Introduction: 2200 Lumen Air Cooled Wooden Flashlight

About: Learn something new everyday


"Light dawns and the darkness is swept away" - James Burke

I am calling this flashlight Maxwell Mark I named after James Clark Maxwell, who did significant work on improving the idea first proposed by Michael Faraday that light was an electromagnetic wave. Faraday didn't have the mathematical prowess to prove his intuition but Maxwell did. Maxwell's equations on electromagnetism is considered one of the most beautiful equations in physics.

Step 1: Brief History of LED

This project came as an idea to me when I was researching into LED’s. LEDs are a recent technology. The traditional light bulb still uses a physical process that Edison invented in 1847 called incandescence which means just heating up a wire with high resistance in near vacuum until it glowed incandescently and therefore used a huge amount of energy and was very inefficient. Red LEDs were invented by Nick Holonyak, a scientist at GE using a process called Electroluminescence. Holonyak was a doctoral student of John Bardeen who co-invented the transistor. Rather than using incandesce to generate light, Holonyak’s LEDs used a P-N Junction in a special transistor (called a Diode) by doping the material to excite electrons until it emitted light. Hence the name Light Emitting Diode (LED). Use of red LEDs became ubiquitous since 1960. It was in clocks, cars, watches and many consumer devices.

It was only in 1993, Japanese scientist Shuji Nakamura (who shared the Nobel Prize in Physics in 2014) developed the first brilliant blue LED and a very efficient LED in the green spectrum range. In 1995, by combining Red, Green and Blue, a brilliant white light from luminescence conversion was introduced to the market. In 2006, white LEDs with 100 lumens/watt was introduced to the world. The LED I am going to use is a CREE XML T6 (available since 2011) produces approximately 158 lumens/watt, which CREE claims is the industry's highest performance, single-die white lighting class LED. There is a global initiative to replace all incandescent bulbs to white LEDs. A CREE LED light bulb with an equivalent brightness of a 60W bulb uses around 9W. My flashlight uses around 6W.

Step 2: Background of Construction

I wanted to create a flashlight completely from scratch. Buy all the components separately and assemble it all together. I wanted to build a flashlight that properly crossed 1000lm mark and lasted longer than most of the commercial ones.

It is difficult to find a usable 1000lm flashlight on the market that uses alkaline batteries. Most are either lower than 900lm or have short battery lives. The well-respected flashlight company Maglite only has a full size flashlight that cranks only up to 694 lumens. Some of the customer reviews from amazon about the Ultrafire 1000lm says they don't really deliver 1000 lumens. There are a few reasons why. One is heat management. Prolonged use of 1000lm requires very capable heat sinks. Even in this model I am using, I actually have to use a fan with the heat sink to maximize heat dissipation. A chip within the LED module lowers the available light output as the junction temperature increases. This is done to protect the LED from overheating. Second reason is alkaline batteries are not very good at giving huge amount of current. If you use multiple AAA batteries to power a flashlight, initially you get brightness but within half an hour it goes dim. Alkaline battery life expectancy relatively is very low. However, The Ultrafire flashlights that use lithium-ion batteries seems to do fine.

When you are working from barebones, there are a lot of considerations to make. First of all, my priority to push the LED chip to its limit. The Cree T6 has a maximum current rate at 3A. That’s a lot of current. That's why most manufacturers of the 1000lm mark prefer to use multiple lithium batteries in parallel. Lithium batteries have a high energy density and are capable of discharging huge amounts of current in a short amount of time. (Also called their high discharge and burst rate).

Step 3: Primary LED Unit

The first part of the primary LED unit is the CREE T6 bead. Assembling this to the unit was the hardest part of all. I bought the CREE XML- T6 bead from eBay. (See figure ). This bead must be handled with care. The plastic lens of the bead can easily come off. During the assembly, I accidentally touched my first bead a bit too hard and the lens came off and the LED had to be thrown away.

If you are testing the LED, do not light up the bead without a heatsink for more than a few seconds. The T6 is a performance LED and will heat up very quickly. The bead can easily over 90 degrees in a few seconds. The T6 is designed to take 5V in reverse voltage, so if you are testing the LED and get the contacts in the wrong way, it's fine.

