High powered LED Lamp project

[DavidOz]

This is part #1 of a series of posts for a DIY LED lamp project. Hopefully the information (and rookie mistakes) I made might help others with similar projects.

Background.
Lounge room was a bit dark so I decided to build a LED lamp on a long arm. This was to be used as a reading lamp and also to brighten up the room, which is fairly large. Sounded simple enough, right?

Part Selection.
The plan was to have a single LED, rather than have a shower head type arrangement. It soon came became apparent that there aren't many options to directly connect a high powered LED light directly to 240V (mains power in Australia). Most of the 240V systems are designed as down light replacements and surprisingly aren't particularly high powered (6W to 10W are common). So this was ruled out, but in hindsight it might have been a easier option to run 4 or 5 of these 240V down lights together.

After doing some reading several people suggest CREE LEDs. After some more reading I selected the CREE CXA2011-3000k. This is a non directional LED about 22mm across (~40Volt, 1 AMP). Claimed Lumen output was around the same as 150W incandescent bulb. Which sounded great.

The next surprise was that high powered LEDs can't be directly connected to a traditional transformer or battery. The problem being that you can't limit the current through the LED (due to their non linear Resistance) with a transformer. Being non linear, small changes in the input voltage can result in large current changes, and each LED is slightly different. Too much current in a LED can burn them out, so you don't want that. So you either need to add a Resistor into the circuit or get yourself a special transformer known as a LED driver.

LED drivers efficiently maintain a constant current in the circuit, and also allow for the LED to be dimmed. My understanding is that these drivers maintain constant current by varying the voltage to achieve the required current.

But for the CREE CXA2011 there weren't many local options for a driver (at least not without importing from overseas). The driver need to have a higher voltage and power rating than the LED.

For the LED in question a driver of at least 40Volts was required. Almost the only option for this LED was the Meanwell ELN-60-48. It can vary the DC output current from 24V to 48V to run a circuit at up to 1.3A.

1.3A was too much for this LED, but the driver has a trimpot on the circuit that can be used to reduce the current (and a separate dimmer circuit).

Here's a couple of photos of the components.

Next post will be the build details, heat problems, more problems with the dimmer, and the final results.

 

[DavidOz]

High powered LED Lamp project - Part #2

Lamp arm
Next step was to build the lamp base and arm. This was going to be a thin curved timber arm with the power cable running up through the center of the arm, and the LED right at the very tip.

Power cable was a few meters of HPM 12V heavy duty garden light cable I had sitting around from an older project. The lower the voltage being used and the longer the run, the more important it is to use heavier low resistance cables (not that 2.5m is that long).

I won't go into the details of building the arm here, except to say it involved a lot of clamps and glue, and that Lon Schleining's book "Wood Bending Made Simple" was of some help. See photo below.

Tasmanian oak hardwood was used, as Oak was meant to be good for bending without fracturing. I found out much later that Tasmanian oak is in fact Eucalyptus, and not Oak at all :-( Turns out that early European timber workers in Tasmanian thought this Eucalyptus was a bit like Oak, and so mis-named it. Further, it turns out that not all hardwood is hard. In fact balsa wood is also classified as hardwood! But getting way off track here, the bending of the arm went mostly OK regardless of the timber being fairly soft Eucalyptus.

Dim witted
The first trial run of the circuit on the bench failed to light up the LED. At this point I had to go back and properly read the specs on the Meanwell driver. Turns out I had made the rookie mistake of assuming that if nothing was connected to the open dimmer circuit then the driver would run at full power. Turns out this is not the case. Further it turns out that there are at least 2 versions of the Meanwell 48V driver. One, the "D" version, has a analog dimmer circuit (using 0V to 10V inputs) and the other, the "P" version, which has a more complex pulse width modulation PWM input. I had a "D" type. You *need* to have an input on the dimmer circuit for it work. This was fixed by a quick trip down the local supermarket to buy a 9Volt battery.

