LED Considerations, thoughts/opinions please

[james2k2]

Hi All,

Long story short, I'm in the market for a replacement LED in a product. The PSU is confirmed capable of delivering at least 2.5A @ 12V (loaded), and the LED driver board 2.1A @ 12V (loaded). The advertised LED in the product was 30W, but in reality, I'm measuring a measly 10W (yes it came from china!). This isn't unheard of in these things as I'm sure you're all aware. I've actually already ordered a replacement: http://www.cree.com/LED-Components-and-Modules/Products/XLamp/Arrays-Directional/XLamp-XHP70

The particular module I've ordered is on a 12V SinkPad II Star PCB and is binned N4 with colour tint 1B. Now by my reckoning, I should be getting anywhere between 2300lm and 3750lm in glorious daylight white.

Have I made the right choice of module? I welcome people's thoughts - including should I be turning the voltage down a touch to lower the forward current?

Thanks,
James

[Audioguru]

We do not know if the LED has a big enough heatsink to survive the maximum allowed power of 30W, probably not.
We do not know if the power supply properly regulates the current because the LED will try to draw more current as it heats which makes it hotter which makes it draw more current which makes it hotter which makes it draw more current etc and it might melt.

[edavid]

Long story short, I'm in the market for a replacement LED in a product. The PSU is confirmed capable of delivering at least 2.5A @ 12V (loaded), and the LED driver board 2.1A @ 12V (loaded). The advertised LED in the product was 30W, but in reality, I'm measuring a measly 10W...

How did you measure 10W?

[james2k2]

The LED should have enough cooling in the product. It has a 60x60mm aluminium heatsink with large fins and a fairly chunky block, and then an 80mm fan (same as the ones used in computer cases). I measured 10W by measuring the current draw and the voltage, I then used a completely separate 12V 2.5A PSU and the LED in the product drew the same amount of power. I then created a load by using 30W of CCFL tubes (it's what I had to hand). Tested this on the external PSU I have and it drew 30W. I hooked it up to the PSU directly in the product... it drew 30W. Hooked it up to the LED driver in the product... it drew 25W.

I don't have a dummy load to hand so this is the best I can do with testing. But it proves the kit is capable of producing the output. Any other thoughts?

[edavid]

I measured 10W by measuring the current draw and the voltage...

What were the voltage and current that you measured?

[james2k2]

Measured 11.96V and 0.846A. A quick VIR calc reveals 10.12W. So yeah, I rounded up/down slightly.

[edavid]

Measured 11.96V and 0.846A. A quick VIR calc reveals 10.12W. So yeah, I rounded up/down slightly.

Did you allow for the drop across your current meter?  It's quite possible that you just didn't have enough voltage to drive 30W through the LED.  Look at the forward voltage chart on page 10 of the Cree datasheet you linked.  You can see that you need well over 12V to get the rated current.

Is the LED driver board supposed to be constant current?  Does it have voltage and/or current adjustments?

[james2k2]

I did not allow for a drop across my DMM (It's a Uni-T UT139C). The LED that I currently have is an unknown. So I can only assume it's a 10/15W LED based upon my testing. I'm still awaiting for the CREE. Just curious as to why I could measure the voltages and current stated previously with a different load?

I'm quite happy to not drive the CREE LED at it's absolute maximum as I want reliability (another reason I picked this particular CREE due to it's high temp rating). So 25W is really what I'm targeting. The PSU has a rating of 12V @ 3A, and 24V @ 3A (It's a dual PSU with the 24V for the stepper motors). The PSU seems to have good quality parts including Nichicon and Samwha caps if they're legit. I think I looked up some of the other components too and found them to be better specified than required.

I'm not sure on the LED driver to be honest. It's a 12V-12V unit with a PWM input for dimming. I couldn't see any identifying parts that would indicate it's capability so that's why I tested using my other lamps (with the PWM disconnected). And achieved a pretty solid 2.100A on my meter. I didn't adjust the voltage up as the v-adj on the PSU controls both the rails together. Saying that, I still had some headroom on the 24V rail. With the 12V rail set to precisely 12V with the LED on, the 24V one was still at 23.6V. It's difficult to really know about the existing parts with so many unknowns.

FWIW, the PSU is DZ-95-2S with the brand HuanQiu.

[edavid]

I did not allow for a drop across my DMM (It's a Uni-T UT139C).

In this case, it could make a big difference.

The LED that I currently have is an unknown. So I can only assume it's a 10/15W LED based upon my testing.

Your test wasn't valid.  It's not really possible to discriminate between a 10W LED and a 30W LED non-destructively.  You would have to apply enough voltage to push 30W though the LED, and see if it blows up.

Just curious as to why I could measure the voltages and current stated previously with a different load?

Because your test load has a different I/V characteristic than the LED.  Look at the I/V graph in the Cree datasheet... you didn't apply enough voltage to drive a nominal 30W, 12V LED to full power.

[james2k2]

Because your test load has a different I/V characteristic than the LED.  Look at the I/V graph in the Cree datasheet... you didn't apply enough voltage to drive a nominal 30W, 12V LED to full power.

