How to Make an Improbably Bright LED Task Lamp

Fiat Lux!  The rallying cry of every scientist, engineer, handyman, or anyone that has to work on small fiddly bits.  You can never have enough light when working on some things.

I wanted a replacement for my 250 watt halogen--a lamp which, while sufficiently bright, was so hot that it was impossible to work with for any length of time.  Enter the world of high-powered LEDs.

In this Instructable I'll show you how to construct a 50-watt LED lamp.  50 watts doesn't sound like a lot compared to 250, but LEDs are around 5-6 times as bright as incandescent per watt.  The 250 watt halogen puts out 3600 lumens--this LED puts out over 4000!

The materials should be less than $100, depending on what you can scrounge together and how exactly you construct it.

The hard parts--the machined mount for the LED--I did on the manual mills at TechShop.  TechShop has everything you need for this entire project, even soldering irons and the like in case you don't have one.  Most of you probably don't have access to a mill otherwise, so sign up today!

WARNING: Parts of this Instructable require experience with AC line power.  If you're uncomfortable with this, get someone to help you out!  If you make this at TechShop, just ask around and someone can give you guidance.

Step 1: Gather your materials

The first step is to gather your materials.  In the picture is everything you need.  Starting from the left, you'll see:
- An old CPU cooler.  LEDs need cooling, and you'll want a pretty beefy heatsink and fan to keep things cool.  Almost any model will do.
- A reflector.  This part is optional--you can use it if you need a bit more directionality to your light.  This one I got from Deal Extreme, part number 39963, "20~100W Silver Plated Plastic Smooth Reflector (58mm)", $2.10.
- A machined mounting plate.  I'll show you later how to make this at TechShop.  You'll want a (approximately)  2x2x0.5" block of aluminum for this part.
- The LED plate!  Deal Extreme part number 157806, "50W 3000K 5000lm 50-LED Emitter Warm White Light Plate (32 ~34V)", $16.00.
- The LED driving circuit.  This is a constant-current circuit that takes AC input and converts to ~34 V output at 1.5 A.  Deal Extreme part number 132902, "JR-50W LED AC Power Driver for 50W LED Light Lamp Bulb", $18.80.
- Some miscellaneous screws.  I needed 4x 3 mm screws and 1x 4 mm.  The exact screws don't matter much, but you'll need taps for whatever thread you choose.
- The mount itself.  I used a "Photo Studio Lighting Light Stand Magic Clamp" from eBay for $14.99, but you can use whatever you think works best.
- Some 2 watt resistors, approximately 400-500 ohms.  This is to reduce the 34 volts of the power supply down to the 8 or so the fan needs.
- An aluminum case.  Aluminum is good because it dissipates the heat of the driver circuit.  
- 22 gauge wire.  Not too critical; just make sure it can handle 1.5 amps.
- Rubber grommets.  For the cabling holes in the aluminum case.
- A GFCI plug.  I cut this one from a $9 hair dryer on Amazon!  Standalone cables were closer to $20, so this seemed like the better option.  I highly recommend a GFCI plug for safety!

Comments

[pfred2]

You're into your task lighting. I don't want to sound mean or anything but the most likely reason you snapped a bit is because the drill bit walked on you when you started it. Lubricant had nothing to do with it. You have to be doing super high speed production for lubrication to be a factor with aluminum. If you insist on using your DRO method use a center bit before you use a little twist drill. Better yet just use transfer punches and skip all of that measuring entirely. But then don't drill out on the mill with a vise bolted down. Small twist drills should be run into divots or they'll walk, flex, and snap. Not every time, but enough that common practice is to punch the work before drilling it.

[spcutler]

Don't worry at all about "sounding mean"; I'm a beginning machinist and I know it.

However, I don't believe that bit wandering was the problem here.  That was in fact my first theory behind the breakage (even though I've never had a problem before on a very smooth, freshly face-milled surface), and so on the next hole I used a center bit (actually, a really beefy sharp-tipped chamfering mill).  I then drilled down just a little ways and inspected the bit before going further.  The aluminum had completely galled to the grooves, and I had to dig it out with a sharp object.

A little spritz of lubricant solved the problem completely, giving me nicely-shaped chips instead of long strings that would gall to the bit deep inside the cut.  The last two holes I stopped using the center drill, and I had no problems at all.

You're probably right that I should always punch or center drill first, but the fact is that I'm lazy :-).  That said, I've been meaning to pick up a set of shorty bits, which would also improve matters. 

