U2 and 18650s redux

[Luna]

I know that a few here are using the u2 with 18650s but it has been reported in an earlier thread that they lose regulation in level 6 within no time. Well I'm currently trying a test. Using the unprotected LG 2400mah, I ran for 40mins at level6 (still had 6 distinct levels at the end) then had to turn it off to take the baby to the Doc for her 9mth checkup. Now that I'm back at the office I started timing again with the same cell. Now I'm 50min again into the test and still have all 6 levels!!!! This cell is on its first cycle and I've used it for the past few days on and off.

So right now I have 90mins+ runtime with the LG2400 and still have a difference noticable between level 5 and 6. I will keep running it but I'm amazed. Maybe the U2s cost (I got it for $225) will be offset by this amazing efficiency with the high current cell

I'll give an update. Man I wish I had a meter with me.

Update:
Ok just lost level 6. That is 95mins (40 + 55) with usage on and off before today and ~90min rest between sessions. Not bad!!!


Opps, I typed 2500 and meant 2400mah sorry

 

[pyro]

have you tested the voltage of the cell when you finished the test?

 

[cy]

My U2 with 18650 has all levels for a long runtime

 

[Luna]

Being at the office I don't have a DMM with me so I haven't tested the voltage. I wish I had one with me.

BTW It just hit moonlight mode (level 1) 12min after losing level 6. I've shut it off now.

 

[SilverFox]

Hello Luna,

The U2 does not run "in regulation" at level 6 with a Li-Ion cell. It will show 6 levels as you have pointed out, but the brightness will be gradually falling off while in level 6. If you look carefully you will notice that the difference in brightness between level 5 and 6 (after an hour or so) is less than the difference between level 4 and 5.

If you need full brightness for a long period of time, use CR-123's. If you run at reduced levels and only occasionally need level 6, the Li-Ion cells will work fine.

I use the 18650 cells in my U2. I charge them up every couple of days and have been very pleased with its performance.

Tom

 

[Luna]

[ QUOTE ]
The U2 does not run "in regulation" at level 6 with a Li-Ion cell.

[/ QUOTE ]

Ok now for a jargon question/explaination of point of view .

I've read this before from you but I have problem with the statement. I tend to take the meaning of "in regulation" meaning the reg circuit can produce the current requirements needed for the minimum output. I'd guess this is where the 80lumen comes from compared to the 100lumen spec(best vs worst).

Now with a big assumption of PK designing at an average spec of 700ma and 6.84vf, this means as long as the battery can source the approx 5w+loss needed, the light is 'in regulation'

I really don't see why the voltage is being considered in the equation, the LED is a constant current device.

But yes, you get more brightness as you increase the power0(which is quite obvious with my A19 testing). I tend to take in regulation as a statement that the minimum current requirements are being satisfied for the level given. So if I can switch between level 5 and 6 and see a distinct output difference (such as visible in kj's runtime charts of 5 vs 6) then I consider it in reg. Level 6 is odd though in that it doesn't portray a straight line/brickwall output like the other levels do. I really think level 6 is programmed differently than the rest that are truly regulated.

Talk at yah later.

 

[SilverFox]

Hello Luna,

Ahh, the many definitions of regulation... /ubbthreads/images/graemlins/grin.gif

I was defining regulation as constant light (lux) output.

If you review KJ's plots you will notice that with the CR-123 cells the light output (lux) at level 6 is fairly constant (it actually drops around 5.5%) and is brighter overall. With the Pila cell, the level 6 light output dims over time (dropping about 14.3%) indicating that the light output (lux) is not "in regulation."

I do not believe level 6 is programmed any differently than the other levels. The CR-123's just can't maintain voltage at the 1.5 amp draw.

Tom

 

[cy]

yep, my U2 draws 1469 milliamps from a 18650 cell

 

[Luna]

I would be very surprised if SF went the voltage regulation route, with the Vf variation of the bins you would have too much flux variation in output across the production line. Chances are, the emitter will be underdriven slightly to decrease warrantee costs. Constant current regulation is the best strategy. Regulation of lum flux will typically correlate well with current regulation of LEDs,so this is another plus.

Obviously the Cr-123 wouldn't have to source 1.5amps @ 6v, it should be pulling max of one amp at that voltage (probably never more than 1C ever). The protection circuit would also have to limit the draw at the 1400ma level to prevent the depleted cr123s from offering up a sacrific.

As for level 6, it appears to allow for boost without any buck, hence the decreasing light output vs flat. Seems to be an inductorless dc-dc using CPO. Has anyone ripped one apart to look at the ics?

