January 01, 2004<br />
<br />		
<br />		
Got everything shrunk down, including a DC-DC synchronous boost switcher into a single cell MagLite AAA Solitare, with a Luxeond inside.  I chose to run the Luxeon at a lower current, at 220mA to gain runtime (0.72 Watts).  With a 700mAH AAA NiMH, I get +15 minutes of runtime at near constant brightness, before I got it thermally potted into the flashlight.  If you look closely, you can see the 0.090&quot; thick copper slug used to transfer heat from the Luxeon to the flashlight housing.  Once I get a rear window defogger repair kit, and get some time to make some final adjustments to the front focusing lens (not shown) I get you all some more information.<br />
<br />
<br />
<img src="http://www.molalla.net/~leeper/solilux.jpg" border="0" alt="" /><br />
<br />
<br />
<br />
Here is a shot without the head and lens, blazing in all it's glory:<br />
<br />
<img src="http://www.molalla.net/~leeper/soliraw.jpg" border="0" alt="" /><br />
<br />
<br />
Wow, this sucker is still running on my desk, after half an hour, on a Eveready Alkaline, wow. And it is only lukewarm, as well as the copper puck inside.
<br />
<br />	
		
Beam Shots:<br />
<br />
Mag AAA Solitare (1 cell) Luxeon with DC-DC Sychronous converter<br />

Mag 2AA <br />
Photon III<br />
<br />
All new batteries<br />
<br />
1 ft Photon III vs. Solitare Luxeon<br />
<img src="http://www.molalla.net/~leeper/solib1.jpg" border="0" alt="" /><br />
<br />
<br />
10 ft Photon III vs. Solitare Luxeon<br />
<img src="http://www.molalla.net/~leeper/solinar.jpg" border="0" alt="" /><br />
<br />
<br />
10 ft Photon III vs. Solitare Luxeon wide beam<br />
<img src="http://www.molalla.net/~leeper/soliwide.jpg" border="0" alt="" /><br />
<br />
<br />
Note the yucky yellow of the Mag bulb below<br />
<br />
1 ft Mag 2AA vs. Solitare Luxeon<br />
<img src="http://www.molalla.net/~leeper/magsol1.jpg" border="0" alt="" /><br />
<br />
<br />
10 ft Mag 2AA vs. Solitare Luxeon <br />
<img src="http://www.molalla.net/~leeper/mag_soli.jpg" border="0" alt="" /><br />
<br />
<br />
Geesh, it is still going after an hour, on just on AAA(moon mode).<br />
<br />
Never realized how large of an impact that high-efficiency DC-DC converters can have on run time, I'm used to the normal BadBoy/MadMax converters, but the high efficiency really helps.<br />
<br />
Guess it was definitely worth my time, spinning my own converter. </div>
<br />
<br />
<br />
<i>[CPF-Doug S}
Very nice!   Looking forward to the electrical data. </i><br />
<br />
<br />
<br />
Thanks for the comments guys, I think you made my day.<br />
<br />

You should have seen the light output at 650mA...This guy is only running at 220mA.<br />
<br />
FYI, one of the main reasons it is so bright is that thermal management helps to make a Luxeon run alot brighter, you can easily loose more than 40% of the light output of a Luxeon if you don't keep it cool.  Also, if you underdrive the Luxeon, it's lm/W goes up. Plus, if you don't load the cells as heavily, you end up getting more usable energy out of them.<br />
<br />
LOL guys, I don't know if I want to build these for folks as a hobby.  I still want to do a 2AA Mag 4W Luxeon mod, and a 20W Mag D-Cell Luxeon mod.  Possibly, after I get my feet wet with these projects, I'll consider it<-NOT!. <br />
<br />
<br />
<br />
Doug S, <br />
<br />

Before I slapped this one with the TPS61020 together, I got these numbers:<br />
<br />
Vin.....Iin....WattsIn..Vout..Iout..WattsOut.Effic  iency<br />
0.933 1.360 1.27.... 3.213 0.272 0.874....68.87%<br />
1.522 0.728 1.11.... 3.260 0.292 0.951....85.83%<br />
<br />
After these numbers, right before I potted it, I decided to go for run time so I lowered the output current to 0.22A<br />
<br />

