5mm AA Simple Build

I would like to attempt my own flashlight build. In a nutshell I want to make a potted, reliable, waterproof (IPX8ish), twisty switch, AA powered, 1 mode low output, floody/candle light. I have limited funds and no experience so I think a 5mm LED would be appropriate, I was considering the cree 5mm leds which I have created a separate thread for (click). I'm not sure when I will have time/funds to do this but I definitely want to do it. Here is what I am looking at/what I could use some help with.

1. Electronics: I want this thing to be 1 mode twisty, between 1-10 lumens and be pretty efficient. I'd like power outage candle/long runtimes type deal. I need help here. I can solder but I don't know what to put between the led and battery, etc.
2. LED: (addressed in another thread, feel free to post here). Preferences in order: long life quality led, durability, efficiency, good beam, warm tint.
3. Body/twist switch: I think I can handle this. I'm pretty sure I can rig something up going to the hardware store a few times. It will probably be PVC pipe or something equally as hideous but it will be tough enough and meet my criteria.
4. Potting: devcon 2 ton epoxy.
5. Power: 1xAA Eneloop/alkaline is preferred. If efficiency/build simplicity can be increased by going to 2 or 3xAA I'll go that route.

Thanks in advance.

TyJo said:

I would like to attempt my own flashlight build. In a nutshell I want to make a potted, reliable, waterproof (IPX8ish), twisty switch, AA powered, 1 mode low output, floody/candle light.I have limited funds and no experience so I think a 5mm LED would be appropriate, I was considering the cree 5mm leds which I have created a separate thread for (click). I'm not sure when I will have time/funds to do this but I definitely want to do it. Here is what I am looking at/what I could use some help with.

1. Electronics: I want this thing to be 1 mode twisty, between 1-10 lumens and be pretty efficient. I'd like power outage candle/long runtimes type deal. I need help here. I can solder but I don't know what to put between the led and battery, etc.
2. LED: (addressed in another thread, feel free to post here). Preferences in order: long life quality led, durability, efficiency, good beam, warm tint.
3. Body/twist switch: I think I can handle this. I'm pretty sure I can rig something up going to the hardware store a few times. It will probably be PVC pipe or something equally as hideous but it will be tough enough and meet my criteria.
4. Potting: devcon 2 ton epoxy.
5. Power: 1xAA Eneloop/alkaline is preferred. If efficiency/build simplicity can be increased by going to 2 or 3xAA I'll go that route.

Thanks in advance.

Click to expand...

Look up the Joule Thief threads here and use that circuit. You'll get about 70% efficiency, working down to 0.7v. Note that this can damage NiMHs, although eneloops seem to be resistant to such damage. Pot that in some body where screwing something in completes the circuit and you're set.

AnAppleSnail said:

Look up the Joule Thief threads here and use that circuit. You'll get about 70% efficiency, working down to 0.7v. Note that this can damage NiMHs, although eneloops seem to be resistant to such damage. Pot that in some body where screwing something in completes the circuit and you're set.

Click to expand...

Thanks for the reply. Any idea what efficiency the Fenix E01 gets? Is it possible to do a 2 or 3 AA direct drive to improve efficiency?

I'm planning a similar build with a driver taken from cheap Chinese 1AA "police" flashlight. LED is very easily swapped since it's not potted. Initial current is 30mA with a significant drop afterwards due to voltage sag (no regulation IMO). I'm not sure about its efficiency or runtimes on alkaline/NiMH, still have to test it. There must be some quality drivers (based on specifically built IC's, using precisely wound toroids etc.) available online. As for E01 efficiency, look here.

Cheers

A white LED requires 3.2 to 3.6V. A single alkaline (1.5V) or NiMH (1.2V) isn't going to light it. You need a driver circuit to boost the voltage. My favorite solution is a ZXLD381 chip and an inductor. It's very simple, but the drawback to simple is you can only run your LED at 15mA (underdriving it) because the ZXLD sends pulses of current that will cause the LED to fade prematurely. If you're willing to add a diode and a capacitor to smooth out those spikes, you can then push 20 to 40mA depending how how much brightness/lifetime you are interested in.

Here are some examples of ZXLD381 builds:
http://www.candlepowerforums.com/vb/showthread.php?226903-ZXLD381-3Cuts-DIY-1.5v-Driver
http://www.candlepowerforums.com/vb...boost-driver&p=3546854&viewfull=1#post3546854

Or, the aforementioned Joule Thief driver circuit is cool too.

You can direct-drive your LED with 3 AAA or AA cells and a resistor, but it starts out bright and ends up dimmer as the batteries discharge (the Joule Thief does this too, to some extent). Resistor direct-drive is very efficient with battery life, though I would not use it with NiMH cells as one cell tends to be damaged when it is overdischarged near 0V before the light goes out -- I recommend alkalines only for 3x cell resistor direct-drive circuits. You can Google online LED resistor calculators that will tell you what value of resistor to use.

