Newbe looking for LED design advice

manappraisal

Newly Enlightened
Joined
Sep 12, 2010
Messages
9
Hello:

This is my first posting. I am trying to design an LED lighting array for use for video production in small indoor spaces. I am just getting my feet wet on LED lighting, but from my research I would anticipate the use of 4-6 10 watt emitters powered by a sealed lead acid 12V 7.5A battery.

The light should be diffuse. The idea is to uniformly light small rooms or provide light support for larger rooms that have indirect sunlight through windows.

I would be grateful for some advice on circuit deisgn and emitter selection and requred heat sinking. Do I use resistors or do I need drivers?

My first attempt to light up a 10 watt LED resulted in a burn out in about one minute. Using an online LED calculator a 1 ohm resistor of atleast 4 watts capacity was suggested. I had the LED mounted on a 3 X 6 inch piece of steel which is approx 1/32 inch thick. Thermal paste was used between the LED back plate and the metal.

The LED used was spec'd at 12V forward voltage, 1000ma current. A 1 ohm, 10 watt 10% resistor was used on the positive lead.

Why did I fry the LED, and could recommendations be made on the best approach for getting the most light output using one of these batteries?

Many thanks.

Jeff in Floirida
 
It's hard to guess without knowing the precise details of the device you used, but it could well be that a '12V/1000mA' LED was more like a nominal 10-12V, and it wouldn't be impossible for such a device to have a forward voltage even below 10V@1000mA.
Using a 1R resistor, it's possible you were getting actual currents of 2A or above, which would end up generating more than twice the heat that the device was designed to cope with.

An electronic driver does avoid the problem of current variation.
It is possible to use the resistor approach, but to do that successfully, you really need to do some measurements, to ensure you know what currents are actually flowing for each circuit.

Steel isn't the best material to use as a heatsink, especially for high power devices, as it conducts heat much less well than aluminium or copper (maybe 10 times less well than aluminium,
1/32 is also extremely thin, which also doesn't help.
If I was running LEDs at a watt or so, I might stick them to 0.5mm aluminium sheet, but for higher power, I'd be looking at something thicker, to reduce the thermal resistance to heat moving sideways.
Steel 1/32" thick equates to ~0.08mm thick aluminium - rather thinner than the average soda can wall.

Is this for personal use, or something more professional?

Does it have to be absolutely silent, or might some fan cooling be possible?

LED-wise, it may be worth thinking about using more smaller LEDs.
Though possibly more expensive per lumen, using something like Cree XP-Gs is likely to give more lumens per watt, so for a given light output you'd get longer runtime and also less heat production.
 
I'm going to say that your heatsink was your problem. You're trying to dissipate several Watts of heat coming from the back of the emitter and putting that heat into a piece of thin steel is almost as good as not having anything behind it at all. If you had placed a fan on the back side of the steel sheet and blew air across it you may have gotten the thing to go for another minute or three.

I'm no expert when it comes to heatsinks nor about metallurgy but I can tell you that copper will be about twice as efficient at drawing the heat away from your emitter compared to aluminum. It depends a lot on the grade and alloy of copper and aluminum, respectively. C110 copper, which is 99.9% pure is only slightly less efficient than C101 copper (99.99%) but is almost half the price so it is probably the best material to use here. Also, if you don't plan on having active cooling on your lighting system then thermal mass is what you'll need. This means more aluminum or copper is needed. If you use active cooling then you can get away with less thermal mass.

I fairly inexpensive cooling system that you may want to try would be a computer CPU cooler/heatsink system. They're designed to draw heat away from computer CPUs and aren't all that expensive. All of them are fan cooled so you would need a 5V power source in addition to the power source to the LED but you may be able to run a resistor to reduce the voltage to the fan.
 
If fan cooling is necessary (which it may well be) and possible (depending on acceptable noise output), most computer fans seem to use 12V power. I'm not sure I've come across one yet that doesn't.

That said, dropping the fan voltage can often give much quieter operation while still giving perfectly adequate cooling - a work light I made with an MC-E running at full power on an old CPU heatsink runs nice and cool with only about 6V supply to the 12V fan, and while not entirely silent, is hugely quieter than if running at higher voltages/speeds.

