Make Match's 3XP-G Mini without a lathe

Flashlight Modding and DIY Parts
1

At this point I must warn you that while I created something like Match's 3XP-G Mini without a lathe, I "cheated" by making a custom driver to fit in the smaller available space. So read on, but don't get yer hopes up too far...

Match's 3XP-G Mini really caught my eye, but that beautifully turned copper pill cup stood in my way. But then it struck me that it looked like a cap from 1/2" copper pipe. I grabbed the calipers and discovered the O.D of the cap is 18mm and the I.D. is just under 16mm. Surely I can find an 18mm drill, so if I could find a 16mm driver that would fit inside the cap then I could make this happen! Well I couldn't find one, at least none that I liked. So I built my own. More on that later. If anyone knows of a 16mm driver that will drive the XP-Gs to tears, please let us know!

I haven't had access to a lathe in 15 years, but I've recently gained access to a PCB milling machine! Basically a miniature CNC mill that can make PCBs in minutes, it is however a much more expensive tool than a lathe. (Please don't hate me just cause I'm blessed with it.)

I scoured Ebay and found an 18mm reamer, 2 for $14, boy is it crappy but worked well enough after a little dressing with a diamond hone. I clamped the minimag head between two pieces of wood and cut about 5mm into the threads. This Then I soldered a piece of pipe inside the cap to act as shoulder for the minimag battery tube to contact against. This reduced the I.D. of the cap to 14mm, mighty small for a driver but I was up to the challenge. I cut down the length of the cap with a pipe cutter, then filed and sanded it to be the right length to place the lower lip of the optics just above the head opening. The cap press fits into the 18mm bored head making good thermal contact (something tells me good heatsinking will be needed :) Drilled a hole for LED wires and soldered a ground wire to the inside of the cap, then mounted the driver and battery spring boards in the cap. I haven't potted anything just yet.

Now I wasn't quite sure how to address the area where the battery spring enters the tube. I didn't want any shorts here. Normally two protected AA lithium ions are too long for the minimag, but because the stock plastic piece is removed, there is actually a gap that needs filling. I don't yet know how big that gap is because I haven't received my flaming TrustFires! So I guessed and cut some washers and a battery contact disc from PCB material. Much to my amazement this contraption actually worked, mostly, with two Energizer lithiums! I say mostly because the energizers don't supply enough voltage for full regulation - my circuit needs about 6 volts. The LEDs were "PWMing" to about 30% power.

So here I sit waiting for the Flaming TFs......... ARRGG!!!

Reaming the MiniMag

Copper cap before and after

Cap and 14mm driver

Stack-up

Need to sand a little more off

Driver and battery spring boards

Battery contact disc and washer boards (before sanding off copper layers)

Everything tucked inside

Those are "pogo" contacts for the battery spring. I had some from a project at work so I used them, a spring would work as well.

The driver I built Uses a Texas Instruments TPS61500 boost converter IC. It drives the three XP-Gs in series at 1amp from an input of 6 to 8.4 volts. Efficiency is just above 91%. http://focus.ti.com/docs/prod/folders/print/tps61500.html

I'd love to post beam shots but guess what? I'm still waiting for the batteries!!!!

Enjoy, Crux

2

Nice work! That driver is what interests me the most. Feel like posting a schematic and parts list for it?

3

Welcome to BLF Crux! Nice first post and pics. great job!

4

Thats some SERIOUSLY beautiful work there!! your traces are beautiful and the work is neat *respect*

What I really like is the way your series connecting the emitters with a 1A boost driver. I parallel connected mine with a 3A driver, but im not so happy with that setup. Im interested in how you made the PCB. the traces are soo fine! Im still manually etching my hand drawn boards lol!

Btw, for those looking for circuit diagrams, have a look at the datasheet for the driver chip.

http://focus.ti.com/lit/ds/symlink/tps61500.pdf second page.

Where are you getting the Ti chips?

5

Welcome aboard!

