LED filament blade?

[profezorn]

Interestingly, the surface area of a lightbulb (~100-120 cm^2 if my google-fu is working) is pretty similar to the surface area of the blade, which would be 4 cm (length of filament) * 4 (number of filaments in a lightbulb) * 1.185 (blade diamater in cm) * PI = 94.245. However, the gas in the lightbulb is probably a decent thermal conductor, while the diffuser in the blade, is decidedly not. Using a standard diffuser will lead to a lot of thermal compression... Either way, I'm just going to have to do some experimentation to figure out how hot it really gets.

[FenixFire]

Also the plastic section of the bulb is therma-tech which has almost the same thermal conductance of aluminum. Polycarbonate is an insulator.

[profezorn]

I think "polycarbonate is an insulator" is too black and white. Polycarbonate, Epoxy and many other plastics have poor thermal conductivity, but it still conducts heat 10x better than air, and slightly better than helium. (Which is partially what's in the filament light bulbs.) Of course, light bulbs also have the added benefit of convection, which plastics do not have. Thus, encapsulating LEDs in epoxy should improve their thermal properties, but not nearly as as much as a proper heat sink. (copper conducts heat nearly 3000x better than epoxy.)

Anyways, I'm currently thinking that active cooling is the way to go. Basically, I want to mount the filament to a very narrow circuit board, cram the circuit board into an inner tube, which should leave some space above and below the circuit board. The inner tube then goes into the diffuser, which goes into the polycarbonate tube. The tip of the blade will have a drilled hole which connects to the inner tube, and the hilt will have a centrifugal fan which pulls air though the blade. It's a bit complicated, especially since I want to keep the blade removable but I'm hopeful that it will keep the filament cool enough. Also, the inner tube can do double-duty as a color filter for the LEDs, as I want my blade to be light blue.

The drawback of this approach is that the filaments will not be directly supported by the diffuser, potentially making them more susceptible to mechanical damage. I might be able to fix that by wrapping the circuit board in something, like thread or mesh.

[FenixFire]

Don't get me wrong, I am very interested in you experiments, as I am also actively working professionally on trying to find better ways to disapate the heat generated from High intensity led for commercial and residential purposes. LED filament bulb at least the ones from the leading MFG that is sister company to the one I work for now, use a mixture of primarily Xenon and Argon. BTW I work for the largest polymer company in the world...so I am familiar with the properties of especially PCs and Acrylics as well as a half dozen specialty plastics designed exclusively for disapating heat off high intensity LEDs. Hence why I was helping out letting you know in advance that a fan and airflow alone will not be enough to cool the LEDs. I like the general idea you have, I just think it could be more effective if you skipped the filaments and made your own, then you could use blue or royal blue LEDs instead of being stuck with warm white...even with filters you would probably get a greenish cyan color.

[profezorn]

I'm not sure I see any advantage (thermally) to making my own strips. I mean, if I put 25W worth of SMD leds in the blade, I'm still going to have roughly the same amount of heat to dissipate.

Pros/cons of using SMD LEDs:
+ no filters needed as I can select whatever mix of colors I want
+ can design for any voltage (with the usual constraints)
+ can easily space LEDs farther apart to reduce heat issues
- a lot more work
- you loose a little bit of efficiency since I don't have transparent circuit boards / led modules

[FenixFire]

Just finished up a call with the LED lighting engineers from France, and had a few moments to ask them about these lower current filaments in particular. Sounds like if run at the 10-15 mA range they are designed for heat mitigation may not be a significant deterrent. The thermal issues we have is that the specific SMDs we are mounting are in the 700-800 mA range so the heat is significantly higher.

[profezorn]

That's encouraging, means it *might* work.

PS: 700mA? Holy Sh*t, 40W in one filament? Lemme see, 25 of those in one blade would be 1000W, which would give off something like 130,000 lumens. That would almost be a real light saber!

PSS: I assume you guys are doing fewer LEDs than the low-power ones, but I couldn't resist some outrageous extrapolation.

[FenixFire]

Depends on the specific application, strip, flood, area, or the hottest is a spotlight application with 28+ Cree XP-D on a 8" disk. It had to use an actual glass optic because...Well lets say it gets insanely hot!

[profezorn]

So I finally got my el-cheap-o LED filaments that I ordered from ebay. Well, the ones I ordered said 50-60 volts, but the ones I got say 70-80 volts... Oh well, that's what I get for ordering from ebay.
Anyways, I have also ordered a pile of coin cells to use for experimentation, so changing the voltage is pretty easy. The good news is that while the filaments get hot, they don't seem to *mind* being hot. I wrapped one up in several layers of tape to emulate some thermal insulation and ran it for about an hour with no problem. (Apart from getting hot enough to be uncomfortable to touch.)

The usable voltage range is a lot narrower than I expected though, the leds don't even turn on until you hit 70 volt, and the light output is pretty low below 75 volt. To really get to full brightness, 80 volts is required. This means that running these directly from a stack of coin cells is not practical, as even a modest drop in battery charge will result in a lot less light. I definitely need something that can boost the output of a few li-ion cells to 80 volts. Texas Instruments has some ICs that can do that, so I'll probably just make my own board for it.

DSC07493.jpg

[profezorn]

I think I'm going to give up on the LED filaments for now, and build an LED string out of APA106 LEDs instead.
The biggest problem with the LED filaments is that I would need to design some circuit boards (one for holding the LED filaments and one for the voltage booster), and frankly I really really hate all circuit board cad software I've tried so far. Eagle seems to be the least bad, and it is horrible.

I might re-visit this later if I can find some way to design circuit boards which is less painful than a razor-eating contest.