As for materials, for anything where absolute minimum physical size isn't an issue, it's often going to be easier to use thicker aluminium than thinner copper, since aluminium is generally more easily available in a wide range of forms and thicknesses, is typically cheaper, and is also generally lighter for a given amount of heat conductivity.
 
Many thanks.

If I use a driver, can 1 driver be used to power a large quantity of LED's. I noticed that drivers are rated as to their current output. However, how does the current rating relate to the number of LED's powered by the driver?

If I used, say, 21 3 watt LEDs. I would assume (although I am not an expert by any means) that I could use arrays of 3 3.7v LEDs in series and have 7 such sets of 3 (12VDC source) run in parallel to each other. Can this array be driven by one driver, or would each set of 3 need to be driven by a separate driver? Can I assume that 3 watt LED's at 3.7V each need to draw slightly under one amp? If the driver had 12VDC output, what would the driver's current ratingneed to be?

Do you know of any published wiring diagrams to show how such an arrangement would be configured? the LED calculators that I have tried just show resistors. However, the driver ideas seems to make more sense.

To answer your question, the lighting is more for my own use in a portable lighting arrnagement for taking video. It does not have to be pretty or finished looking. I just need the light. All other considerations are secondary.

Thanks for your help and sorry about any misspellings or typos, there does not appear to be a spell check on this forum.

Jeff in Palmetto, FL


It's hard to guess without knowing the precise details of the device you used, but it could well be that a '12V/1000mA' LED was more like a nominal 10-12V, and it wouldn't be impossible for such a device to have a forward voltage even below 10V@1000mA.
Using a 1R resistor, it's possible you were getting actual currents of 2A or above, which would end up generating more than twice the heat that the device was designed to cope with.

An electronic driver does avoid the problem of current variation.
It is possible to use the resistor approach, but to do that successfully, you really need to do some measurements, to ensure you know what currents are actually flowing for each circuit.

Steel isn't the best material to use as a heatsink, especially for high power devices, as it conducts heat much less well than aluminium or copper (maybe 10 times less well than aluminium,
1/32 is also extremely thin, which also doesn't help.
If I was running LEDs at a watt or so, I might stick them to 0.5mm aluminium sheet, but for higher power, I'd be looking at something thicker, to reduce the thermal resistance to heat moving sideways.
Steel 1/32" thick equates to ~0.08mm thick aluminium - rather thinner than the average soda can wall.

Is this for personal use, or something more professional?

Does it have to be absolutely silent, or might some fan cooling be possible?

LED-wise, it may be worth thinking about using more smaller LEDs.
Though possibly more expensive per lumen, using something like Cree XP-Gs is likely to give more lumens per watt, so for a given light output you'd get longer runtime and also less heat production.
 
I guess that I need to brush up on heat sinking as well as power management.

I there rule of thumb regarding the wattage of the LED versus the surface area/volume of the heat snik material.

I did obtain some plate aluminum at Lowes today. it is 1/8 inch X 2 inches wide. I will use an 18 inch length mounted on a partoble rolling cart. The aluminum will be "free air" with nop contact with other materials with the exception of the vertical steel tubes on each side to which the aluminum plate would be mounted.

Can anyone suggest what the highest power LED's that can be mounted on this piece of aluminum and effectively utilize passive heat sinking. Fans and other apparattus are not agood options. Also, what spacing along the aluminum plate should be implemented. A series of these aluminum plates can be implemented i.e. 5 3 watt LEDs per plate?

Should I use thicher or wider aluminum plate?

Thanks.

Jeff
 
The idea is to uniformly light small rooms or provide light support for larger rooms that have indirect sunlight through windows.

You do realize that without correction gels the LED light is going to be much different in terms of color temp and not match the window light, correct? I'm assuming you got the 'Satistronics' variety 10watt whites, which in any respect might not be a good choice for video work because of their erratic CRI and low efficiency.

Do I use resistors or do I need drivers?