With the 9Volt battery attached to the dimmer circuit it was possible to crack open the driver case and adjust the internal trimpot in the driver with a screwdriver to set it up for 900mAmps, using a multimeter in line with the LED. I had to wear sun glasses to get this done, as the LED was massively bright. I also could only run it for a few seconds before it all got too hot.

Why Meanwell designed the driver like this I don't know. Firstly why require an external power supply for the dimmer? Having an internal supply would have made much more sense. Secondly why require it to be 10V? Almost nothing is 10V. On the other hand 9V batteries and transformers are easy to source.

Here's a couple more photos,

Next post will be about the heat sink and the lamp head.

 

[DavidOz]

High powered LED Lamp project - Part #3

Heat Sinks & Thermal Resistance
Before starting this project I imagined that LEDs would give off almost no heat. Looking back, that seems silly now, but I think this is the expectation of people who are just getting started and who are only familiar with low power output LEDs.

Traditional tungsten incandescent bulbs are roughly 2% efficient. Meaning 98% of the electrical power gets converted into heat and invisible infrared radition. So it normal that they get hot. LEDs are way better, but still only manage to get to roughly 10% efficiency. There is a good comparision here,
http://en.wikipedia.org/wiki/Luminous_efficacy
So ~90% of the power you feed in still gets converted to heat with LEDs!

Another big difference between incandescent lights and LEDs however is that LEDs will deteriorate and fail if temperatures get too high. In the case of the CXA2011 the maximum junction temperature is 150C (the 'junction' being the internal temperature inside the diode). But the CXA2011 puts out around 35% more light at 25C compared to 150C. So you want to keep them cool for efficiency as well as longevity.

From the published specs, the CXA2011 has a effective thermal resistance, junction to case, of just 0.4C/W. Which I understand to be pretty low comapred to other LEDs. This means that the heat can escape fairly quickly, which is a good thing. My (maybe incorrect?) understanding of this is that if the junction temperature is at 85C and you are running the LED at 43W then there will be a 17C temperature gradient between the junction and the external case of the LED. So a measurement of 50C on a heat sink means the interal temperature is around 67C. Someone step in here an correct me if this is all wrong?

Being cheap I didn't want to buy a heatsink, so I hunted around and salvaged old heatsinks from various old computers and other electronics. See photo below,

Deciding I didn't want a fan, I picked the biggest most effectively looking unit with a copper base, heat pipes and large surface area. It isn't in the photo above, as I had already attached it with some thermal paste to the LED at this point, which has a aluminium plate on the back for this purpose.

As the LED was just so bright I also decided to add a frosted glass panel (acid exteched) in front of the LED to make it more disfuse. Didn't want to blind the kids. Also stained the wood and gave it a coat of polyurethane at this point.

Next post will be some temperature & power measurements and the lamp in action.

 

[DavidOz]

Re: High powered LED Lamp project - Part #4

Power measurements

I needed to find a better solution for the dimmer circuit than a 9 volt supermarket battery (which went flat after a few days in any case), so replaced it with a low profile 9V transformer and put that under the base of the lamp and connected it up to a 10K ohm linear potentiometer. Again I would have like to used a 10V unit, as per the Meanwell specs, but the local electronics store didn't have one on the shelf. But 9V was already retina searing bright.

With the lamp undimmed the whole system (driver, LED and dimmer) was pulling 48.1W from the wall. Dimmed down to the min level dropped the usage to 13.7W

Temperature measurements

Running the lamp at full tilt slowly ramped up the heat sink temperature over 15min to around 60C, with an ambiant air temp of 18C. I expect this to get significantly hotter in the summer months when ambiant reaches 35C, as the room isn't air conditioned. So in summer I am guessing the internal LED juntion temp might get to ~94C (201 degree Fahrenheit)

Love that temperature gun with the twin laser red dots. Great toy!

The result - the Miniature Sun
Overall it was way more work than expected, and significantly more expensive that I initially imagined. You definitely wouldn't do it to save on the power bill. A cheap Chinese lamp fixture with a CFL would be way cheaper. On the other hand I now have a unique piece of furniture with a steam punk look.

A couple of photos of the finished product are attached.

Don't stare at it too long, you'll go blind!

