My attempt at home interior lighting

[James Jackson]

Well, I've been reading a lot here - and learning a lot, too.

Over the last year or so, I've put together several 'prototype' LED lights, and I think that I've finally got one right.

What I've done is designed a PCB that is about 1-inch by 12-inches and has 6 LEDs on it - along with its' constant-current drive circuitry.

I am driving the 6 LEDs at about 700mA. They are Avago Moonstone LEDs (yeah, not the best - but very user-friendly when it comes to hand soldering). These particular LEDs - I believe - are about 140-something Lumens - nominal, at 700mA. They call them 3-watt LEDs.

I have this LED Tube installed in my home - on the wall - ant it is able to be rotated - to allow for the light to be aimed higher - or lower - into the house. It is controlled by the controller box that I have next to it, which allows me to turn on/off the LED Tube using an IR remote. The controller also monitors the battery voltage, and when it gets too low, it turns off the load - to protect the battery.

The LED Tube does not heat up too much - it gets up to about 45-degrees C (about 110F), as measured the best I could using a thermocouple placed as close to the LED as I could get it.

This system is running off of an SLA 12-Volt battery that gets recharged during the day by a solar panel. The battery is almost 3-years old, and I may have to replace it soon, though - especially when I start adding more LED Tubes to the system.

I will try to add a few photos of what I have set up on the wall...

Next, is the LED Tube lit... (I apologize for the poor quality of these photos - I'm still learning how to use my new camera)

Lastly, is a photo of the controller - well - the box that the controller is in.
You will notice the IR receiver, and the two indicator LEDs - one Green, the other Red. I also have a small speaker inside the box for audible indication.

I hope I have not violated some forum 'rules' by posting this. I am just trying to share what I am doing - that seems to be working.

Thanks for reading,

James Jackson

 

[Tekno_Cowboy]

That's pretty interesting, but I'll bet that PCB was pretty expensive to have fabricated...

 

[James Jackson]

RE: Expense of the pcb -
Nope - the PCB is standard 2-layer, FR-4 material - 0.062" thick.

Of course - 'expense' is relative. I use a place called pcbfabexpress - where one can get prototype pcbs for about $15.00 each. I usually get 5 pcbs - which comes out to right at $100 when one factors in the shipping.

Regards,

James

 

[Tekno_Cowboy]

Wow, that's a quite a bit cheaper than I expected to hear. I've been meaning to get some PCB's made, but I assumed that they'd be a bit expensive. I'll have to look into it again.

Thanks!

 

[jeffosborne]

Very nice, James! I wonder what your constant-current circuit is like, switching or otherwise? Also, are the fat lands on the PCB the only heatsink, or is there some aluminum lurking in the lamp somewhere? Also, the optics are narrow beam or wide? We would like to see abeamshot, if you can! Thanks for sharing your design. Jeff O.

 

[James Jackson]

Jeff,

Thanks for the positive words.

RE: What the constant-current circuit is -
It is just using a Catalyst Semi CAT4240 (I think it is now ON Semi) circuit - straight from the datasheets. It is a boost switching circuit. Very simple in design - and I've used it several times. It always seems to work.

RE: 'fat leads on the PCB' -
Ummm... are you referring to the copper pour? This is just a ground pour - that connects across the board. There's also a ground copper pour on the back side of the pcb. Not really much of a heatsink. The heatsink is behind the pcb - and is my own creation.
It is NOT aluminum - but copper - which is much better than aluminum at dissipating heat.

One of my goals in making this LED Tube - was to use commonly available items - that one can readily purchase. The FR-4 PCBs. The copper for the heatsinks. The cost for the copper is not that much - when you consider what I am doing with it.

I've never really liked the aluminum-core pcb concept. (I have designed a few in the past.) You have to design your circuit on one layer (top layer), and it is costly if you design a pcb that is very large. The aluminum dissipates the heat... but only does a fair job at doing it - and one usually winds up attaching a large aluminum finned heatsink on the backside.

RE: Optics -
These are 30-degree optics - that just snap on over the Avago Moonstone LEDs. I think that they cost about $2.15 or so each. There are also 15-degree and 6-degree optics. I just purchased the 15-degree optics - which arrived today. I am going to experiment with them.

RE: beamshot -
I will try... I am not sure what it will look like though... or how exactly to do one.

Regards,

James

 

[James Jackson]

More photos - I've finished the second LED Light Tube.