[pfred2]

In light of the new information I wonder if you were running high enough RPM? Aluminum surface speed is high, about 300 surface feet per minute. Also, you know you are supposed to raise the bit out of the hole often to chuck the chips don't you? You can't expect the twist helix to completely clear the hole for you. You have to start raising after you've gone about the depth of the diameter you're drilling.

I have seen materials stick to bits working it though. That can be because of poor tool finish. In which case then you would have to use some kind of a coating to overcome the deficiency, or buy better bits.

For laughs I figured out what RPM you should have been running for the size hole it looks like you drilled and no milling machine can go that fast. It is slightly over 9,000 RPM. If you were going that fast the chips probably would have flown off the bit don't you think?

I don't know what the top speed of the machine you were using is but I'd be surprised if it is over 4,500 RPM so being as you could only run about half of the speed you should you would have to adjust your technique some.

I just got done running some little jobs here where I had to center drill start a number of holes. It is a pain. Works though. I made bearings brackets and lead screw lock collars for a CNC machine I'm making. The machine has 2 sets of double lead screws too so it added up to 8 of each. On the bearing brackets I drilled for 3 8-32 set screws a piece too. Oh, and I made a couple of extras in case I messed one up (I didn't but I felt better setting up different operations having made extras). So for the brackets alone I had to do about 30 chuck changes just for the center drill. Then another 30 to mount the bits. I'm striving to be lazy, so far it hasn't worked out for me yet though.

P.S. Being lazy always ends up costing me extra work, materials, tools, time, etc. machining.

[spcutler]

You're probably right about the spindle speed; I was at about 3k (machine didn't go much higher), and I think your 9k figure is the right ballpark.

I was using peck drilling but a single diameter seems like overkill; I was going about 3 diameters at a time. Maybe shorter pecks would have prevented the break, but adding lube solved the problem equally well (this is starting to sound dirty...). The chips were actually being cleared well enough that I doubt I needed to peck drill at all.

Sounds like a cool project you've got going. I have a mini mill at home that I've thought about converting to CNC. That's kinda on hold since TechShop has a nice Tormach CNC that's better than any home conversion that I could do.

[pfred2]

I clear chips to stop birds nests from forming as much as for any other reason. There are too many variables to make any specific statements as to when chips should be cleared other than they need to be.

As far as surface speeds go they are all known at this point. I use variations of this formula

SFPM = PI * Dia. * RPM / 12

294.5 = 3.1415927 * 0.125 * 9000 / 12

Aluminum working surface speed is between 250-300 SFPM but small diameter tools should be run faster to stress them less. The additional speed translates into less torque on the tool.

There is a metric analogue to this formula but I am not very familiar with it. Whenever I need to work in metric units I convert to imperial measurements.

Speeds and feeds is intermediate machining magic so it pays to become familiar with the topic as best as you can. For aluminum it is less important, when you work with harder materials like steel it becomes more critical. Perhaps this is because the relative hardness of your tools, and materials gets closer to one and another. Then you must rely less on sheer brute force, and more on finesse to get the job done.

P.S. I didn't bother to look up what drill was used for your tap but just guessed it was about an eighth of an inch (0.125). I was only looking for a ballpark figure and knew you likely weren't even close. Calculating machining speeds is one of those rare places where math is really cool. heh

[lutam]

I did a similar setup for a SAD Light. I have problem with the power driver or the led array. It seems to be overheating after some times, only the strip of leds in the middle of the array stay lit.

[spcutler]

What kind of heatsink setup did you use? A good rule of thumb is that the LED should stay cool enough to touch without burning your finger. If that's not happening, you need a bigger heatsink or one that's in better thermal contact with the LED. A very smooth surface, thermal paste, and a reasonably high clamping force are important.

[Satrek]

I checked a local supplier (Germany), and while they sell 50W LEDs, they cost 80€ and above (~100$)... I wonder where the price difference comes from...

[aeszok]

Ebay is your friend, I found a supplier once that offered 5 of these 50 watt ones for about $40-50.

[Satrek]

Interresting. I wonder whether their quality would be an issue, i.e. whether you would notice the extra money spent or not... At that price, I'd call it a bargain...

[spcutler]

I'm not sure. Your local source may be higher quality, but the $16 unit I used seems perfectly good for my purposes. My source has even larger emitters but they start at $66 for 100W, so the 50W version is the best deal.

[Tomdf]

You sir, are improbably talented at naming projects. Nice work.