Bench testing Cree XML2, doesn't match spec sheet

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
I am bench testing a cree xml2 U2 led. I have the led mounted to a heatsink and being powered by an LM350 (adjustable output) with a 12v lipo source. Using a dvom to monitor the amperage and an oscope probing the power leads going directly to the led. I am seeing 3.30V at 1A.

My concern is, the data sheet says I should be at ~2.92v at 1 amp. Is this normal because the LED is cold? If I left it on, would it start drawing more current?
 

RetroTechie

Flashlight Enthusiast
Joined
Oct 11, 2013
Messages
1,007
Location
Hengelo, NL
A schematic, or photo of your test setup with all connections visible, would help. The point is to measure current through and voltage over the LED simultaneous, and there's different ways to do that (or to be exact: you can't :duh2: ). Also if thermal contact between LED and heatsink is poor (thermal paste? ;) ), the LED could run much hotter than feeling the heatsink would suggest.

And of course there is specimen-to-specimen variation, datasheets inaccuracies etc. Which might be all there is to it.
 

DIWdiver

Flashlight Enthusiast
Joined
Jan 27, 2010
Messages
2,725
Location
Connecticut, USA
The point is to measure current through and voltage over the LED simultaneous, and there's different ways to do that (or to be exact: you can't :duh2: ).

Of course you can. You just need a voltmeter with infinite input impedance. They don't make them any more (AFAIK) but they used to. My dad has a Wheatstone Galvanometer (which I've never known anyone to use), and I gutted a Leeds & Northrup galvanometer for the case, which makes a nice toolbox. You can simulate these devices by using a voltage source to match the voltage you are measuring. You measure the voltage between them, and when it measures zero, you have the correct voltage reading and zero load on your measured device. This is exactly what the old devices did.

But a modern voltmeter with 1 or 10 megohm input impedance will be just as accurate if you are looking at currents of tens of milliamps or more. I'm sure you know that but others reading this thread may not. Using modern digital meters, measuring the voltage across a device and the current through it simultaneously with reasonable accuracy is quite easy, in fact quite trivial.

But to the OP's question, 3.3V is probably wrong. If you are measuring 3.3V at 1A, I'd guess that there's a bad connection somewhere, or you are measuring at a point too far removed from the LED to get an accurate reading (and you are using too small wire or getting poor connections).

But not definitely. The max voltage of an XM-L2 at 0.7A, 85C, is 3.5V. So it's quite possible you have a part that drops 3.3V at 1A, at whatever temp the part is running when you measure it. It's just not likely.

That's the problem with a lot of parts - the 'typical' and 'worst case' specs are so different that you have a hard time designing a product that takes advantage of 'typical' specs and still works at 'worst case'.
 

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
A schematic, or photo of your test setup with all connections visible, would help. The point is to measure current through and voltage over the LED simultaneous, and there's different ways to do that (or to be exact: you can't :duh2: ). Also if thermal contact between LED and heatsink is poor (thermal paste? ;) ), the LED could run much hotter than feeling the heatsink would suggest.

And of course there is specimen-to-specimen variation, datasheets inaccuracies etc. Which might be all there is to it.

It's got thermal paste under it.

I bought these exact ones from ebay, genuine Chinese: http://www.ebay.com/itm/221330652929

Images of setup: http://imgur.com/a/k8CxL

Images of measurements: http://imgur.com/a/GIaOR
Not sure what the 600mv VPP is?

Ran another run, got something different: http://imgur.com/a/olQPb
Something janky with my little power supply? The LM350 gets very hot
 
Last edited:

RoGuE_StreaK

Enlightened
Joined
Apr 15, 2010
Messages
615
Location
Brisbane, Australia
As an aside, why are you running it from a 3S lipo when you can tap into a single cell for ~4V? You are burning off a hell of a lot of energy for no gain?

Is the LM350 in constant current mode? Maybe something related to this thread that google popped up? I didn't read much of it, but seemed to be relating to issues with a similar setup
 

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
As an aside, why are you running it from a 3S lipo when you can tap into a single cell for ~4V? You are burning off a hell of a lot of energy for no gain?

Is the LM350 in constant current mode? Maybe something related to this thread that google popped up? I didn't read much of it, but seemed to be relating to issues with a similar setup

I guess the only gain I was seeking was keeping the lipo pack in balance. That thread you provided, the first reply was in regards to using such a high input voltage. I tested it with 2S and got a similar result (but the LM didn't heat up as much). With 1S the dropout V is too much with the LM350. I used this datasheet and it's typical application circuit, I already had it on the bread board from some bms testing so I could adjust voltage so thought I could try it out on this cree.

I am going to take it to the maker space and use an actual adjustable power supply and see may be just what I needed in the first place.
 
Last edited:

alpg88

Flashlight Enthusiast
Joined
Apr 19, 2005
Messages
5,228
that is strange, almost every xml i tested dropped about 3,3-3,4v at 3A.
it is very easy to mesure V and A at the same time, all you need is 2 multimeters.
 

