High Wattage Halogen Replacements

Huh? Thermal cycling the LEDs to shift the wavelength a nm or two? The plants sure won't care. As long as there is light in the wavelength they can use for photosynthesis, they will do so.

Jon

What IS the optimum wavelength for "indoor plant" growth ?>:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142   Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
           The touchstone of liberalism is intolerance

That's a large room. Many young people here rent a flat that has a smaller total area. May just as well need the 10k lumens after all.

Certainly achievable with LEDs, but trying to do it with one only (as opposed to several smaller power ones apart from each other) is sure not easy. Besides, the lumen ratings are tricky and hard to compare.

With a halogen lamp, the maker specifies a lumen rating and it's pretty much constant. Until the lamp darkens from old age, it doesn't change. In particular, it does not depend on ambient temperature, heatsink efficacy and such external environmental factors.

With LEDs you should take some care about how to interpret the ratings and it doesn't hurt to know some caveats and limitations as well.

Here's a list. It's in fairly generic terms, as they apply to such light sources. It may be incomplete, but should do for a start:

1. LED efficiency drops considerably with rising temperature (the same led at the same drive current will put out less light when it is hot). This is an important point because high power LEDs must be operated at elevated temperatures out of necessity, affecting the efficiency.

(they measure the output with a short pulse so that it won't heat up).

2a. Makers often claim lifetimes based on unrealistic temperatures.

1. The maximum luminous efficacy is not located at the maximum rated current (to get the stated number of lumens per watt it is often needed to run the LED at some power considerably smaller than abs max rating).

3a. Sellers will sometimes multiply the maximum lumens per watt number with the maximum allowable power (often at ridiculous temperatures that would need some serious freon plumbing to achieve) and claim that their device can output this much light. They don't tell you that they have arrived at the number by taking values that don't fit together because they apply to different operating regimes and are not attainable all together at the same time even in theory.

1. They also don't tell you how much the light output is supposed to reduce as the LED ages over the course of the years. The life expectancy is usually calculated based on the failure rate that includes both total failures and reduction of the light output below a specified minimum. The minimum can be quite low (like 50 % of initial). Less than honest sellers, and that's the vast majority, don't like to tell you, on what basis their ratings are specified / calculated.

4a. This applies to fluorescent and discharge lamps too. They all drop in light output over their useful life. But reputable manufacturers at least specify the ratings somewhat more clearly and honestly, reading the datasheet, one can at least understand the acceptance criteria.

1. LED lamps are still a new technology that many people don't fully understand them or lack reliable comparisons with what they have known for years. As long as many people are "jumping" on this "bandwagon" with great expectations but much lesser certainties, there is also a lot of quackery going on and a lot of snake oil is being sold. Honest business is a rarity, and in new products it's always much rarer than elsewhere. Expect to see sellers claiming unbelievable ratings under dubious conditions as it may take some years as the dust settles and people learn what realistic values are, and how to spot obvious quacks.

As for your particular light that claims 5k lm at 50 W, some care is

temperature, which may be as realistic as putting the lamp in a fridge, this will probably be the least of your problems. The biggest problem will be to keep if from burning up by overheating. If it really has a 50 W rating (which is, for a 10 $ product, again very dubious), then it will need serious heatsinking, and most likely a sizeable fan too.

Fortunately heatsinking is, to some extent, amenable to calculations, so you can at least find out some minimum requirements that way.

For one thing, you need to know the ambient and the maximum permitted chip temperature. You also need the thermal power, but for the sake of simplicity, you can equate it with 85 % of the electrical power of the LED (this assumes a 100 lm/W luminous efficacy = ca. 15 % efficiency).

Let the thermal power be 42 W (let's say your LED really has 50 W electrical power and that it also really has 100 lm/W efficiency).

First, you have to account for the thermal resistance in the LED itself and in the heat spreader (usually intergated with high power LEDs).

I don't know the RthJ-C value of your LED, and if it's from some nameless seller, you might never know it yourself. So for the sake of simplicity I will assume the value from a "Super-247" power transistor package (ca. 0.2 to 0.3 K/W). Transistors in this package are very powerful, they are often rated for hundreds of W of dissipation, so that value may be somewhat optimistic and the LED may turn out to be worse.

Also you'll need some sort of heat spreader, and there is some thermal interface material between the parts, so for simplicity let's take 0.5 K/W as the overall thermal resistance of your LED with heat spreader.

The maximum allowed thermal resistance of a heatsink is given by

Rth_heatsink = ((Tmax - Tamb) / Pth) - Rth_interface

that is ((the maximum junction temperature minus the ambient temperature) divided by the dissipated power) and minus the thermal resistance of the interface from the chip to the heatsink.

In this example it would be

= 1.40 K/W

This means the heatsink would need a maximum 1.4 Kelvin per Watt thermal resistance.

= 0.93 K/W

with the obvious consequence of a 0.9 K/W heatsink required instead.

very hot and would not work anyway), you would find that:

= 0.00 K/W

Which means that this physical problem has no solution because the heatsink would be required to be infinite - that is to have a thermal resistance of zero exactly. To make it work, either a lower power or a higher maximum temperature or a better thermal interface would be needed instead. You can look up maximum temperature ratings in LED datasheets, and you will find that high temperatures strongly affect both efficiency and longevity of the devices in question. Also, the thermal interface is basically a fixed constant for each type, so it leaves only the ambient temperature and the power as free variables.

more or less realistic in your case, you need a 0.9 K/W heatsink.

Take a look at Mouser and Digikey and look what 1.0 down to 0.9 K/W heatsinks look like. Here's one that claims a 0.9 K/W rating:

It's really very big and heavy. And expensive.

But it only claims 0.9 K/W when oriented vertically, in still air, in an empty room with no obstruction to airflow. Put it in any realistic mounting position and there go the claimed 0.9 K/W.