I am using a 21mm reflector collimator lens with 15 degree lens degree. So the beam is going to be extremely focused and will go farther i.e. it will have a long beam distance. You should first position the collimator lens first on top of the bead and then solder the leads. The contact surface for the leads are tiny. You have to use a third hand and make sure there the soldering is seamless. If not, the solder will not stick to the smooth contacts. It took me over a week to get it right.

Second part is the heat sink. I found a really wonderful way to glue LEDs to heatsinks. Using solder to fix the bead to the heatsink is almost impossible. As you apply soldering iron to a heatsink, it absorbs the heat energy and renders the iron useless. The best way is to mix thermal paste and super glue together. 70% thermal paste and 30% quick glue seems to work for me. Remember, the super glue is a ticking time bomb. I recommend you put the bead on top of the heatsink within 10 seconds of mixing the paste. Apply light pressure on the assembly.

The third part of the unit is the fan that's going to dissipate the heat from the heatsink. As I will come to later in the stress test, the heatsink is pretty good in itself to dissipate the heat itself. But adding the fan adds the longevity of the LED. CREE LED’s brightness is dependent on the junction temperature (i.e. temp of heat sink). (Figure) I used superglue for this as well.

The LED unit needs to be set on top of stand so that air can circulate freely. I used thin dowels and shaped them until the lens was leveled and pointed directly ahead.

Step 4: LED Driver

Cree comes with custom LED drivers that fit nicely with the heatsink and have modes for High beam, low beam and flash but they have their limitations. They seemed to be designed for lithium batteries in mind .i.e. they usually have input 3.7V. I am not too fond of them.

In the spirit of overclocking, I am going to use a driver of my choice, that's basically a DC-DC step down supply circuit with the LM2596S regulator from 5V to 3.7V. The transformer module containing the regulator has two parts that are doing the heavy lifting. The regulator itself, which should not exceed 150 degrees Celsius and the inductor which temporarily stores voltage and lowers it (also should not exceed 150 degree). Out of the box, with my LED setup the transformer module with the LM2595S regulator chip was running close to 90 degrees when it was powering the LEDs. So, I placed two tiny heatsinks on both the inductor and the regulator to dissipate the heat and lower the temperature. (Figure)

Since I am custom designing this LED setup, I have to do a stress test on my LED flashlight driver to monitor several parameters. See stress test at bottom.

They are sort of the active components in my circuit and I want to make sure that my configuration is absolutely safe under any conditions. I did a 30 minutes continuous duty stress test before I give this flashlight away.

In my stress test for 30 minutes, the internal components were around 40 degrees. That's acceptable.

The fan has an XL6009 step up transformer that boosts 5V to 12V.

Step 5: My Battery Epiphany. (light Bulb!)

Now came the battery choice. I was planning to use C batteries which are 1.5V (they are smaller than D) and quite robust. The Energizer MN1400 has 7AH. I got 12 C batteries in series parallel connection with hookup wires and aluminum foil and did a trial run. (See picture and excel file). I was very disappointed. The problem is that whatever I did, I was only getting 0.7A from the batteries. The battery life plummeted with each use. Then there was the problem of replacing the dead batteries. I was pondering over this for a few days and explained the problem to my brother. He said “Why don't you use the power bank?”. A lightbulb went off my head. In this case an LED went off in my head. Pardon the puns. I did consider my 13000mAh Anker Astro E4 (also called power bank) before, but my initial concern with it was it always needed a button to be pressed for it to turn on. Another reason was that a USB only gave out 2A. I expected the C batteries to give out 3A. But it didn't turn out that way.

The epiphany came because I remembered the several technologies that was featured in Anker power bank. Firstly the Anker sends out a pulse of current several times a second to the USB ports to check if a device is plugged in. Which means as long as I have a switch that gives a solid 0 or 1 signal to the power bank, it will work. If the switch gives 1, the Anker powers the device. Once the switch gives 0 (off), Anker stays on for 30 seconds and then automatically turns off. Then there are other features like voltage protection circuit, smart current regulation, battery temperature monitoring and several other features under Anker’s PowerIQ technology. Brilliant.