This first shot is of the two LED Light Tubes mounted on the wall. Note the lenses. Inner ones are 6-degrees, next are 15-degrees, and outside ones are 30-degrees.

The next photo is of the LED Tube lit up. It draws about 1.4 Amps at 12.5V - give or take.

Lastly, is my attempt at getting a beamshot of what it illiminates. The far wall is about 26 feet away. If you look real closely, you can read the time on the clocks.

I made a mistake - and left the floorlamp on - but otherwise there are no other light sources (well... other than the TV) on in the room, and I didn't use a flash on the camera when I took the photo.

 

[blasterman]

Anybody that 'rolls their own' driver gets a big thumbs up in my book. Wish I had more time to work with this end of things.

I ma a bit concerned about heat-sinking though. At 700mA I like to have at least 16 square inches per emitter or things start to get too warm.

 

[Vesper]

Nice project and very interesting, thanks.

 

[Dave_H]

I ma a bit concerned about heat-sinking though. At 700mA I like to have at least 16 square inches per emitter or things start to get too warm.

I also commend you for this initiative, but am puzzled as to getting
away (so far) with almost no heatsinking. These LEDs seem to be
consuming 2-2.5W each, most of which is going into heat.

Dave

 

[James Jackson]

I also commend you for this initiative, but am puzzled as to getting
away (so far) with almost no heatsinking. These LEDs seem to be
consuming 2-2.5W each, most of which is going into heat.

Dave

Dave,

I'm curious as to why you feel that most of the power is going into heat? I have measured the heat on the LEDs - and it does not go above 110F - well within allowable temps for the LEDs.

Also, remember - I am not using an aluminum heatsink - I am using copper, which is much more suited as a heatsink.

Regards,

James Jackson

 

[Dave_H]

James,

Well over half the LED power is getting dumped into heat. I don't know
thermal efficiency for these LEDs, but I recall reading it may be as high as
about 40% though if so I doubt it is for power LEDs run at high currents.
Perhaps someone can shed light on this?

DOE site quotes 15-25% although it may have improved. Let's be
generous, say 30%. Checked the specs for Avago 3W Moonstone LEDs,
using vf=4.0v at I=0.7A, about 2W per LED is dissipated.

http://www1.eere.energy.gov/buildings/ssl/comparing_lights.html

You measured case temperature, which is lower than the die (junction)
temperature. Package die-slug (junction-case) thermal resistance is
10C/W, so the die is 20C hotter, around 65C for your test, but still OK.
What was the ambient for your measurement? Keep in mind that if the
ambient goes up, so does the die temp. I've heard 80C as typical limit
for long life; all below is better.

Regarding copper v. aluminum I don't think it will make a large difference
in this case. Thermal resistance of the PCB copper is increased by its
thin-ness (0.7mm for 1/2-oz copper, 1.4mm for 1oz); can't compare
directly to a thicker aluminum heastink.

Dave

 

[James Jackson]

"Regarding copper v. aluminum I don't think it will make a large difference
in this case. Thermal resistance of the PCB copper is increased by its
thin-ness (0.7mm for 1/2-oz copper, 1.4mm for 1oz); can't compare
directly to a thicker aluminum heastink."

Dave... ummm... you may want to take a closer look at the photos that I posted. There is more to my heatsink than just 1 oz copper on the PCB.

Lots of good words... but they mean nothing - as I am getting lots of LED light output. I know that you may be knowledgeable and all...

Regards,

James Jackson

 

[Dave_H]

Dave... ummm... you may want to take a closer look at the photos that I posted. There is more to my heatsink than just 1 oz copper on the PCB.

OK, now that I look carefully and re-read the text I see the
tube heatsink running along the back. That explains the name,
thought it was a generic reference to the form factor. Since
I can't see the back, just curious how the copper is attached/
coupled to the PCB. Do you use a bunch of "thermal vias" around
the LEDs for heat transfer to the back side?

As for meaning nothing, well I hope the info is useful to others
(maybe myself) who attempt similar projects. I believe I've
raised a couple of good points.

Dave

 

[James Jackson]

"DOE site quotes 15-25% although it may have improved."

Looking at this data, it appears to be from 2006 - or earlier. High-Power LED technology has improved tremendously since 2006. I have been working with High-Power LEDs since about 2002 or so. Back then, the Hi-Power LEDs were at about 25 to 30 Lumens per watt - if that.

"You measured case temperature, which is lower than the die (junction) temperature. Package die-slug (junction-case) thermal resistance is 10C/W, so the die is 20C hotter, around 65C for your test, but still OK. What was the ambient for your measurement?"

