What kind of lights should I use?
Fluorescent lights buzz. Compact fluorescents buzz less, to be sure, but they also make noise. Incandescents are much quieter, but to make them really quiet, they should run on direct current. Still, unless I use a dimmer, even on AC they are so quiet I should not worry about it.
The problems with incandescent lights are
- they generate heat
- with dimmers, they become little sound transducers and can become quite loud at low light levels
- they still generate lots of heat
- using dimmers with them is right out, since the tungsten cycle stops working
The issue of LED dimmers is still up in the air. LEDs running on DC should be completely silent, but if they run on AC or are pulsed (which is the common dimmer technology and is also supposed to increase their lifetime), they may emit audible or electronic noise. So I am investigating that. The main problem with LEDs is their cost per lumen, which in terms of initial investment is much higher than either of the others, though total cost of ownership can be the lowest.
So I have been doing research, especially into LED systems. One of the front runners seemed to be the Xtreme Linear LED light fixture. At the specified luminosity and price, it looked like I could have adequate light for about $500, which is not too bad.
However, there seem to be some problems with the data sheet. My analysis showed that one possible explanation for the contradictions would cost me $5,000 for 6,000 lumens, far beyond my budget. So I wrote them the following (rather long) letter.
Subject: Xtreme Linear Light To: Sales@theLEDlight.Com Does this product exist? If so, I am intensely interested in using a number of them to light my sound-recording studio, but I have a number of questions and problems with its data sheet. Can these lamps be used as room lighting? The 135-degree view angle certainly gives this impression. My control room has approximately 15 square meters of floor space, so illuminating this to an average level of 400 lux would require 6,000 lumens. Since the walls would also be illuminated, additional lumens would be required, but I don't require 400 lux uniformly throughout the room. According to the data sheet, using 5 XLLB2s would provide 6,400 lumens at a cost of $481.50, plus taxes, shipping, and the cost of other system components. These fixtures are clearly specified to operate at between 11 and 14 volts, DC, and to work with a (nominally) 12-volt dimmer. They are also specified to use small amounts of current, but this can not also be correct: [Table 1] mA Watts Lumens Lm/W XLLB2 350 4.2 1,280 304 XLLB3 400 4.8 1,680 350 XLLB5 750 9.0 2,620 291 XLLB6 800 9.6 3,180 331 XLLB9 1,200 14.4 3,980 276 The power-dissipation is impressively low. Too low. White light (one idealized type) is 242 lumens per radiated watt; all of these numbers would require getting more energy out of the fixture than is put in, i.e., efficiencies greater than 100%. If the light need not be white, these numbers would be remarkable but at least possible. Either the current is wrong, or the lumens are wrong, or the voltage is wrong. My guess is that the lumens are wrong. I'll try to construct plausible values below. It clearly states that there are 58 lumens per LED. Assuming that each emitter is one LED, then we see [Table 2] Lumens Lm/W XLLB2 116 27 XLLB3 174 36 XLLB5 290 32 XLLB6 348 36 XLLB9 522 36 At that rate, I would need more than 100 emitters to get 6,000 lumens. At XLLB prices, that would be somewhere around $5,000. The Lumens/Watt measured at the LED is probably better. The variation in Lumens/Watt works if you run series strings of 2 at one current (310 mA), and strings of 3 at another (400 mA); this suggests that other circuitry (limiting current) inside the fixture is dissipating some of the power; the actual LEDs are plausibly producing about 50 Lumens/Watt, given the spec of 58 Lumens and a typical dissipation of 1.16 Watts (350 mA at 3.3 volts typical, per the data sheet). However, if the full current passes through the LED as I would expect from series current-limiting, then the 400 mA strings should be brighter, and so the spec of uniform luminance suggests something else might be going on. If the luminance is uniform for all LEDs in all fixtures (neglecting variations in individual LEDs from manufacturing), and assuming that no switching power supplies (which would generate radio-frequency noise) is used, I would expect the maximum current specs to be as follows: [Table 3] mA Watts XLLB2 350 4.2 XLLB3 350 4.2 XLLB5 700 8.4 XLLB6 700 8.4 XLLB9 1,050 12.6 Note that the XLLB2 and XLLB3 would use the same amount of power, but one third the power that the XLLB3 would emit as light would be wasted as heat by the XLLB2's current regulator. Similarly for the XLLB5 and XLLB6. The Cree XLamp XR-E LED data sheet gives ratings in Lumens rather higher than 58, but the values for the 7090-XR are not given specifically, and one assumes the lower values for the warm-white color. Some additional loss may come from the 135-degree spec, assuming a diffuser is used. I can't tell from the product picture; it looks like three three yellow dots may be bare LEDs, but if so, what is the translucent part of the fixture for? If the luminance ratings are other than the rather discouraging ones in my Table 2, then the XLLB strips must use more than one LED per emitter, i.e., they would be mounted in clusters. I can not say how likely this would be, since the individual LEDs cost more than $10 in 100 quantity, but there are 6 square inches of heat-sink per emitter in these fixtures. To reach the luminance specs, you'd need something like 10 LEDs per emitter. [Table 4] Spec 10x58 XLLB2 1,280 1,160 XLLB3 1,680 1,720 XLLB5 2,620 2,900 XLLB6 3,180 3,480 XLLB9 3,980 5,220 A 100W incandescent light bulb typically radiates about 1,700 lumens. (Yes, I know the incandescent may deliver fewer useful lumens, depending on the fixture.) This is comparable in light output to the XLLB3 and is only twice as much power. A compact fluorescent with similar luminance uses only 23 watts, and is therefore twice as efficient as the LED. However, only about 35 watts of the XLLB3's (or 23 watts of the XLLB2's, i.e., 1.15 watts typical per LED) output are actually used to emit light. The rest of the power is used by the current limiter. For my purposes, it would help if I could put the current source outside the studio, where waste heat is a particular problem. At the top of the Xtreme Linear Light web page, it says that I should always use a dimmer, because its pulse-width-modulation component should extend the life of the LEDs. LED lighting interests me because LEDs are completely silent, as DC-powered incandescent bulbs would be, and unlike fluorescent or AC-driven incandescent bulbs. A dimmer module could cause the LED fixture to emit sound and electromagnetic noise, and so the frequency of those pulses becomes an important factor, and the wires must be shielded. (At least the strobe effect would not be important in a sound-recording studio.) Is it really the case that these fixtures sold without a dimmer are not warranted? How does the dimmer module affect the efficiency and lifetime of the LEDs? Can you point me at some reference material on this? Could a DC dimmer be implemented in the current limiter instead? The brightness of the LED should vary with the current. It looks like reading color codes (e.g., on electronic components) would benefit from adding some red and green light to compensate for the white LED spectrum, which has a narrow peak in the blue and a wider peak in the yellow, but is definitely deficient in the red and green. The diffusing fixtures, however, are not available in any colors other than white. What would you recommend? -- Spencer
This was their reply:
From: the LED Light Subject: Re: Xtreme Linear Light Spencer, You are too much smarter than the engineers who designed the Xtremes. Therefore, we cannot sell the product to you because you will not be happy with them. Regards, The LED Light, Inc. Ph 775-841-4490 Fax 775-841-4491 http://www.theledlight.com/index.html
That's what I get for writing such a long letter, I guess.
(I have redacted the personal name of the respondent.)