Step 6: Wooden Chassis and Assembly

My main interest is in electronics and seeing how theory meets practice. I haven't taken a woodworking course in high school. I wish I did. So my finish is not perfect. But I can get by.

I started by making two thin cylinders with 2.5 inch diameter. This is roughly the diameter of the Anker power bank. The rest of the components are smaller. One of the cylinder is going to house the primary LED unit. A hole has to be made in the center of the cylinder for the wires to go inside. Then I left some space for the transformers and started nailing 0.5 inch dowels to the two cylinders on opposite sides and built the skeleton of the chassis. I filled the space with the dowels and went half way through the cylinder and placed all the components inside. Here is a picture with the transformers and Anker wired inside the chassis (Picture 4).

I had to cut off a slice for the power bank and made sure to superglue the two USB cables on the power bank so that if someone yanks on the USB, It won’t give stress to the components inside. The USB pinout diagrams can be searched online. USB has red, black, white and green wires. As usual, when splicing the red wire is positive and black wire is ground. The other two wires are data and can be removed.

The 0.5 inch dowels are expensive. I am attaching a diagram I used to calculate the number of dowels needed to cover the cylinder, so you can plan how many dowels you will need. (picture)

The full schematic of wiring up the components are given here. (See picture and Multisim file)

Step 7: Full Duty Stress Test (30 Min)

For the stress test, I attached a k-type thermocouple thermometer to the heatsink in the primary LED unit. Then I used a timer and logged down the temperature values. A trend line was produced from the data of 30 minutes and was extrapolated to 4 hours. As you can see from the graph (picture), the best fit to the values was a logarithmic curve. The equation has a good R squared value (it was a good fit) and as you can see, if you use the flashlight for a long period of time, the predicted temperature values plateaus around 70 degrees. Which is not bad, considering the passive heatsink is only getting rid of heat through convection and radiation.As you can see from the CREE datasheet picture, at 70 degrees even without the fan, the LED will give more than 90% of its capacity.

As a safe measure and to ensure longevity, I would recommend using the flashlight without fan for about 5 minutes and any use after 5 minutes should be with the fan turned on.

The Inductor and LM2596S at the end had 34.5 degrees and 40.1 degrees respectively. As you turn on the flashlight, the current going into the LED (measured by an ammeter) is around 1.38A and as time progresses and the ions in the battery circulate, the current slowly rises to exactly 1.6A. So the longer you use it, the brighter it gets. The fan uses only 60mA. So if the fan is turned on, the maximum current going to the LED is 1.54A.

Step 8: Conclusion

I think this is a very powerful flashlight. Anything that that is this bright and can last this long is very hard find on the market. At 30 cm, the flashlight gives out 2580 lux at peak. Using an online calculator, that turns out to be 2200 lumens!!! The Maxwell I flashlight (including the battery) weighs 1.1kg.

With a rate of 1.7A and with a 13000mAh battery, the flashlight theoretically should last about 5 hours. You can also charge your phone while using the flashlight. It's still a power bank. This flashlight can still charge an iPhone with half the time as a laptop USB. I haven't found anything like it yet anywhere.

For example, take this military grade TM15 Nitecore flashlight. In turbo mode at 2450 lumens, TM15 lasts only 1 hour and it costs $225.

For Maxwell Mark II, I want to figure out a way to boost the current from 1.6A to 3A.(See picture from datasheet). I am not sure what's limiting the Anker E4 from giving out more Amperes. It should in theory be giving 3A in total from both USB ports. Right now I am getting 1.1A from one port and 0.5A from another with a total 1.6A. That's only using 40% of available power.

Additional Note: I found out that according to the USB Charging standards, the maximum allowed current draw for a USB port is 1.5A. If you need to draw more current, the data lines of the USB has to used with a dedicated charging port controller like TPS2513A-Q1.The charging port controller handshakes with the Anker powerbank using the data lines to send up to 2A from each port. That's something I will implement in Maxwell Mark 2. More to do, more to learn.