I need more spare time! I've been meaning to build a high-speed spindle for my Sherline mill so I can mill PCBs too... The mill comes with two options, 2800 rpm and 10000rpm, and I have the 2800rpm one. I'm thinking of building a 50000rpm spindle. Need those high revs for making traces come out clean.

Have some spare pogo pins you can send this way? Or maybe some 60 degree V cutters?

Good work!

--Bushytails

6

Ingenuity at its finest. Well Done, Crux! It's always nice to see folks get something useful out of my posts. Also, as others have said...fantastic job on the driver!

7

Aloha and welcome to BLF Crux!

8

10 Watt boost driver for 3 x XP-G leds


Reference the Texas Instruments data sheet for the TPS61500 and Application Report SLVA419A for info.

There is a wealth of information in the data sheet; this circuit just uses the basics. This part can support Analog and PWM dimming, and can drive up to 8 series LEDs at 1 Amp from a 12v supply.

Keep in mind that the total LED voltage must be greater than the battery voltage for the circuit to work. The output is not short circuit protected. Definitely use protected cells with this driver.

Some notes on the circuit:

R1 and R2 provide about half the current sense feedback voltage. This means the current sense resistor R3 can be half the usual value and thereby reduce power loss in R3. It’s only 1/10w of power savings, but “waste not, want not” (For reference, an LED at 30mA is 1/10w). To save space, R1 and R2 can be omitted by connecting R3 to pin9, but R3 must change to 0.2 ohm.

R4 sets the switching frequency. 200k ohms = 500 kHz; Lower frequencies are more efficient, but require physically larger inductors… so it’s a trade off.

R5 and R6 set the over-voltage protection limit to about 13.5V. Because this is a boost converter an open circuit of the LEDs would cause the output voltage to increase to the point where it kills the regulator or the output capacitor or both. To live on the edge you may eliminate R5 and R6 to save space, just connect pin 11 to ground.

U1 TPS61500PWPR Digi-Key 296-24014-1-ND

C1, C2 10uF 16V X5R 1206 Digi-Key 445-4052-1-ND

C3, C5 0.47uF 0603 Digi-Key 445-5152-1-ND

D1 3A 40V Schottky Diode Digi-Key 641-1258-1-ND

R1, R5 100k ohms 0402 Digi-Key

R2 1000 ohms Digi-Key

R3 0.1 ohm 1206 Digi-Key 73L4R10JCT-ND

R4 200k ohms Digi-Key

R6 10k ohms Digi-Key

L1 15uH .050 ohm 3.2A Coilcraft MSS1038-153MLB or Digi-Key 732-1077-1-ND

The 1206, 0603, 0402 in the above list refer to the component’s length and width in hundredths of an inch. These are just what I used, many other sizes will work. As my eyes get older the parts get smaller – it’s a cruel world. And no, I can’t live without my microscope.

I hope you find this useful, I didn't fully optimize the circuit because it worked quite well when I first fired it up... I have mixed emotions when that happens. Its nice to sit back and enjoy the moment, but I learn much more from my failures than my successes!

-Crux

9

Thank you for that detailed write-up on the driver. I'm going to try this myself. I was also looking at this IC, and it looks like it may be good for multi-xml applications.

10

First off - thanks to all for the compliments. Normally when I show my latest creation to my wife and daughter they just roll their eyes... Does anyone else get that reception?

The PCB mill I use at work is a ProtoMat C30/S from LPKF. I think I've made more "after hours" boards than work related boards.

Okwchin - I like the idea of series connecting the LEDs too. There is less power loss in the catch diode at lower current than at higher current. Syncronous rectification addresses this problem, but those drivers are few in number.

