage drops, the current drops, and the output drops.

Yes, this is completely non-regulated. So, as the input voltage drops, the input current does not increase, as it would do, in a regulated design.

You can make things look artificially regulated, if you use something like a Lithium cell (or even a NiMH), which have pretty flat discharge curves at these currents.

Regulation by definition means you are holding something in the output constant.

With regulation, something is held constant, like constant voltage or constant power, or constant current. Regulation results in a flat line with a drop off. There may be a very slight general curve to the line, depending on heatsinking and such, especially with non-current regulated designs.

There are a number of lights that claim to be regulated, but different cells result in different curves, different output levels with different cells, and other wierd things. Most of these are not even regulated, but just have a boost circuit to boost the voltage up. This product is a good example, as well as the Mini-Pro, which was ran wide open. Some folks have started using the term voltage multiplier for circuits like these. Some circuits do not even hold a constant multiplier.

Idleprocess, I took apart one of the ones that was degrading, and there is nothing wrong with the electronics, it is the LED dying from being blasted with too much current and voltage, which it cannot handle.

If the duty cycle was like an hour on, and four hours off, I would not expect that much of a slower degradation, there would be a very slight improvement- maybe a minute or two slower per hour of runtime.

I would expect one of the best things to do would be to buy one of those big solid aluminum hotlips (think they were like 30.00 dollars or so), modify it to accept the PR, and put it in a metal body light, like the 2D Maglite. The improvement would probably be significant at that point, but nothing spectacular.

Nerdengineer-
I have zero data on the earlier devices, nor the conditions the original device was specified at- which would influence the lifetimes that were specified.

Maybe somebody here on cpf tested them, and has the graphs of the test data, I actually don't follow the ArcMania/Michael Jordan stuff that close. Has anyone see it?

On this note, I will have to refer to the designer's expertise, ArcMania's a member here, so it would be best to defer these questions to the designer, who would be able to answer all of your questions and provide the test data for you.

Chimo-
As far as recommendations, I do not know how much each member values their investments. For some, money is hard to come by, and for others, it is easy to come by. As such, it makes it a difficult call. There are a number of new PR based devices on the market, but I do not have any of them for evaluation purposes.

I do know, for myself, I've made a very simple modification to the MagLED, to create a long runtime device, which lowers the drive current to the LED. This also serves to increase the efficiency of the LED a lot, and puts out an amazing amount of light, for the little amount of power it draws. In this device, the thermal resistance from the LED die, to the PR body is much lower, and as such, it would create an underdriven Luxeon, which would extend it's lifetime towards the 50,000 hour mark to only 30% loss in output (70% line on the graph above). Sorry, I'm not going to do a 5.7 year runtime test, you will need to rely upon the data provided by LumiLEDs.

Hondo · #**162** 09-16-2006, 10:37 AM

Thanks NewBie, the non-regulated behaviour is exactly what I observe. In fact, when the input voltage drops, input current also DROPS. NiMH's do give a fair psudo-regulation, at a lower brightness than lithiums, or fresh alkalines. What I was looking to do is take advantage of this situation by underdriving it a bit to a brightness still usefull to me. Since I am inputting less than 120 ma, and I am guessing that the input current at 2.8 volts in the test was much higher, I was hoping that I could expect a much better lifespan at lower brightness. While I would give up the near-Luxeon brightness, it is still good for 90% of what I reach for a light for, and I can grab a Luxeon light for the rest. Did you happen to record the current being delivered from the 2.8v source?

Hondo

NewBie · #**163** 09-16-2006, 09:45 PM

Okay, well it is time to put this one to rest.

The LED's light output has dropped by 20 times it's peak value in the begining.

Thus it has reached only 5% of its original light output level.

The final chart is below, as well as the actual data source file, if you'd like to download it and play with it a bit yourself.

Source data for download:
http://www.molalla.net/~leeper/smjled2.xls

Here you can see how the phosphor is now really dark over the die:

Contrast that to this:

I took a Royal Blue LED, to excite the phosphor, and notice how the phosphor is no longer excited by the blue light over the die area:

Here is the die lit up, notice how there is very little yellow being produced above and near the die, and it is darkish over the die:

Oh, and before I forget:
- The current draw by the bulb was 0.2A @ 2.8V, the same as it was at the beginning of the test.

And finally, we have the comparision beamshots against some low output flashlights I dug up, after ArcMania's SMJLED2 PR SS has had ~245 hours on it (fyi, it doesn't really show up in photos against the beamshots of the earlier lights now):

idleprocess · #**164** 09-16-2006, 10:02 PM

That phosphor-excitation shot says quite a bit...

NewBie · #**165** 09-16-2006, 10:34 PM

Yes it does idleprocess.

I added the direct shot of the excited LED, and also some beamshots, but I had to go dig around for some really low output lights, it wouldn't show up against the lights I used before, anymore.

NewBie · #**166** 09-16-2006, 10:58 PM

Well, I was asked to do a comparision against the Fenix P1, so it could be used as a reference from the early beamshots. I had to tweak the brightness adjustment in the photo program to make it show up. Yes, the P1 has an oversaturated beamshot now:

chevrofreak · #**167** 09-17-2006, 07:37 AM

Holy crap, only 10 days of life

vortechs · #**168** 09-17-2006, 09:27 AM

What if the PR bulb was re-designed to make the LED easily replaceable? If there were tiny sockets to hold the 5mm LED in the PR base (like the PR-base bulb adapter for the TerraLUX MiniStar2), it would be possible for the user to change the LED if it started to loose brightness over time, or if better LED's are introduced. The 4-die 5mm LED's seemed to work better in the PR bulbs, probably because they could better tolerate the high drive current.