Using a resistor on a battery powered LED, especially big ones like this is kind of like towing an anchor with a sailboat; You're wasting a lot of power. If the LED is spec'd for 12volt and you have it adequately heat-sinked then a 12volt battery shouldn't hurt it.

However, the heatsink you are using is orders of magnitude insufficient for a 10watt emitter. I work with a lot of LEDs up to 20watts, and you need aluminum that's at least 1/4 inch thick for a 10watter, and forget steel. Without a fan it will need to be at least double the size of a computer heat-sink, or it will still burn up. That's just for one emitter. For 40-60 watts worth you're going to need a something quite large, like this.
 
For the type of video we are doing, we do not have precise control of the lighting conditons since we video throughout buildings which are going to have influences of a wide variety including: intensity, varying amounts of reflected vs. direct light via windows, time of day which results in different color temps of ambient reflected sunlight, windows on N vs S faces of the building and also influences of any interior lighting.

Auto (or sometimes manual) white balancing of the camera will take care of a lot of these problems. As such, precise matching of our lighting to ambient window light, while desirable, is not an insurmountable problem. Razoring the footage into a number of clips and individuallly color correcting/white balancing/shadow & highlight adjustments via software further helps. The bigger issue really has to do with AE effects which are problematic since we cannot lock down the iris in this application.

Our task is further complicated by the affects which insufficient light intensity has on compression of video for web devliery as well as auto focusing. I have had acceptable results using multiple production "on camera" led video lighting systems. However, these devices tend to create "spotlight" effects on surfaces. Hence the need to experiment with diffuse lighting from unfocused LED emitters. We are admittedly in the experimental stage and conclusions regarding the efficacy are forthcoming.

Unfortunately, geling the windows in an entire building, especially with 2nd and 3rd floor locations is not a practical solution and I am pretty sure would not be acceptable practice in the eyes of our clients. Therefore we need to make the best of an imperfect circumstance.

Regarding direct hook up of a 12 VDC emitter to a 12 VDC battery, could this approach result in unconstrained current draw to the emitter and burn it out even with appropriate heat sinking? If this is not an issue, then I would certainly hook them up directly. I am really not knowledgable in this area and trying to learn. The use of the resistor was determined by an on-line LED calculator.

All advice is appreciated.

Thanks.

Jeff


You do realize that without correction gels the LED light is going to be much different in terms of color temp and not match the window light, correct? I'm assuming you got the 'Satistronics' variety 10watt whites, which in any respect might not be a good choice for video work because of their erratic CRI and low efficiency.



Using a resistor on a battery powered LED, especially big ones like this is kind of like towing an anchor with a sailboat; You're wasting a lot of power. If the LED is spec'd for 12volt and you have it adequately heat-sinked then a 12volt battery shouldn't hurt it.

However, the heatsink you are using is orders of magnitude insufficient for a 10watt emitter. I work with a lot of LEDs up to 20watts, and you need aluminum that's at least 1/4 inch thick for a 10watter, and forget steel. Without a fan it will need to be at least double the size of a computer heat-sink, or it will still burn up. That's just for one emitter. For 40-60 watts worth you're going to need a something quite large, like this.
 
Using a resistor on a battery powered LED, especially big ones like this is kind of like towing an anchor with a sailboat; You're wasting a lot of power. If the LED is spec'd for 12volt and you have it adequately heat-sinked then a 12volt battery shouldn't hurt it.
Some of the 10Watt/12V emitters I've seen around are described one place as 12V, elsewhere as 10-12V, which I'd guess means 3 white dies in series.

Driven off a low impedance nominal 12V source via a 1R0 resistor, such an LED could easily end up with an pretty excessive amount of current flowing, depending on the particular LED, how hot it was, how used it was, and how charged the battery was.

Efficiency-wise, a resistor might not be too bad if the LED Vf is close to 12V, but there is an unavoidable trade-off if using resistors - the closer the LED Vf is to the power source voltage, the more efficient the resistor-based setup, but the more sensitive to variations in battery voltage or LED Vf, and if the supply voltage is going to be somewhat variable, there can come a point where it's hard to get a sufficiently stable setup, however tuned the resistor is to the particular LED.
 