Match - I looked at the data sheet for the TPS40211 - 47 pages! What application do you have in mind? Maybe a string of 5 XM-L @ 3amps? That's a mere 50 watts. (Just carry a Dewar flask of liquid nitrogen to cool it.) ;)

I'm quite impressed by the amount of light this MiniMatch makes, it's by far the brightest light I own. The beam is far from tight, but there is so much light that it doesn't matter. I can't post any meaningful beam shots because I don't own any "reference" lights. Well I do have a XP-G MagLite running at either 1 or 1.5 amps, but otherwise all my lights are one off mods or home brew plumbing nightmares. (3/4" copper pipe is a reoccurring theme in my collection)

11

Yup...all the time. Every mod I've shown her I get that same look that says "So..? "

But ya...I've been using my v1.0 minimatch for awhile now. It's toned down to @ 600 lumens, but I find it's a good compromise between output, heat, and battery life.

As far as the TPS40211, I've been thinking of making a multi-xml driver out of it. Problem is, it appears to only be available in lots of 1k. Currently the only workable one out there is the H6Flex from Taskled.

12

Not to mention the solderability of most of these LED drivers we want to use. Unless we can do reflow soldering on perfect PCBs... that puts my hand made hand soldering jobs out of contention. :(

How Long can you run the triple XP-G setup before it gets too warm. From an ambient of about 20 deg C, it takes mine 20 seconds to start feeling warm, by 40 seconds its blueshifting, and by 60-90 minutes its already quite warm. At this stage I turn it off.

Im driving mine off a 105C driver, but with a trustfire Flame 18650 so It should be providing currents in the 2A region due to the voltage sag, so about 700mA per emitter.

13

My MiniMatch with the custom TPS61500 driver begins to get warm at 20 seconds. As long as I switch hands periodically I can hold it till the batteries run dry. But it does get uncomfortable after about 5 minutes. I don't think I'll run this test too often as it does stress both the LEDs and driver. I really should do a temperature test on it, maybe this weekend if I can find the time. I don't have any light meters though...

The 61500 IC I used does have pins (HTSSOP package), but there is a big pad on the bottom that must get soldered for heatsinking. I do reflow soldering in a toaster oven :) - really. I outfitted it with a temperature controller and thermocouple. It works surprisingly well.

14

Hi Crux,

I am using TPS65100 Driver IC for LCD backlight powering. my LCD need backlight voltage 25.6V, 80mA input. I have verified lcd power requirements using variable supply directly connecting to LEDA (25.6V/80mA). LEDK - GND.

I feeding +5V/120mA current to TPS65100 using variable supply. with no lcd case, i can see the output voltage +26.5V. OVP fixed for this.

when i connect lcd, then tps651xx voltage fall to 21.5V and the input voltage also drops from +5V to 2.56V and Variable supply shows the current 119mA at CC mode, Vfb at ~19mV.

Can you please tell me where is going wrong. i am very new to back light supply with constant current source.

sorry for my poor English.

Regards,
shiv

15

Welcome to BLF Shiv,

Make no excuses about your English, it is nearly as good as mine.

I have not used this TPS61500 as a backlight driver, so I'm not sure if I can help. But if we do some mathematics on the numbers you posted we see that the backlight needs 25.6V x .08A = 2.048Watts.

And if the converter is 85% efficient then we need 2.048W / .85 = 2.4W from the input supply.

This means the current drawn from the 5V supply will be 2.4W / 5V = 0.48A.

If the 5V supply can't supply this much current (and probably a fair bit more during start-up) then the voltage will drop. Perhaps this is what you are seeing?

You state the variable supply goes into CC mode (119mA), try raising the CC limit to above 500mA and see how the circuit works.

Good luck!

-Crux

16

Wow…nice board!

If you have an Eagle PCB .brd you can upload it to OSHPark and have some pretty nice professionally made boards done up!

Excellent job!!

I wonder if this in conjunction with an ATtiny could make a 17mm board?

17

Maybe you could post a description/build of your oven in This thread .

18

The problem with most boost regulators is if the input voltage is low enough to generate a low moon mode then they aren’t capable of boosting to a high current efficiently. In other words, the input voltage determines how low the lowest mode can be. This ic looks like it can run pwm modes for lower modes or analog for true (non pwm) current regulation (not at the same time)but the non pwm regulation can’t be as low for that reason. Is 2200khz the maximum pwm input? Pardon me if I’ve completely misunderstood this.