Retrofitting an incand uplight. Bridgelux or Cree CXA2011?

Retrofitting a torchiere-style uplight for LED. It’s designed for A19-style bulbs (the reflector is deeper than a halogen-style torchiere) so there should be plenty of room for a driver and an adequate heat sink. Judging from the output of a (separate) 300W halogen torchiere, I’m going to shoot for something in the 4500-5000lm range with CRI > 90. Dimmable, of course. For simplicity, I think I’m going to restrict myself to LED arrays that can be screw-mounted rather than SMT-style packages. Since I’m looking to do an uplight to illuminate a vaulted ceiling, I’m not sure I need to bother with LED reflectors or lenses though I’ll probably want some sort of diffuser.

I figure I have a few options:

1) Single emitter. The Bridgelux RS series BXRA-30G4000-H-00 can both be had for around $40 and is rated at 5700lm with CRI > 90. Rated at 24.4V at up to 3A. Cree’s arrays in this output range appear to be surface-mount so that nixes them from consideration.

2) 2-3 smaller emitters in the 2000-3000lm range. Bridgelux RS and ES (37.4V @ 1A) and Cree CXA2011 (40V @ 1A) have some high-CRI candidates that appear to meet the above requirements. Haven’t looked at OSRAM or others.

Anyone have experience with Cree CXA2011 vs Bridgelux? Any recommendations for power supplies/drivers?

Bridgelux is easier way to go but I’m quite sure that A-19 bulb format fixture does not provide enough space for adequate cooling without active (fan) help. I haven’t checked data sheets but bridgelux is probably less efficient than cree combination but bridgelux should have better light quality (despite similar CRI).

Pictures of light fixture and room would help a lot…

’fraid I don’t have any pics of the room. The lamp itself is a generic torchiere-style uplight similar to this one . The reflector bowl for mine measures about 13” diameter and about 4” deep.

After seeing passively-cooled DIY reef lights with a couple hundred watts of LED power, I’d hoped to be able to rely on passive cooling for this project but you’re probably right. Even if I were to mount the LED to an aluminum block 13” in diameter and a couple inches thick, there’s probably not enough mass there to keep a 60-70W LED cool for hours on end. I do have a few dozen fan-cooled heatsinks upstairs in a closet that can handle a 70W load so in principle active cooling is doable though powering the fan becomes an issue…

Mass is irrelevant, surface area is what matters. Bridgelux recommends at least 10 square inches per watt. Also check out the Nuventix active LED coolers.

Nice beefy heat sink with vertical slits and large diameter fan (8+ cm) would be a good choice (CPU heatsink). You can power the fan with separate cheap plug in ac adapter. Fan does not have to work at high speed just enough to help things moving. Nevertheless I would recommend to use at least basic thermal fuse mounted on heat sink to protect expensive LED array in case of fan failure.

Another thing to consider about torchiere style lights is just how unstable they are. LOTS of houses have burned to the ground when a halogen torchiere tipped over and set fire to the world. Add a few pounds of heat sink up there and you can expect even more stability problems.

Good point about the torchiere stability problems. ’tis one of the reasons why I stopped using a halogen torchiere years ago and switched to an A19-style with a CFL. Doesn’t eliminate the tipover risk, though. I simply had this lamp sitting unused in a corner so it became my candidate for LED conversion but I’m open to other ideas. I’m very much a fan of indirect lighting and it seems that consumers don’t have a lot of choices. Torchiere floor lamps, wall sconces and track lights are about all that’s available (and affordable) when it comes to uplighting. I suppose a ceiling fan could be modified to provide an uplight though the LED housing would have to be mounted above the motor/blades to avoid the strobe effect. That would be a fairly complicated modification.

Two thoughts sprung to mind as I looked at the Nuventix stuff. First, the Synjet looks interesting though I couldn’t find any real performance specs (CFM, static pressure, etc). Second, their heat sink designs seem a little primitive. No heat pipes or vapor chambers…and so the performance is fairly poor. 0.49 C/W for an active-cooled heat sink measuring 5.5” x 4.5” x 1.5” is about where CPU cooling was 15 years ago. (For comparison, the Dynatron T318 is a solid copper heat sink w/heat pipe for 1U servers that measures around 4” x 3” x 1” and can cool a 130W CPU at around 0.20 C/W at 20CFM for around $30)

Then it occurred to me: I wonder if there’s much exchange of ideas between heat sink designers for different industries? I’d hate to think the high-powered LED cooler folks are going to spend the next 10-15 years evolving towards current CPU cooler designs (They share similar design requirements: keep a highly-localized heat source below, say, 85C. As LEDs catch up to CPUs and GPUs in terms of power dissipation, it seems reasonable that their coolers will converge on similar designs).

The driving issue in active cooling LED lights is noise. The Nuventix stuff is designed for ultra low noise.

The Philips 10W MR16 bulbs have a tiny fan in them. I can’t hear the fan over a foot away, but one has to be aware of resonance effects in the housing you put it in. The fan is much more noticeable if I stick the bulb in my integrating sphere.

Large (8-10 cm) fan rotating at less then 1k RPM should be quite noiseless especially since air flow won’t be much restricted…