Blowing Cree's left and right - help!

[blasterman]

The past month or so I've been losing a lot of Cree XR-Es on the test bench, and have been banging my head trying to figure out what the problem is. I now have suspicion what it is, but would like some advice from some others here before I approach LedSupply on the issue. I want to make sure I've ruled out every other possibility.

A couple months ago I bought a 350mA LighTech driver from LedSupply. The driver, has worked fine in the arrays I've built, but it seems to be the culprit behind my high death rate of Crees, but it's not consistent.
Originally I thought I was doing something wrong, but after yet another XR-E died half a second after I pulled it out of the shipping box I noticed the common thread. Previously to using the driver I opted for 12volt supplies and power resistors, and I've never lost an LED that way...provided I remembered to use the right series of resistors. However, *every* Cree I've lost has been *only* when connected to the LightTech driver, and they always die within a few seconds of being connected to it. There's a brief '*spark*, and the LED refuses to light again.

What made the trouble-shooting even tougher is that I've hooked the LighTech up to an even larger variety of 3Watt BestHongKong ProLights, Luxeon IIIs, and K2's. *None* of them have died - it's only Crees, and only certain ones. Today I got Red K2 and a Blue Cree XR-E to test. The K2 fired up fine with the LighTech. The Cree sparked and died within a fraction of a second.

WTF? Am I supposed to use power resistor with a 350mA dedicated driver on a XR-E? Seriously. What am I missing here.

Taking all the variables into account, the common culprit is the LighTech combined with Crees, and this includes Warm-Whites, Neutral White's and Blues. I've never lost a Cree using 12-volt supplies with proper resistors, and the LighTech has never 'popped' another brand of 3-watt LED. When connected with several in series, the LighTech has never 'popped' a Cree.

This leads me to initially conclude (and realize I'm a novice when it comes to electronics) the LighTech driver is either defective, and/or when is initially connected to a single Cree isn't regulated correctly, or quickly enough.

I have to also conclude there's enough deviation with XR-Es themselves that some are more resistant to the problem given I have several R2's and Warm-Whites that work just fine with the LighTech.

Is there anything structural with XR-Es that would make them more delicate with the Driver?

Should I assume the driver is defective even though it powers up other LED types just fine?

 

[spencer]

It probably doesn't have open circuit protection. That means when its plugged in and nothing is connected it charges to x number of volts and when you connect the LED, x number of volts are discharged all at once. I know the CCHIPO charges to 48 volts I believe and if you read the website it says this and a warning that you could kill your LEDs. The Crees must be more sensitive than other brands to this voltage spike.

 

[csshih]

have you even tried measuring the voltage output from the leads with a DMM?

 

[blasterman]

and if you read the website it says this and a warning

Could you link that please?

have you even tried measuring the voltage output from the leads with a DMM?

No, because if there *is* a problem wouldn't the voltage spike be transient anyways?

 

[spencer]

Could you link that please?

I meant for the CCHIPO is says this. I would link to that page but the site is down and has been for awhile.

 

[JohnR66]

You'd have to scope the output of the driver. Working with switching type supplies, I've seen switching transients you can't see with a DVM. You might see big spikes, negative going pulses and such.

 

[Mr_Light]

I fried a couple of CREEs when I tried testing them by connecting them to an Xitanium driver that was already plugged in. I think that constant current drivers build up too high an initial voltage if they are open circuit before being connected to the LED. I think that as long as you hook up the LED to the driver before you connect it to power you should be ok.

 

[joeparker54]

To be on the safe side i would still preload the output with a resistor, then hook LED parallel to it, then remove the resistor. It would take minutes to make with a small piece of perfboard/breadboard, a resistor and a couple switches of your choice. For someone poor like me, you can never take enough precautions! Mistakes cost too much time and money that I don't have.

 

[yellow]

from blowing drivers without connected led, I am used to do all the wiring an THEN power up the driver, not the other way round.

Did You power the thing and then add the led?
Would also think there was a much too high spike then

 

[blasterman]

I think that constant current drivers build up too high an initial voltage if they are open circuit before being connected to the LED.

Ditto on your theory - that's my conclusion as well. The fact you encountered the same problem with Xitanium brings me to the other conclusion that switching drivers won't fix the problem because LEDs drivers are pretty much designed the same.

I think that as long as you hook up the LED to the driver before you connect it to power you should be ok.