After I lowered the output current, it went over 90% efficient, with this synchronous switcher.<br />
<br />
I can't wait until I get put the other one into a flashlight, the TPS61030, which will put me above 95% efficiency at 1A, in a 2AA Mag, and still over 90% when the battery voltage drops to 2.2V. And about 86% efficient at 1.6V, but I'll probably stop it 1.8V so that I don't run the risk of cell reversal.  The numbers are in the other thread in the electronics forum.<br />
<br />
Then comes the 20W design with multiple Synchronus Switchers, running +95% efficiency with larger inductors, for the D-cell Mod.  (Heheh, you ever buy a 1 1/2" copper rod? 23" of it cost 52 dollars, yikes)<br />
<br />
<br />
<br />
<i>[CPF-phoneguy]<br />
<br />
That must have been a nightmare to solder.<br />
<br />
Awesome job!!!!  I would also be interested in getting one if you decide to make more<br />
Bryan </i><br />
<br />
<br />
Nitemare to solder...humm...not too bad, but a challenge to do at home, I "haywired" it.<br />
<br />
Package is 3mm x 3mm x 0.9mm, pins are 0.3mm width on 0.5mm spacing. <br />
<br />
Here is the package diagram:<br />
TPS61020/24/25/27 96% Efficient Synchronous Boost Converter with 1.5-A Switch.<br />

<br />
Package Drawing:<br />
<a href="http://focus.ti.com/lit/ml/mpds117a/mpds117a.pdf" target="_blank">http://focus.ti.com/lit/ml/mpds117a/mpds117a.pdf</a><br />
<br />
Datasheet:<br />
<a href="http://www-s.ti.com/sc/ds/tps61020.pdf" target="_blank">http://www-s.ti.com/sc/ds/tps61020.pdf</a> <br />
<br />
<br />
<br />
Oh, found an old photo of the build I did to try the circuit out, looks messy since I kept changing things during testing, plus I had no magnification:<br />
<br />
<img src="http://www.molalla.net/~leeper/sminver.jpg" border="0" alt="" /><br />
<br />
<br />
<br />
Geesh, the thing has been running since about noon, its still putting out useful light after six hours. <br />
<br />
<br />
<br />
<i>[CPF-john_bud]<br />
<br />
Jarhead,<br />
<br />
Do you have a schematic and parts list that you are willing to share?  It looks like a fun project, maybe one that I could duplicate with instructions, but it ain't one that I could invent!<br />
<br />
Hats off on the great job!<br />
<br />
jb<br />
  </i>
<br />
<br />
<br />
Thanks for the comments.<br />
<br />
Heheh, this mod definitely ain't for the faint.<br />
<br />
Maybe I will spin a little board in the future.<br />
<br />
It is basically the same circuit right off the datasheet<br />
<a href="http://www-s.ti.com/sc/ds/tps61025.pdf" target="_blank">http://www-s.ti.com/sc/ds/tps61025.pdf</a><br />
<br />
With the following:<br />
<br />
L1 is a 4.7uh inductor.<br />
R3 is a 0603 package 1% 1M ohm<br />
R4 is a 0603 package 1% 191k ohm <br />
C1 is a 0805 10uf Ceramic<br />

C2 and C3 are combined for one single cap, the same value as C1<br />
<br />
The circuit had some issues starting up from 1.5V under heavy load, so I modified it a bit.<br />
<br />
R1 is the same as my R3<br />
R2 is is a 0603 package 1% 187k ohm <br />
I took R1 and connected it from LBI to Vout.<br />
I put a 0.1 uf 0603 across R2.<br />
This gives a delay time for LBO output to go high, once the output reaches the trip point determined by R1 and R2.<br />
<br />