The ambient was in an office environment - which is/was about 25-C.

"Keep in mind that if the ambient goes up, so does the die temp. I've heard 80C as typical limit for long life; all below is better."

Using Figure 9 of the Avago Datasheet - I can go up to about 95C on the slug - and like I said... I am well below that - at only 45C.

"Do you use a bunch of "thermal vias" around the LEDs for heat transfer to the back side?"

No, I do not. The method I am using is proprietary. Again let me say that I am using a copper heatsink attached to a plain FR-4 Printed Circuit Board. Something that can be easily fabricated - which I did on our dining room table.

What I will say is that I need to use a soldering gun - no... not iron... 'gun' - which is rated at 140 Watts on High heat. When soldering to the copper - the copper pulls the heat away from my soldering gun - which makes it difficult to make a good soldering joint. The copper, meanwhile - is getting hotter and hotter. Something that I don't want to touch until it cools down a bit.

Another thing that I will say is that most people think 2-dimensional when talking about heatsinks. I am going 3-dimensional. X - Y & Z.

Comparing copper to aluminum - according to my research - copper is twice as good as aluminum at dissipating heat. Period.

"I believe I've raised a couple of good points."

Dave... I hope you realize that I have researched this. I didn't just get out the soldering iron - buy a few LEDs and throw something together.

Keep in mind that I am talking about 12 High-Power LEDs. They may not be the most efficient LEDs on the market, but I could build the Light Tubes using almost any High-Power slug-isolated LEDs - and be successful, using my technique.

As it is - with these 12 LEDs, they are rated at a nominal 125 Lumens each. Even if I derate them to 100 Lumens - that's a minimum of 1200 Lumens of LED Light being powered by a 12-Volt Battery - for lighting our home in the evening (or whenever).

Also, keep in mind that the LED Light Tubes are one-half of my system. The controller is the other half. It does more than just turn on/off the LED Light Tubes. It has a Basic Stamp inside as a controller - and it samples the DC voltage, and will turn off the load if the battery voltage drops too low (LVD). It has a few other features, too.

Currently, it only switches the LED Tubes on and off. In a future spin of the board, I am going to add a dimming feature which will allow me to set the LED Tubes at various settings of brightness from full-on - to full-off.

But that is in the future. Right now, my wife and I are enjoying the ability to use an IR remote to turn on some bright lights.

Regards,

James Jackson

 

[blasterman]

Comparing copper to aluminum - according to my research - copper is twice as good as aluminum at dissipating heat. Period.

Not sure why you are getting so defensive on this, but they were logical questions.

However, for the sake of arguement, the Thermal Conductivity of copper vs aluminum is pretty irrelevant because it's surface area required to dissipate the generated heat that's critical. Copper 'transmits' heat faster than aluminum, but the heat has to go somewhere, and that's typically the air. Note that very, very few computer heatsinks are made of solid copper but typically aluminum with an occasional copper slug core.

So, that's cool you are using copper, but it wouldn't make a difference if you were using aluminum (or solid silver).

Also, last time I checked there was no such thing as a 2-D heatsink (they are all 3-D). :naughty: However, when you get away from simple heatsinks like 1/8 alu bar convection issues and orientation start to come into play.

I've mounted a lot of LEDs on copper pipe, and it works better if the pipe is cut into sections because otherwise the heat inside the pipe doesn't efficiently get to the outside. However, if you aren't running into thermal problems then it's nothing to worry about because you have enough area on the outside.

In regards to the PCB, this is one way to do it, but all it takes is for one component on the circuit board to malfunction and you have to chuck the entire board.

 

[Dave_H]

Blasterman made several points I was going to, no point repeating
all, except to say that thermal resistance of heatsink material will have
some non-zero affect on the outcome, but relatively minor in this case.

2D heatsinks are thin/flat such as copper foil on PCBs...not terribly
efficient but are cheap/free and do help the overall cooling; object
being to spread out heat over the largest area to let convection take it
away.

Copper pipe is (fairly) cheap, available, and solderable, and it all
works. Blasterman's idea about segmenting the pipe makes sense,
allowing more convection including inside surface of the pipe(s).
Another idea would be cutting slots across the pipe in such a way as
to not weaken it too much, but allow air circulation. Yet another is to
mount a small 12v fan at the end of a single pipe, pulling air through it.
This of course for future consideration given that current design works.