I'm vaguely certain all the LED's I've lost we're blown by contact with a powered up driver, but not 100% sure. Having a load on the driver when it's first powered up *might* help temper the initial discharge, or using a resistor of course. Still, I wish this issue were given a proper warning so it could be avoided, and the fact it only happens with Cree......grrrrrr.

I'll now stick to 12volt PSUs and resistors for initial testing and reserve LED drivers for wired configs. Not sure of how to approach this issue with LEDsupply because the reason for the blown Crees would now seem to be a grey area in terms of who's at fault.

 

[tebore]

Not sure of how to approach this issue with LEDsupply because the reason for the blown Crees would now seem to be a grey area in terms of who's at fault.

If you're powering the driver before you power connect the LEDs to the driver then it would be you.:laughing: Having a load on the driver would have and "not might" have prevented all those deaths. Older LED drivers would have just blown when no load.

I'm surprised you didn't blow a RED K2. I figured a Red LED would be easier to kill since it's got a lower vf then your average white LED. Well at least you've proved what Philips has been pushing and that is the K2 is a more robust LED than Cree even if they are less efficient.

 

[jtr1962]

It sounds like the problem is indeed connecting an LED to a powered up driver. The output on many (most?) constant current drivers rises to approximately the input voltage in the absense of a load. Not a problem connecting an LED if there is no capacitor on the output. However, some drivers do indeed have an output cap to smooth the output so it's close to DC. I know the drivers I design usually do. Anyway, when connecting an LED to a powered-up driver with a output cap what happens is the cap will discharge from 12V or whatever down to the LED's Vf. This produces a current spike (which is actually an energy spike) in the LED, perhaps as high as some tens of amps. Apparently the Crees can't deal with this much energy. Other LEDs may have more area to sink the energy spike, or just thicker bond wires which short term can absorb the energy without failing.

Bottom line-never connect an LED to a powered-up driver. I've been advising my customers not to do so with my drivers from day one.

 

[blasterman]

Having a load on the driver would have and "not might" have prevented all those deaths. Older LED drivers would have just blown when no load.......

I'm surprised you didn't blow a RED K2. I figured a Red LED would be easier to kill since it's got a lower vf then your average white LED.

Recently I 'hot' tested several red Chinese ProLights with a Vf of 2.2 when I built this fixture. No problem with the Chinese LEDs either, but the Blue Cree 'fizzled' on contact. So, if your assumption was wrong that lower Vf should make the LED more prone to blowing when 'hot' connected (as per the ill advised fashion we now pretty much agree on is causing the problem) then I'm logically a bit hesitant to just blindly believe the first part of your statement.

So, it's 'me bad' for not being aware that 'hot' connecting LEDs was bad and I'll avoid doing it at risk of my own expense. However, the fact that only Crees are sensitive to this brings me to the conclusion that Crees might have other issues in terms of long term longevity.... even though such a claim might be considered blasphemy around here. What-ever makes the Crees more sensitive to the power spike *has* to be relevant over long term use - especially if there's frequent power cycles.

Thanks for the explanation jtr1962. This might also explain why fully rectified LED bulbs seem to have higher failure rates than simple strings.

 

[LukeA]

What-ever makes the Crees more sensitive to the power spike *has* to be relevant over long term use - especially if there's frequent power cycles.

That strikes me rather like saying the wire going to my table lamp won't last as long as a 0 AWG power line because the 18 AWG wire going to the lamp can't handle 100A.

 

[yellow]

K2 is built to run on double the current continuously, than the Cree's max is.
Any wonder such a device stands a spike better?
:thinking:

PS: I dont see it a "failure" of any of the led mentionned to die, when being connected to a driver that already is powered up.
Users fault no matter what.
Too bad about the money wasted.

 

[kuksul08]

Maybe you should change your username from blasterman and they won't burn out anymore

 

[LEDriver]

Blasterman - I am famiiar with the Lightech drivers, and would like to explore this further, but I'd like to take this offline. Please send me a Private Message, so we can discuss.
LEDriver

 

[saabluster]

Blasterman - I am famiiar with the Lightech drivers, and would like to explore this further, but I'd like to take this offline. Please send me a Private Message, so we can discuss.
LEDriver

I do not believe you have PM privileges yet as you do not have enough posts.

 

[jonboy79]

Sorry to bring up an old thread, but wouldnt this also be the same if you had a relay or switch after the driver, it would be better to place a switch before the driver, extra wiring on the 240v side but nothing to drastic.