For R5 I used my R3 part, then I took the LBO output and wired it to the Gate pin of a Fairchild FDS6064N7 MOSFET with a on resistance of less than 0.004 ohms.  The Source pins on the device go to ground.  The Drain pin (pad) goes to the minus side of the Luxeon, with the plus side of the Luxeon to Vo.<br />
<br />
I used 4 oz copper clad PCB to build this on, which is a very rare thickness of copper for PCBs.  I used it for thermal transfer for the chip.<br />
<br />
I then mounted the emitter on a 0.1" thick copper slug, and thermally potted everything, then thermally potted the whole thing in the Solitare.<br />
<br />
Next spin, I will probably do a current regulated setup, but this works just fine right now for me, and I will go to smaller parts, problably 0402 or 0201 for the resistors.<br />
<br />
One really nice advantage of this circuit, is when the batteries get to low to keep the output voltage up, it disconnects the output so the boost works up the voltage under no load, then it hits the threshold point, and the trip goes off, and reconnects the load.  This causes a flashing effect when the battery is near dead.<br />
<br />
I skipped the compensation cap across R3 (talked about for low ESR caps), it didn't seem to make any difference for my setup<i>(with the values, voltages, currents, layouts, in this converter, but it means the world in other ones.) </i>
<br />
<br />
<br />
<i>[CPF-BLU3_SHOCK]<br />
hey jarhead how did you get so good at soldering realing small stuff did you pratice a lot or something oh and i bet that you use one of those smd holders and those light magnifing glass combos. man you are really good.  oh and it is a very fun hobby but it can get expesive </i><br />
<br />
<br />
<br />
Humm, well, with the right equipment, such as a 300x stereoscope, I can solder to NAVAIR and NASA specs, including certification for internal repair on traces within multilayer boards, repair of the traces themselves.<br />
<br />

Here is a shot of something that would never pass certification (quality), small, but it is not tiny by any standard.  It is a solder job of L1 jumpers on an AthlonXP.<br />

<br />
<img src="http://www.molalla.net/~leeper/l1jump.jpg" border="0" alt="" /><br />
<br />
<br />
<br /> 
I just did a run time on this guy with a light meter, and noticed some interesting differences between Alkaline and NiMH batteries.  The Alkalines fade, where the NiMH just drops off like a cliff at the end.<br />
<br />
FYI at 12" with the current lens in it(which is nothing more than a single side convex acrylic focusing lens), it puts out 320 Foot-Candles, which according to Craig from the LED Museum equals 320,000 millicandela.  In other units, at 12" I get 3450 Lux.  Measurements are done with the Fluorescent light setting, which is probably the closest, other settings, such as Tungsten Filament yield a lot higher readings.<br />
<br />
Right now, the light stays at 3450 Lux for +30 minutes on one single 700 mAH NiMH Rayovac AAA, then drops rapidly.  <br />
<br />
I have hooked up the serial port, but I'm out of charged batteries, so I am recharging them on a 5 hour charger (all I have).  Then I'll be able to show runtime plots once they get charged. <br />
<br />
<br />
<br />
Here we go, runtime graph:<br />
<br />
<img src="http://www.molalla.net/~leeper/luxaaa.jpg" border="0" alt="" /><br />
<br />
<br />
<br /> 
*Keep in mind this is a 3W Luxeon running at only 0.72W*<br />
<br />
With an lens that I cannot fit into a AAA Soliare, I can hit +18,000 lux, fluorescent corrected (which results in lower numbers than tungsten settings), here is a shot trying to hold the emitter and lens with one hand and trying to aim and take a picture with the other hand, with the light meter on the bottom F means fluorescent corrected, 0 means add a zero to the reading, but I can't do everything well enough at the same time to get a photo with the optimium reading (at one meter):<br />
<br />
<img src="http://www.molalla.net/~leeper/1mlens.jpg" border="0" alt="" /><br />
<br />
<br />
Neighbor's yellow house, over 75ft away (probably close to 90 ft), same problem trying to hold everything centered, so I get lots of color fringing since I can't hold it well and shoot, I need four more hands:<br />
<br />
<img src="http://www.molalla.net/~leeper/75ftbeam.jpg" border="0" alt="" /><br />
<br />
<br />
Runtime is +30 minutes from a Rayovac AAA 700mAH battery<br />
<br />
<br />
<br />
<i>[CPF-b2p]<br />
so how do you make this? im lost.</i> <br />
<br />
Thanks for the comments.<br />
Heheh, this mod definitely ain't for the faint.<br />
Maybe I will spin a little board in the future.<br />
It is basically the same circuit right off the datasheet<br />
<a href="http://www-s.ti.com/sc/ds/tps61025.pdf" target="_blank">http://www-s.ti.com/sc/ds/tps61025.pdf</a><br />