Hey thanks for the link. I will get one of these big heat sinks right away. Don't want to have another 10W LED get cooked.

Regards.

Jeff


You do realize that without correction gels the LED light is going to be much different in terms of color temp and not match the window light, correct? I'm assuming you got the 'Satistronics' variety 10watt whites, which in any respect might not be a good choice for video work because of their erratic CRI and low efficiency.



Using a resistor on a battery powered LED, especially big ones like this is kind of like towing an anchor with a sailboat; You're wasting a lot of power. If the LED is spec'd for 12volt and you have it adequately heat-sinked then a 12volt battery shouldn't hurt it.

However, the heatsink you are using is orders of magnitude insufficient for a 10watt emitter. I work with a lot of LEDs up to 20watts, and you need aluminum that's at least 1/4 inch thick for a 10watter, and forget steel. Without a fan it will need to be at least double the size of a computer heat-sink, or it will still burn up. That's just for one emitter. For 40-60 watts worth you're going to need a something quite large, like this.
 
Unfortunately, geling the windows in an entire building,

Was talking about geling the emitter array to get it to match window light. Obviously if you've tested this in post and the color works then it's a non-problem. However, I was just trying to warn you that the color quality of generic Asian LEDs, especially cool-whites, can be dismal and erratic. Kind of on the same order as using non corrected fluorescent tubes.

At the least, you need more heat-sinking (lots more), and the resistor won't hurt anything and at worst smooths out voltage a bit.
 
Yes, interesting thought to gel the LED's

After I get the rig set up, I will put a color meter on the light to see where we are with color temp. It would be interesting to see how much fall off a gel will create. Not an issue with a 1000 watt quartz light, but wth LEDs, it would be a conern.

Thanks for the suggestion.

Jeff

Was talking about geling the emitter array to get it to match window light. Obviously if you've tested this in post and the color works then it's a non-problem. However, I was just trying to warn you that the color quality of generic Asian LEDs, especially cool-whites, can be dismal and erratic. Kind of on the same order as using non corrected fluorescent tubes.

At the least, you need more heat-sinking (lots more), and the resistor won't hurt anything and at worst smooths out voltage a bit.
 
The professional LED lights I've seen on film sets have a Colour Temperature Dial on them.

At one end of the dial they send all power to Warm LEDs and at the other they send all power to Cool LEDs - with all options in between available.

Sure is more efficient than losing light in Gels.
 
It seems like every couple weeks someone stumbles into this forum wanting to make an led video light. I've toyed around with the idea myself and for all the other people who posted I would recommend HID lighting before LED because it doesn't require custom fabricating control circuits, cooling systems, and is fairly efficient. 4-6 10w led's is 40-60 watts of energy.

Even though I could easily build an led video light should I choose, I doubt the video light guys have much experience with electronics and the specific requirements of high power led's so that the video light will consume a great deal of effort to get up and running. Many just think an led is just like a light bulb, connect to a battery and you're done but that couldn't be further from the truth.

With that much consumption you could equal with just one simple HID. The upside too is cost as a pair of HID's is much less than anything retail you could scrounge for in terms of led equivalents of 110w of HID output. A single 55w bulb can output around 5000 lumens equivalent to a 300w halogen par and you get two in the kit so gelling for temperature correction won't reduce output to useless levels. Plus heat isn't as much concern and the ballast regulates automatically.

This is the easiest for novices to get started with and in terms of efficiency it's on par with most high power neutral white led's driven at their max with a decent cri. I happen to know a 4300k 55w bulb shows shows up excellently on camcorders and won't require as much gel correction as high output cool white led's especially since indoors you'll almost always need warm tinted lights.
 
Hey thanks for the tip.

I would like to investigate this further. Can you make specific recommendations as to HID bulbs? Sources for these bulbs? I need something that runs on battery power, has very wide dispersion (has to avoid spotlighting), and is light weight. This will be mounted on a rolling cart that may need to be carried up steps and transported out of cars etc.