I recently worked on a design where up to 25W needed to removed
from an area not much more than a square inch; definitely a forced
convection case. Heatsink material was aluminum, not copper. BTW
has anyone seen/used plastic heatsinks...they do exist.

This still leaves unanswered questions: what is the power efficiency of
these and other power LEDs, and how much do they dissipate? I don't
see how this relates directly to luminous efficacy (lumens/Watt).

Dave

 

[James Jackson]

"Not sure why you are getting so defensive on this, but they were logical questions."

If it sounds like I am being defensive - perhaps you should re-read the questions. They appear to be attacking my design - which is working.

The 'help' being offered only seems to be that I need to do it differently - or that it is not working - which is incorrect. I have taken the time to document - to the best of my abilities - what the temperatures are (using a Fluke Thermocouple multimeter - the model number escapes me) - and not give a 'it doesn't get warm' type of posting - like some folks do. Nothing wrong with that - but when someone then comes back and tells me that 'most of it is going to 'heat' - it doesn't make any sense.

"Also, last time I checked there was no such thing as a 2-D heatsink..."

Then why do you say this?

"At 700mA I like to have at least 16 square inches per emitter or things start to get too warm."

You are giving your heatsink dimensions in 'square inches' - a 2-dimensional measurement. Correct?

"I've mounted a lot of LEDs on copper pipe,..."

Yes... I have read some of your threads on this - but you didn't post any photos of your work - if I am not mistaken. That's okay. Also - from reading some of your posts, it appears that you are using 'stars' with LEDs mounted on them - and then attaching the stars to whatever heatsinks that you are using.

You are losing some of the 'thermal conductivity' by doing this - as the material that is being used is similar to Bergquist T-Clad material - which is good, but has a thin 'Conductive Dielectric Layer' that - while good, insulates the copper thermal slug from the heatsink.

Also - you are mixing metals - attaching a copper (well, if it wasn't insulated from the aluminum 'star' base) slug to an aluminum heatsink. I am using a copper-to-copper transition for heat dissipation. More heat is conducted by doing this - as you know.

"...and it works better if the pipe is cut into sections ..."

For you. I think that it is important to remember this. For you - using your technique - this is probably true. For me, My goal is to create a low-cost - easy solution using readily available componets - and very little modification/work. Using one long tube like I am doing, and am documenting - is working very well. Again - it is how I am connecting the LED to the copper tube that makes the difference - which you are confirming.

"In regards to the PCB, this is one way to do it, but all it takes is for one component on the circuit board to malfunction and you have to chuck the entire board."

I don't think that this is even a valid statement. What you are doing is 'assuming' that I am using faulty electronics to drive the LEDs. What I am using is a circuit that I have proven to be good over the last few years. It is just as reliable - if not more-so than other solutions that I have seen - or read about. Plus, it will allow me to add dimming later on - when I re-spin my control board adding the needed PWM circuitry.


I think that you - and Dave - may be missing out on a very important feature of my lighting system. It runs off of 12-volts - which is a SLA battery - that gets recharged by a solar panel.

If you look real close at the photos I've already posted - you will notice a tan wire running along the wall. This wire is a 40-foot extension cord - modified by cutting off both ends (so that it cannot ever be plugged into 120-VAC by accident), and wired up to the battery on the far end - which - if you squint - you can see the battery box right next to the door in my beam-shot photo (look for the green blur - that's the voltmeter that constantly displays battery voltage).

I am running my LED LightTubes as a separate lighting fixture - which is in no way connected to the 120VAC wiring. We use this light in the evenings - or during the day when more light is needed. The electric bill has dropped since I implemented my first set of LED lights - which are 3 MR-16's.

My lighting system works very well. It even works when the power goes out in the neighborhood during bad thunderstorms - or when someone hits a light pole.

Regards,

James Jackson

 

[Dave_H]

James, I'm baffled as to why the questions, comments and recommendations are construed as an attack on your design. As you don't seem too open to ideas that aren't yours, consider that my recommendations are targetted at others who may want to take your design (which as you point out and we all seem to agree, works fine) and make their own variations. It seemed a logical place to discuss possible improvements to the heatsink and how it works, but this being your thread I could take it to a different discussion.

Dave

 

[biggerrigger]

James, I wanted to put my .02 in if I may. I find your design very well thought out and exucuted. This is a fine example of one finding a DIY solution to a problem. Please post any future mods to your system as there are some that would like to see this project progress into its final state. Again, very good job and I look forward to seeing more DIY solutions.

Aaron