Obviously a simple flat plate would have nowhere near this rating and a couple of fins spread across the periphery won't achieve it either.

To put it simply, cooling a 50W LED without a fan is really hard.

With a fan, 50 W is at least achievable. In still air, 20 W in one spot would be a much better limit when it comes to realistic designs. More power would need better distribution, so many smaller lamps would be better than one large lamp from the thermal point of view.

Dimitrij

LED's are about 5 times the efficiency of the halogens. Still need heat sinking.

Dan

What IS the optimum wavelength for "indoor plant" growth ?>:-}

...Jim Thompson [end quote] =====================================================

I didn't write the section that Jon quoted, GH did, and in the next section of my post that wasn't quoted I said this:

Some plants only need the red, some need both blue and red. This is the datasheet for Sylvania's aquarium Gro-Lux fluorescent bulbs, that shows the action spectrum for chlorophyll and the spectral output of the Gro-Lux bulbs:

They also have Gro-Lux bulbs for plants but to me the spectra are pretty similar (and I found this one first on Google :-)). From the reading I did a few years ago the normal setup is to just use blue and red LED's whose spectral output is centered on the chlorophyll spectrum, and if you really care you need to vary the red/blue ratio to find the optimum for your plants.

----- Regards, Carl Ijames

Looking at the torchiere lamp, I realize an aluminum frying pan would make a good heat sink. A nice cheap one from a yard sale would be great. When is that 127 sale?

--

Rick

Yeah, it's a third of the house. The upstairs is only half a floor and the kitchen, bath and a third bedroom is the other half of the downstairs.

I'm not putting much faith in the eBay lumens. There are tons of vendors selling these, but I'm sure they are all just quoting the same no-name maker's numbers. Still, it should be bright compared to anything else in the room.

I am only putting $10 into the light for this, so it will be an experiment. It may or may not work well. I'm not going to spend a lot of time optimizing it until I see how it works, first on the bench. Then I'll try it in the lamp based on the bench results.

--

Rick

That was my whole idea with the fluorescent retrofit, was to spread out the LEDs and give plenty of room for heat removal. The commercial drop-in lamp replacements all pack a fantastic number of high power LEDs into as small a space as possible, to make it cheap. But, that makes the cooling problem a nightmare.

Cree seems to be doing serious long-term testing. But, yes, the Chinese, etc. are certainly not.

Cree has real data, they have apparently been running their early while LEDs for 10 years or so. They have some data that shows very good life at the "test" current, which is what I'm running my lights at. Something like 30 % worst-case reduction at 60K hours. That should be well over 10 years in our application, and I'm running them (I think!!) cooler than their test data.

Jon

So it is indirect lighting.

I would suggest installing a shelf on the high wall maybe 3-4 m from the floor.Install sockets for four 1,5 m fluorescent tubes.

Find LED replacement tubes that can be connected to the mains without ballasts. Most replacement LEDs have less than 180 degrees, which is good thing for this kind of indirect illumination, in fact 90 degrees would be even better, since all light would go directly into the ceiling and no need for reflections through the light fixture.

Some 1,5 m replacement tubes have 2100 lm output, so your ceiling would get 8400 lm total. The power would be 4x21.5 W so a significant saving compared to the 500 W halogen.

With sufficient size of the shelf, the lamps would not be visible from below, thus avoiding glare and the lamp construction would not be visible. The lamps would be in free air, reducing heat problems.

On the top side of the shelf, you could add mirrors to redirect the downward rays, if the radiation angle is large (close to 180 degrees).

Yes definitively if you insist to drive them at Imax i.e. 1A for a "3 W" LED. This makes the thermal design very hard. In addition, the light output drops rapidly after a few thousand hours due to the intense radiation to the phosphor.

Running the LED at Imax/3, the thermal design is greatly simplified and due to the lower radiation from the chip itself, the phosphor may last the claimed 50000 hours. Since efficiency is slightly increased when running below Imax, a relatively larger portion of the electric input "escapes" as visible light and hence the heat generation is used, simplifying the thermal design.

Of course about 2.5x the number of LED chips are required to create the same amount of lumens, increasing the cost.

Depending of the exact dimensions, you could even fit six 1.5 m tubes back to back, thus 12600 lm for 129 W. Alternatively seven 1.2 m tubes

7x1700 lm = 11900 lm for 133 W. These numbers were taken from the Osram catalog.

Sure enough. Cree was traditionally known for quality and they are a supplier of products for more demanding industrial uses. So is Philips (Lumileds), Osram Opto-Semi and Nichia. They tend to have reliable data and they produce some LEDs for rather extreme environments as well (e.g. Philips makes the "Altilon" LED for car headlights, rated for continuous

extreme environment for an LED).

But neither is cheap, and many people reading here will be looking on fleabay for the cheapest "xxx-watts-for-yy-bucks" noname modules of dubious provenance. That will get them neither Cree nor Philips.

So, of course, for quality products, one can get somewhat honest data (which one still has to learn how to read and interpret if one has never seen a power LED datasheet before). But my comments were mostly about those devices where the data is barely worth the ink (or disk sectors).

Dimitrij

Not really workable in this case. More than half of that wall is a balcony. Think hallway to the rooms. There is maybe 12 or so feet of wall that goes from normal ceiling height (7'6" or 8', not sure which) to the peak height. I've thought about hanging some art there or maybe mounting a small mirror in a skylight to reflect the sun to make a spot on this wall and mark it at noon on each day to create a calendar in the shape of the analemma. Certainly this could be lit the way you describe. I may try that.

--

Rick

Yeah, reading some CPSC and other info, those cheap LED drop-in lamps might even get you a fire! I think some brands had to be recalled due to more than several fires.

Jon