What is the current that the LED is seeing from the driver in the bulb? From reading the posts in this thread it sounds like the bulb draws about 200mA at 2.8V from the power supply, but delivers less than that to the bulb due to the voltage increase and the losses in the converter.

Hondo · #**169** 09-17-2006, 11:00 AM

I kind of like the idea of a "bi-pin" plug-in. I think it was mentioned elsewhere about the reliability of such a thing, and being able to be knocked loose in a fall. Given the hosts these are used in, I would not be using these in a life-threatening situation, and this would be as reliable as a MiniMag SMJ mod.

The current at the LED I believe can be determined from Newbie's fourth graph in post #111. It appears to be just over 0.16 amps at 2.8 volts input. This portion of the graph is puzzling, as it appears the LED would be driven harder at 2.6 volts than 2.8 volts. Efficiency is clearly dropping above 2.5 volts, but the dropping output current seems like an extreme result. What I had trouble finding was the input current at the power source - 200 ma, thanks Newbie!

Quote:

Originally Posted by chevrofreak

Holy crap, only 10 days of life

Well, yes, when driving with 200 ma input current. It is only drawing about 100 ma from my AA NiMH's. I would be curious what this test looked like at that level of input. The "hottest" host I have one of these in is a 2-D Ray-O-Vac with alkalines resting at about 1.48 volts, and that measures 120 ma on the 10 amp scale. Fresh D's I recall giving in the high 100's, and I just tried a pair of E2 AA lithiums and got 200 ma. BTW, it was the brightest I have seen that unit, but no where NEAR enough brighter than one pulling 100 ma on fresh NiMH's to justify double the current, at a higher voltage, so well over double the power, and the resulting carnage to the LED. I think this all supports Newbie's findings on the huge drop in efficiency at high drive voltage/current. And at least for me, why I will use these with particularly tame batteries, and grab something other than a single 5 mm light when I need more light for a job.

For what some folks are looking for in these, I think I would like to see a circuit like Newbie's "tuned down" Mag LED drop-in, underdriving a 3W Luxeon, in a PR base that fits all of the hosts. At that level, the heat would not be an issue for the plastic lights, runtime would still be good, and the LED would last a loooong time.

Hondo

vortechs · #**170** 09-17-2006, 04:43 PM

Quote:

Originally Posted by Hondo

The current at the LED I believe can be determined from Newbie's fourth graph in post #111. It appears to be just over 0.16 amps at 2.8 volts input. This portion of the graph is puzzling, as it appears the LED would be driven harder at 2.6 volts than 2.8 volts. Efficiency is clearly dropping above 2.5 volts, but the dropping output current seems like an extreme result. What I had trouble finding was the input current at the power source - 200 ma, thanks Newbie!

I did think about post #111. However, the graphs in post #111 use a different LED than the SS SMJLED bulb uses, to separate the performance of the driver from the performance of the LED. I was not sure if the data in post #111 regarding the output current would be what the LED in the SS bulb would receive, since it has a higher Vf than the one used for those tests. I thought the data in post #111 was primarily just to show the efficiency of the driver circuit in the PR base.

If the efficiency of the driver circuit at 2.8V is less than 50% and the input current to the SS PR bulb is 200mA, and the Vf of the LED in the bulb is greater than 2.8V, it seems the current the LED would receive would be less than 100mA (rather than 160mA).

PlayboyJoeShmoe · #**171** 09-17-2006, 06:40 PM

Glad!!! that I got my seven "old" 4 die SMJLED PRs when I did!

If I wern't so fumble fingered I would turn down a M*gled or two!

NewBie · #**172** 09-18-2006, 08:14 PM

Okay, I decided to cross-section the LED, to see what happened. I was able to just take off the side of the internal reflector, and leave 99% of the leads intact, so it still operates fine:

Notice how the epoxy material over the die area burned black:

Here I underdrove the LED die, because it thrashed the photo by overloading the camera:

.

In this last photo, notice how the light is brighter at the edges and darker over the center of the die (where it is rather black now):

.

shiftd · #**173** 09-18-2006, 08:28 PM

Ewwww, that looks nasty
Thanks for the nice pictures

Looks like many tiny boogers

So, how did you cut it that small and precise?

Can we see comparison photo with a good led die?

Haz · #**174** 09-18-2006, 09:10 PM

great pics you have there, looks like something from National Geographics!,

thanks for your efforts NewBie in this detailed review.

chimo · #**175** 09-19-2006, 05:36 AM

Wow! Outstanding photos! Great job on the cut-away while maintaining function of the LED. Those energized photos are very telling as well wrt phosphor degradation.

What did you use to file down the LED?

Paul

NewBie · #**176** 09-19-2006, 06:41 AM

Thanks fellas.

Paul-
Elbow grease.

NewBie · #**177** 09-21-2006, 07:04 AM

And sandpaper.

chimo · #**178** 09-21-2006, 07:22 AM

I have to re-iterate how well that was done. I tried a quick and dirty sand-down and my results were not nearly as good. Although the sanding grooves look "large" in the photos, they are not! Great job on the macro photos as well. What did you use for the extreme magnification?

Paul

NewBie · #**179** 09-21-2006, 07:06 PM

I used a microscope.

ViReN · #**180** 10-02-2006, 03:32 AM

One More reason NOT to overdrive LED's ...

they ALL do it (except Fenix E0 and Peak LED Standard power lights)

Be it HDS Systems or Arc.... Dorcy or Streamlight... everyone...

Thanks to Fenix E0 & Peak LED Solutions for NOT overdriving LED's