DollarIn

Newly Enlightened
Joined
Mar 28, 2014
Messages
19
Are you sure they're genuine? It seems fake to me. Maybe the company has Cree PCB boards and are putting on their own LED modules.
 

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
Are you sure they're genuine? It seems fake to me. Maybe the company has Cree PCB boards and are putting on their own LED modules.

When I first got them I looked at them under a magnifier and they looked just XML2 based on the image provided here: http://flashlightwiki.com/Cree

I noted the dots and some other features, but I'm no expert. I took some more pictures of the ones I purchased, maybe you or someone can tell if they're knockoffs, album: http://imgur.com/a/TFvpl

Bigger images here:
http://i.imgur.com/xBYx4EC.jpg
http://i.imgur.com/4JxPJls.jpg
http://i.imgur.com/rdeXxIN.jpg
http://i.imgur.com/GeEbVSc.jpg
http://i.imgur.com/x657kUP.jpg
http://i.imgur.com/ZERg8tR.jpg
 

RetroTechie

Flashlight Enthusiast
Joined
Oct 11, 2013
Messages
1,007
Location
Hengelo, NL
it is very easy to mesure V and A at the same time, all you need is 2 multimeters.
Indeed. One to measure voltage directly across the LED, another to measure current through [LED + voltmeter]. Current through the voltmeter is ignored as close enough to 0 in this case. Don't see why a scope would be used here.

Another way I often use is this: LED + series resistor, resistor value measured or using a 1% / 0.5% one (metal film). Measure voltage across the LED, measure voltage across the resistor, current through resistor = current through LED is then calculated from voltage & known resistor value. That way you only need a single voltmeter for quick & accurate measurements. Also it gets around the problem that for cheap DMM's, voltage measurements are often more accurate/reliable than current measurements. :eek:oo: With a fixed voltage supply (say, 5.0V regulated) you can just swap resistors, measure voltage across LED or resistor, and calculate everything needed.

Power supply looks a bit fishy... 100 mV+ ripple voltage is a lot for a linear regulator. Maybe it's oscillating, missing some decoupling capacitors, or too long/thin wires in places? Not to mention burning 8W or so in that little cooler... Are you using the LM350 as a current source, or as a voltage regulator?
 

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
Indeed. One to measure voltage directly across the LED, another to measure current through [LED + voltmeter]. Current through the voltmeter is ignored as close enough to 0 in this case. Don't see why a scope would be used here.

Another way I often use is this: LED + series resistor, resistor value measured or using a 1% / 0.5% one (metal film). Measure voltage across the LED, measure voltage across the resistor, current through resistor = current through LED is then calculated from voltage & known resistor value. That way you only need a single voltmeter for quick & accurate measurements. Also it gets around the problem that for cheap DMM's, voltage measurements are often more accurate/reliable than current measurements. :eek:oo: With a fixed voltage supply (say, 5.0V regulated) you can just swap resistors, measure voltage across LED or resistor, and calculate everything needed.

Power supply looks a bit fishy... 100 mV+ ripple voltage is a lot for a linear regulator. Maybe it's oscillating, missing some decoupling capacitors, or too long/thin wires in places? Not to mention burning 8W or so in that little cooler... Are you using the LM350 as a current source, or as a voltage regulator?

I am assuming it is a voltage regulator. I made some changes to the power supply system, I added a 100uf cap on the output; along with the existing 0.1uf on the input and 1uf output the datasheet recommended. I changed over to some soldered jumper wires, instead of the crimped jumpers I was using in the previous tests. I also switched out the trim pot.

Please refer to this image:
http://i.imgur.com/UaHjUEu.jpg

The voltages are being taken from the ground (where the yellow jumper is connected to the large cap leg stuck in the breadboard bus), the positive from the tinned end of the soldered wire going to the led. The volt meter and oscope are within 0.02V of each other and you can see a 320mV Vpp.

I am not doubting there could be something wrong with the power supply or connections that I have not exposed yet. Or like someone eluded, these could be knock off XML2s.
 

alpg88

Flashlight Enthusiast
Joined
Apr 19, 2005
Messages
5,228
they seem to be real cree xml2, on your pics. now that i think of it, i never seen fake xml\xml2, may be they exist somewhere, but i neither seen nor heard of them
 
Last edited:

Steelmesh

Newly Enlightened
Joined
Mar 31, 2014
Messages
11
I hooked it up to a precision power supply: http://i.imgur.com/JHoUbfK.png

Meter 1 was measure Amps, Meter 2 measured the voltage at the leads soldered to the LED.

V on the chart is what I measured, PDF is what I attempted to guesstimate (why it's wiggly)

So it's about 0.1 volts higher than the spec sheet; I assume that is in spec
 

RetroTechie

Flashlight Enthusiast
Joined
Oct 11, 2013
Messages
1,007
Location
Hengelo, NL
In both setups, LED voltage includes voltage across the red/black wires that go up to the LED. Plus perhaps a contact resistance or two (crocodile clamp, breadboard contact strip). That extra voltage measured is probably low, but non-zero.

LED voltage should be measured on the pcb star, directly next to the LED (the points where red/black are soldered onto the LED pcb).
 
Top