As for complexity, there are LED's which come premounted onto heat sinks and include a driver. All you need to do is mount and connect to a 12VDC power source. That is not too complicated providing they work correctly. However, weight is an issue and if HID lighting is lighter that might tip the balance.

Best regards.

Jeff



It seems like every couple weeks someone stumbles into this forum wanting to make an led video light. I've toyed around with the idea myself and for all the other people who posted I would recommend HID lighting before LED because it doesn't require custom fabricating control circuits, cooling systems, and is fairly efficient. 4-6 10w led's is 40-60 watts of energy.

Even though I could easily build an led video light should I choose, I doubt the video light guys have much experience with electronics and the specific requirements of high power led's so that the video light will consume a great deal of effort to get up and running. Many just think an led is just like a light bulb, connect to a battery and you're done but that couldn't be further from the truth.

With that much consumption you could equal with just one simple HID. The upside too is cost as a pair of HID's is much less than anything retail you could scrounge for in terms of led equivalents of 110w of HID output. A single 55w bulb can output around 5000 lumens equivalent to a 300w halogen par and you get two in the kit so gelling for temperature correction won't reduce output to useless levels. Plus heat isn't as much concern and the ballast regulates automatically.

This is the easiest for novices to get started with and in terms of efficiency it's on par with most high power neutral white led's driven at their max with a decent cri. I happen to know a 4300k 55w bulb shows shows up excellently on camcorders and won't require as much gel correction as high output cool white led's especially since indoors you'll almost always need warm tinted lights.
 
Ebay. Just type in "HID 4300k kit". These are automotive car headlight kits so 12v is perfect. They have the same light dispersion as a regular light bulb so avoiding spotlighting is no problem, they'll even work in standard generic soft boxes. I doubt you'll find a decent single premount led with driver for less than $45, let alone 4-6 of them. Also one thing HID has going that most people don't consider is the warmer the HID the brighter it gets whereas led it's the opposite, warmer equals less bright. Oh and as for weight my whole HID kit weight about 3 pounds, that's two bulbs and two drivers.
 
Last edited:
In some respects low wattage HIDs bulbs are actually more efficient than Chinese white LED's. I checked the specs for a number of different 120volt based low wattage HID brands and found even the low CCT (3000k) HID bulbs with 80'ish CRI had 70-80lumens per watt efficiency. That's better than many popular Chinese LED arrays at cool-white. Certainly better than 5-10mm beehive video lights.

I'm not sure how this translates into automotive HID, but I would assume the bulb compositions are similiar. My only reservation is that the 12volt HID market seems to be flooded with a lot of junk bulbs that likely have inferior color and performance to cheap Chinese LED arrays.
 
Well I have an array of several warm white ~3800k mr-16 bulbs. These are generic and comparing how the lights photograph these are terrible at reproducing red color correctly even when correcting in a photo editor there's a difference in color rendition between cheap led's, good led's and HID. HID still wins tops, no one looks sickly on HID but led can be very unflattering.

My HID on the other hand looks just like daylight on the camera, so I'd still pick the automotive HID over generic led light, and even a good led light like a an SSC P7 doesn't match the color rendition of a 4300k HID.

I've use my HID light on my motorcycle and it is exposed to vibration and water from washing, sadly it hasn't rained enough for me to test rain performance but it has held up fine for over a year of use. Which translates to about 1000 hours.
 
and even a good led light like a an SSC P7 doesn't match the color rendition of a 4300k HID.

Not sure if I'd go that far, but we're getting into a subjective category. I do agree that cool-white LED color rendition is way over rated and over sold. To be fair, color rendition with some automotive HIDs isn't exactly stellar either. I've seen CRI numbers as low as 50 for the later.

However, if we were to compare 4100k HID to 4100k LEDs of good breeding when both are around 80'ish CRI I don't think one could make an arguement for color quality of HID over the LED, but working opinions are welcome of course because they will be different. If anything, LEDs have a smoother spectral curve than HID for most of their range. Making a video light from top flux rebels or XP-Gs would also be significantly more efficient than HID, but that's a different arguement.
 
Top