With the following:<br />
L1 is a 4.7uh inductor.<br />
R3 is a 0603 package 1% 1M ohm<br />
****Fixed**** R4 is a 0603 package 1% 187k ohm (thanks for the catch, Doug_S)<br />
C1 is a 0805 10uf Ceramic<br />
C2 and C3 are combined for one single cap, the same value as C1<br />
The circuit had some issues starting up from 1.5V under heavy load, so I modified it a bit.<br />
R1 is the same as my R3<br />
***Fixed****R2 is is a 0603 package 1% 191k ohm (thanks for the catch, Doug_S)<br />

I took R1 and connected it from LBI to Vout.<br />
I put a 0.1 uf 0603 across R2.<br />
This gives a delay time for LBO output to go high, once the output reaches the trip point determined by R1 and R2.<br />
For R5 I used my R3 part, then I took the LBO output and wired it to the Gate pin of a Fairchild FDS6064N7 MOSFET with a on resistance of less than 0.004 ohms. The Source pins on the device go to ground. The Drain pin (pad) goes to the minus side of the Luxeon, with the plus side of the Luxeon to Vo.<br />
I used 4 oz copper clad PCB to build this on, which is a very rare thickness of copper for PCBs. I used it for thermal transfer for the chip.<br />
I then mounted the emitter on a 0.1&quot; copper slug, and thermally potted everything, then thermally potted the whole thing in the Solitare.<br />
Next spin, I will probably do a current regulated setup, but this works just fine right now for me, and I will go<br />
to smaller parts, problably 0402 or 0201 for the resistors.<br />

One really nice advantage of this circuit, is when the batteries get to low to keep the output voltage up, it disconnects the output so the boost works up the voltage under no load, then it hits the threshold point, and the trip goes off, and reconnects the load. This causes a flashing effect when the battery is near dead.<br />
I skipped the compensation cap across R3 (talked about for low ESR caps), it didn't seem to make any difference for my setup.
<br />
<br />
<br />
<i>[CPF-Doug S]<br />
In your description, I believe that you have reversed your values of R2 and R4. As described, LBO will never go high.</i>
<br />
<br />
Good Catch!  Vout flows into R1 to R2 to ground.  I put a 0.1 uf 0603 across R2., so R1+R2 and the 0.1uF form an RC for the delay.<br />
<br />
R2 = 191k<br />
R4= 187k<br />
<br />
Output voltage is determined by R3 1,000,000 ohms and R4 187,000 ohms.  The output reference is 0.5V.  0.5V/187,000 ohms = 2.674e-6 uA.  R3+R4= 1,187,000 ohms.  1,187,000 ohms * 2.674e-6 = 3.174 Volts.<br />
<br />
The trip point for LBI/LBO voltage is determined by R1 1,000,000 ohms and R2 191,000 ohms.  The output reference is 0.5V.  0.5V/191,000 ohms = 2.62 uA.  R1+R2= 1,191,000 ohms.  1,191,000 ohms * 2.62e-6 = 3.12 Volts<br />
<br />
So, what happens during startup, is the MOSFET keeps the load disconnected until the output voltage reaches 3.12V, which allows the regulator to come out of it's soft-start and pre-regulator phase, and keep the load disconnected.  Once this 3.12V is reached, the regulator is already out of soft-start, the LBI threshold exceeds the internal 0.5V reference, causing LBO to go high, which turns on the external MOSFET, and allowing current to flow through the Luxeon.<br />
<br />
Thanks *very much* for catching that typo Doug_S
<br />
<br />
<br />	
<i>[CPF-mike101]<br />
Hey, I got lost here. Is this Newbie and Jarhead the same person?</i>
<br />
<br />
LOL, yes, look at my tag line, they are both me.<br />
<br />
GreenLED,<br />
<br />
Ask LED-ASAP how it compares, I really don't have a clue, I was just hotrodding my solitare.<br />
<br />
<br />
<br />
<br />
<br />

<a href="http://www.molalla.net/~leeper/led.htm">Main site</a><br>
Mirror:<br>
<a href="http://jarhead.hobby-site.com/~jarhead/led.htm">Main site mirror</a><br>

<br>
<br>
<script src="http://www.google-analytics.com/urchin.js" type="text/javascript">
</script>
<script type="text/javascript">
_uacct = "UA-2114182-1";
urchinTracker();
</script>