 

[BM_lol]

Missing some optics; maybe try and look for a frosted glass diffuser. Wow, that's a HUGE heatsink you got up there, and good job on balancing the weight from the base.

 

[Neondiod]

Very nice :twothumbs :rock:

 

[blasterman]

Looks good. Love the arm, and good work getting through the thermal math.

Design suggestion though. Standard frosted glass has the annoying green tint, which may / may not bother you. A piece of 1/4 clear plexi with window film designed to emulate frosted glass works really good and there's no green tint. Plus, you can stack increasing layers to get the right amount of diffusion. You've probably also learned that the bigger the diffuser the further back the LED has to be mounted. So, it's basically a matter of how small you are willing to tolerate the light source -vs- how you want it to look.

Again, nice job.

 

[netprince]

Nice project! I like the stand, well implemented.

 

[slebans]

Re: High powered LED Lamp project - Part #3

Heat Sinks & Thermal Resistance

Traditional tungsten incandescent bulbs are roughly 2% efficient. Meaning 98% of the electrical power gets converted into heat and invisible infrared radition. So it normal that they get hot. LEDs are way better, but still only manage to get to roughly 10% efficiency. There is a good comparision here,
http://en.wikipedia.org/wiki/Luminous_efficacy
So ~90% of the power you feed in still gets converted to heat with LEDs!

That article is misleading as it uses the LER value of 683 lumens per watt for monochromatic green at 555 nm to calculate the efficiency of various ight sources. You need to use a LER value that corresponds to the specific wavelength(s) of the LED in question. For example, a common White(Blue with YAG phosphor) LED might have a LER value of 300 lumens per watt. If the LED outputs 100 lumens per watt then it has an efficiency of 33.3%.

Finally, incandescent bulbs are closer to 4% efficient while CFL bulbs are around 20% efficient.

Stephen Lebans

 

[mds82]

Re: High powered LED Lamp project - Part #3

Very cool project - Good Job!!! One thing you can do to make the light less blinding is move the frosted glass farther away from the LED. In a way, create a light mixing chamber so that the light is less intense to look at. Maybe as a test, take a glass jar like the picture below, frost the glass with a glass etching kit, and then aim the LED into that. this will give you a more diffused look... Just an idea

Edit.. or even to make it look nicer get a bowl shape and make that frosted

 

[DavidOz]

Re: High powered LED Lamp project - Part #3

Thanks for the feedback and suggestions.

Yes, the glass has slight green tinge, but I don't mind it. You can't see it from the photos, but it goes with the glass bench tops that are also in the room. It is also about double the thickness (10mm) of glass than you'd normally put on a light or in a window, which gives it a more solid look. Besides being frosted, the edges were beveled and two holes drilled through the glass plate for the mounting brackets. BTW: If you are drilling glass don't drill near the edge. I cheated and had the local glass supplier cut, frost, bevel and drill it.

As it's main purpose is a reading light, it will be above and behind the chair for the most part. So won't be staring into it.

 

[samgab]

Re: High powered LED Lamp project - Part #3

Wow. I'm very impressed!

 

[foxtrot824]

Re: High powered LED Lamp project - Part #3

Very nice! I like the wooden arm style.

 

[MikeAusC]

Re: High powered LED Lamp project - Part #3

Congratulations David on completing this challenging project.

Did you get the CXA2011s from LEDsales ?

If you want to eliminate the power wastage in the 9 volt supply, you could use a resistor and Zener Diode to generate 9 volts from the LED voltage.

 

[samgab]

Re: High powered LED Lamp project - Part #3

Congratulations David on completing this challenging project.

Did you get the CXA2011s from LEDsales ?

If you want to eliminate the power wastage in the 9 volt supply, you could use a resistor and Zener Diode to generate 9 volts from the LED voltage.

In my opinion, this project looks like a piece of art, as well as a functional and state of the art LED light source. I love it.

I just discovered Cutter Electronics.... I wish we had a store like that here in NZ! Still I bought a bunch of stuff from them online.
I think I saw this emitter on their website as I was browsing though, FYI.