 

[wasBlinded]

My U2s all "try" to draw 6 watts from the power source on level 6 - this is very consistent between 4.2 and 6.8 volts. Once voltage begins to drop below 4.2v, the boost circuit will not draw enough current to maintain a 6 watt draw, and the light begins to dim. This is with a regulated power supply, so available current is not the problem - below 4.2v the boost circuit can no longer supply the LED with its intended current on level 6.

Thats why a U2 with a Li-ion like the 18650 will gradually lose brightness on level 6. The Li-ion will be at 4.2 volts very briefly before heading for its more stable voltage of 3.7 volts, and the light will gradually dim.

There is no "buck" ability in the U2, once input voltage exceeds 6.8 volts, current draw rises rapidly.

 

[KevinL]

My apologies if I gave any of you the wrong impression. I did post early on that my U2 seemed to COMPLETELY lose level 6 after 10-15 minutes and only give level 5, but as I later discovered, that was due to inferior cells (I know, I know, 80 octane leaded gas in a Ferrari syndrome).. I plead guilty not having my teleporter gateway operational, thus being forced to wait for air mail and 4sevens' 18650s /ubbthreads/images/graemlins/wink.gif

Once I got the gray-jacketed LG 2400s from him, I've never been happier with the U2. Level 6 is available bright and white for a nice long time. /ubbthreads/images/graemlins/smile.gif

 

[Luna]

wasBlinded, level 6 obviously has no buck but all other levels appear to due to the constant output per kjs.

Also did you verify your tests with a lux meter? Once again assuming the circuit is a charge put, they tend to be more efficient as the voltage decreases, well to a point then it tanks. So I guess what I am saying is that the power requirements will decrease as the voltage decreases and the efficiency increases. So that 6watt draw should decrease to around 5.2watts as the voltage of the cell is in the 3.5v range

Given I don't want to take the light apart to test the circuit, here is my emperical take.


Look at the LTC3216 for example: LTC2116 DC-DC

At with a 200ma load at 3.2 volts it is 92% effcient but at 4.2v it drops to approx 72%. At some point you would want to bypass it and go dd. The drive has to produce 4.8watts+ loses to give the LED 100% rated power. So at 92% we have 5.2watts and at 72% we have 6.7watts.

Due to the to the safety precautions incorporated in the light that protect li-ions, I'd bet the U2 has a similar chip in it. However the same light has to work with Cr123 so a limit on the draw had to be inplace. This of course places a limit on the draw of the li/ion cell reducing its ability to source the current of needed at around 3.5volts.

I can see a drop off in brightness over time in level 6, but with a fresh cell, I do not have a noticable difference. I'd venture to guess the Lg2400s do not have the voltage sag that the Pilas 168s do. They are of less capacity and have to drive a protection IC. Also we can't establish the level6 output benchmark as being the spike you get with a fresh set of CR123s, but I will tell you that the output with the LG2400 is comparable to what I get with CR123s plus I have a greater runtime.

I wish KJ would do a level 6 graph using a set of these LG 2400mah cells. /ubbthreads/images/graemlins/wink.gif

Kevin, I concur. I don't feel as if I'm compromising using the LGs. Coming from a fellow flashaholic, that carries considerable weight /ubbthreads/images/graemlins/smile.gif

-Craig

 

[wasBlinded]

OK, I have some numbers for you to contemplate. This is with the head removed from the body and cooled with a muffin fan. Power supplied with a BK1743 bench supply, voltage measured at the head with a Fluke Scopemeter 123, and relative lux readings taken with a Extech 401020 Light adapter attached to a Fluke 88 multimeter. Current readings taken directly from the BK1743 digital display, which I have verified to be accurate to within 3% and even more precise than that. Also, the head has been "broken in" by 24 hours of continuous Level 6 on the power supply. This is important, since the Vf of the LED often drops a few tenths in the first 24 hours of life. Lux readings are given as a fraction of the brightest the light head will deliver in regulation on Level 6.
<ul type="square">
Volts Amps Lux (relative)
3.40 1.30 .819
3.50 1.38 .865
3.60 1.42 .897
3.70 1.44 .924
3.80 1.46 .943
3.90 1.47 .965
4.00 1.48 .986
4.10 1.48 1.00
4.20 1.43 1.00
4.30 1.38 .997
[/list]

Beyond 4.3 volts, lux readings are very stable and the volts x amps product is very stable at around 5.85 until 6.8 volts.

So this light does very well on an 18650, giving 92% or more of maximum lumens throughout the major life of the cell (down to 3.7 volts).

 

[Luna]

wasBlinded, man thanks for the numbers. They are much appreciated.

One question, did you start the test with lower voltage or higher? If higher to lower then I wonder if a form of thermal management by the reg came into play.

Let me do a bit of thinking. Thanks again.

-Craig