You mention floody. Usually the high quality, really bright 5mm LEDs (Nichia, Cree) have narrow 15° throwy beams to concentrate the light in a little hotspot. You can make them floody by rubbing them with steel wool or fine sandpaper. Another alternative is an "100mA 5mm LED" which is theoretically 5x brighter than a normal 20mA 5mm LED. I wouldn't try to drive them at 100mA though, as I feel they are massively over-rated. I'd do 40-50mA tops, but you should still come out ahead in brightness and still have some measure of longevity.

For resistor direct drive I prefer 4AA. Your batteries need to provide 3.6V to drive the LED.
3AA would be ineffective when the voltage drops to 1.2V per battery. But you may get away without a resistor if you can find an LED with a high enough Vf.
4AA would be ineffective when the voltage drops to 0.9V per battery. You get more total juice from 1.5V-0.9V than 1.5V-1.2V out of the battery. But you need the resistor for sure.

For calculating the resistor here are a few sites:
http://led.linear1.org/led.wiz
http://www.pcboard.ca/kits/led_notes/
http://ledcalc.com/

For applications with extended runs like a power outage light run the LED at 1/2 the rated current. Use 2 LED/resistor circuits running at 10 mA each rather than 1 LED/resistor circuit running at 20 mA.

One question.
Have you considered a more rugged LED like an SSC-P4 with the solder tabs. If you keep the current down you would still have low light levels but the LED won't die as easily as a 5mm.

Available as bare emitter or mounted on stars from that unmentionable Hong Kong site. Also available from Newark in the states.

LEDninja said:

Have you considered a more rugged LED like an SSC-P4 with the solder tabs. If you keep the current down you would still have low light levels but the LED won't die as easily as a 5mm.

Click to expand...

I agree, the 5mm's primary advantage is packaging: drill a 5mm hole and jam it in. If you have some kind of substrate to mount a power LED to the surface of, not only do you get the option of higher total brightness, but you also typically get better lumens per watt -- even if you keep it dim.

For something like a homemade PVC light, you'll need to get a little creative with mounting it.

I think you need to first decide how many lumens output you want, between 1-10 can make a difference in both runtime and circuit efficiency and design. An output around 1 lumen or so can easily be had without a boost circuit direct driving off 2 batteries with fresh alkalines being brighter than nimh cells but I have used a 5mm 110 degree cree LED off nimh direct drive and it runs for weeks at outputs starting around a few lumens dropping down to a fraction of a lumen over time. If you want brighter than direct drive off 2 cells but still less than 10 lumens a resistored 3-4 cell approach is the easiest to manage and has the advantage of being able to adjust the output by either swapping out the resistor or putting in a variable one. Low lumen output LED lights have rather long runtimes such that higher efficiency is often less important than getting a good quality light engine setup and ease of use.

tylernt said:

I agree, the 5mm's primary advantage is packaging: drill a 5mm hole and jam it in. If you have some kind of substrate to mount a power LED to the surface of, not only do you get the option of higher total brightness, but you also typically get better lumens per watt -- even if you keep it dim.

For something like a homemade PVC light, you'll need to get a little creative with mounting it.

Click to expand...

An SSC-P4 on a star has 2 advantages. The star acts as a heatsink and at these low levels does not require attaching to a metal flashlight body. Soldering to the tabs on the star is further than the tabs on the LED itself further from the delicate die. Also with buying the LED pre-mounted on a star, your choice of LED increases. Cree XP-C, XP-E, XP-G, XM-L though anything bigger than an XP-C is overkill. Get a proper star board instead of a round board and you can simply screw the star down to any flat surface.

Thanks for all of the replies. I was considering 5mm LEDs because they are cheap and I could have spares for more lights or if I screw one up. I didn't want to make a housing/lens assembly to protect the LEDs (due to cost and complexity), I figured 5mm did not need this protection. I want to pot the light so it is very durable. I only want the light to be used as a light/candle in pitch black conditions, so I guess 1 or 2 lumens would be ideal. The following sounds good:

Lynx_Arc said:

I have used a 5mm 110 degree cree LED off nimh direct drive and it runs for weeks at outputs starting around a few lumens dropping down to a fraction of a lumen over time.

Click to expand...

I would be willing to add batteries, resistors, etc. to gain runtime or any sort of (psuedo)regulation. My main concerns are durability, efficiency, and simplicity of the build. I won't be able to do ZXLD381 builds, those seemed to complex. I guess the less batteries the better, if possible. I'm not too concerned about damaging Eneloops as they seem pretty tough, if I go multicell configuration I will be more cautious. These cells won't be put in other lights and will be charged frequently to avoid over discharge.

I'm not sure when I will be able to get to this (time and money are limited), but I want to do it right when I do get to it. Thanks for all of the replies and quality information (and please keep it coming if you have the time).