Bridgelux, XP-G etc electrically isolated -- Really?

[kethd]

Many power LEDs have "electrically isolated" heatsink bases. Supposedly.

In the case of Bridgelux, they mention this in passing in AN14. There don't seem to be any details anywhere.

The Cree XP-G datasheet is pathetic -- the mechanical drawing shows three areas on the bottom, two corner pads on the top, with no markings or annotations at all as to the electrical connections.

If we mount a big series string on a common heatsink, which might be very tempting, the voltage stress via this potential electrical conduction path could get up to over a hundred volts.

What are the safe-design guidelines for this? How good is this electrical isolation, really -- in Bridgelux or any other power LEDs?

(Interesting find of the day: How to build your DIY LED array
http://www.gardenscure.com/420/lighting/117772-how-build-your-diy-led-array.html
by KNNA, thread started in 2008
A complete-from-scratch description of building large arrays of crude power LED lighting bars; good intro and lots of educational photos. Interesting approach to practical problems of mounting emitters. Intriguing mention in passing of using water-cooled copper pipe heatsinks.)

 

[AnAppleSnail]

What are the safe-design guidelines for this? How good is this electrical isolation, really -- in Bridgelux or any other power LEDs?

Hadn't thought of it that way. All I've got to look at are XR-Es, which do not have an electrical connection between the metal ring and the thermal path. Even "negligibly electrically conductive" thermal epoxies warn against using the product near pins and leads. If those problems are due to capacitance it might not be due to current flowing, but if it IS from current flowing, 100+V will be interesting.

Anyone wanna see what happens to a Cree with high potential between a contact and the thermal pad? I can test resistance of some that I've already killed. I can't measure the resistance between the thermal pad and either contact - although at 100V, who knows?

 

[purduephotog]

Many power LEDs have "electrically isolated" heatsink bases. Supposedly.

In the case of Bridgelux, they mention this in passing in AN14. There don't seem to be any details anywhere.

The Cree XP-G datasheet is pathetic -- the mechanical drawing shows three areas on the bottom, two corner pads on the top, with no markings or annotations at all as to the electrical connections.

If we mount a big series string on a common heatsink, which might be very tempting, the voltage stress via this potential electrical conduction path could get up to over a hundred volts.

What are the safe-design guidelines for this? How good is this electrical isolation, really -- in Bridgelux or any other power LEDs?

(Interesting find of the day: How to build your DIY LED array
http://www.gardenscure.com/420/lighting/117772-how-build-your-diy-led-array.html
by KNNA, thread started in 2008
A complete-from-scratch description of building large arrays of crude power LED lighting bars; good intro and lots of educational photos. Interesting approach to practical problems of mounting emitters. Intriguing mention in passing of using water-cooled copper pipe heatsinks.)

The only unit I've ever had trouble with is LedEngin. The latest 'reason' was that the dieletric was breaking down when soldering.

*shrug* I've been waiting now 3 weeks to hear what temperature they said it breaks down at.

 

[Optical Inferno]

I do agree that the Cree datasheets are a little cumbersome at times always leaving you wanting more, but with careful viewing you can figure it out.

Also, Cree tends to put the thermal pad characteristic on the first page under the features section (it's there for the XP-G) but your right a small diagram would be nice. The three images that they use to show the top,side, and bottom mean the LED is being rotated about one axis for each pic. Therefore the anode is on the left pad in the datasheet image.

I do agree that there could be a little more info regarding this. but I think they assume you have already used the XP-C and XP-E. I am also interested in the dielectric strength of the isolation. You may be able to get that info from a cree dealer like Arrow or Avnet.

Can't speak for BridgeLux though...never used them before.

 

[kethd]

Do we get any monitoring/responses here at CPF from Cree, Bridgelux, etc staff or do I need to try to contact them officially?

 

[bshanahan14rulz]

I'm sure there's a distance to voltage coefficient you can use or look up somewhere. Just measure the distance between the thermal pad and the closest contact, and assuming air, see what the voltage would have to be to induce arcing.

 

[kethd]

I'm sure there's a distance to voltage coefficient you can use or look up somewhere. Just measure the distance between the thermal pad and the closest contact, and assuming air, see what the voltage would have to be to induce arcing.

My main interest is in the internal electrical isolation, inside the LED, between the die and the base heatsink surface. When two or more such LEDs are mounted directly on a shared metal heatsink, those two respective internal isolation layers in series are providing all of the electrical isolation between the two die junctions.

 

[AnAppleSnail]

My main interest is in the internal electrical isolation, inside the LED, between the die and the base heatsink surface. When two or more such LEDs are mounted directly on a shared metal heatsink, those two respective internal isolation layers in series are providing all of the electrical isolation between the two die junctions.

Most particularly, there are two [Solder Pad to heatsink] connections keeping voltage from flowing. At what point is voltage high enough to exceed that? And if you solder the thermal pad on, at what point could the voltage between the solder pad (heatsink) and one of the contacts be high enough to damage the LED?

 

[bshanahan14rulz]

I think that ceramic substrate is as thick, if not thicker than, the gap between the heat exchange pad and the anode or cathode pad. With Cree, the heat pad is just there so that reflow soldering can attach to the LED to create better heat exchange. The actual contact pads have vias through the ceramic.

Edit: if you look at the LED from the side, you will see how thick the ceramic substrate, or "isolation layer" actually is.

 

[kethd]

(The "Homemade AC LED Driver Design" thread active now mentions this http://www.powerint.com/sites/default/files/PDFFiles/der186.pdf interesting non-isolated driver. Non-isolated drivers make exact understanding of "electrically isolated" LEDs very important for safety reasons -- or else you could end up with a heatsink that electrocutes someone.

I am new to CPF. I know there are concerns about double-posting, cross-posting, etc. These two threads now have closely related subject matter, one focusing on the electrical isolation inside some LEDs, the other AC driver design. Readers of each thread should be made aware of the other thread. What are the right ways to do that around here?)