The existing LED won't the same as the CREE though.

Don't suppose anybody knows the burden voltage of my meter? Tried googling... Considering the high currents I'm measuring it could be quite interesting.

[james2k2]

The new led arrived, and it's easily 30% brighter than the old one (by eye side by side comparison anyway). Now being rated at 30w there's obviously room for improvement as it's only drawing an amp. I think the psu is probably at fault. The led driver doesn't appear to be more than a dimmer to be honest.

On that note, I probably need a new constant current driver. 2A would be fine, and it needs a pwm input. Does anybody have any suggestions?

Also, I measured the temperature and found the led to only be 35C when lit for an extended period. It's drawing about an amp at the moment.

[james2k2]

Hi All,

I'm getting quite frustrated that I'm not getting the kind of results I was hoping for with this new CREE. So I've been trying various PSU's. The built in one I tweaked up to about 12.7V, which when loaded is dropping to about 12.55V.

The current flow I'm getting is just above the 1A mark.

Other PSU's I've tried is a 12V 2.5A, and even a 12V 10A. The second one is far beyond the need, but based on the character graph I'm thinking it should be drawing around 1.5A, and yet I'm getting a measly 900mA. The 10A PSU was measured at 12.05V under load, and 12.6V when unloaded.

Am I measuring wrong, is my meter not giving me accurate enough results? My meter is a UNI-T UT139C

I understand that using a Constant Current driver would be ideal, but it must have PWM input, and I'm only finding ones that are capable of 700mA - I'm after around the 2A mark.

I just don't know where to look. Even better, has anybody had any experience with these XHP70 LED modules?

As a side note, I've just checked the power draw from the main socket to be only 13.5W. That was driving the LED directly from the 12V 2.5A PSU.

Thanks, James

[edavid]

Am I measuring wrong, is my meter not giving me accurate enough results? My meter is a UNI-T UT139C

Burden voltage is too high... bump up your supply voltage, or use an 0.1 ohm or lower shunt to measure the current.

I understand that using a Constant Current driver would be ideal, but it must have PWM input, and I'm only finding ones that are capable of 700mA - I'm after around the 2A mark.

If you try to use a constant voltage supply, you will probably blow up your LED.

[james2k2]

Thanks David.

You mention a constant voltage supply, but I was looking at constant current? None of my supplies I'm using are constant voltage. Would you recommend a CC driver? Also, I'm aware that you can get DC-DC CC drivers, and AC-DC CC drivers. I seem to recall reading somewhere that using an AC-DC PSU and then a DC-DC CC driver is the best solution?

Aside from the readings from my multimeter, my plug-in watt-meter is also only showing 13.5W of power draw. (Multimeter showed around 12.5W, I guess this is the burden voltage affecting this).

I just don't get how I'm finding it so difficult to drive this LED at it's max potential.

Also, I don't suppose you could provide a link to the burden voltage for my meter? I was unable to find it in the manual or the website. I know older models of UT meters have had poor burden, but was unable to find anything for my particular model.

Thanks again for your input.

[edavid]

You mention a constant voltage supply, but I was looking at constant current? None of my supplies I'm using are constant voltage.

Sorry, I don't understand what you are saying here.  Why are you talking about 12V power supplies if they are not constant voltage?

Would you recommend a CC driver?

Certainly.

Also, I'm aware that you can get DC-DC CC drivers, and AC-DC CC drivers. I seem to recall reading somewhere that using an AC-DC PSU and then a DC-DC CC driver is the best solution?

I don't see any advantage to having a separate power supply and driver.

I just don't get how I'm finding it so difficult to drive this LED at it's max potential.

I think your mistake was assuming you could drive a nominal 12V LED with a 12V power supply 

Also, I don't suppose you could provide a link to the burden voltage for my meter? I was unable to find it in the manual or the website. I know older models of UT meters have had poor burden, but was unable to find anything for my particular model.

I couldn't find anything either.  Buy or borrow another meter and measure it.

[james2k2]

Sorry, my understanding of a constant voltage would be that regardless of loaded/unloaded it would output the same voltage with no droop. My mistake if this isn't correct.

Well, I checked the I/V graph and at 12V it should draw around 1.5A. Upping the voltage to just over 12.5V does push the current draw to 1A. I think this is my annoyance is that it's not drawing the right current in relation to the characteristic graph.

The graph also shows that it should be drawing maximum current at around 12.8V, so I don't really want to go above this and risk damaging the part - it's not the cheapest of LED modules and I don't really care to keep replacing them

Thanks for confirming some of my other points.

[james2k2]

I've found this device on ebay: http://www.ebay.co.uk/itm/181865630237

So it looks to be a Buck/Boost device with the ability to set a constant current, and even has the PWM input. The PSU in my device has 2 rails: 24v as well as the 12v. So my figuring is that I could use it in Buck mode with the 24v rail (24v rail is 3A according to the spec), and the unit is only drawing 40W from that rail.

This would also allow me finer control over the voltage to the LED as well.

Thoughts please?

[HackedFridgeMagnet]

Just use a lab style PSU, as long as it has current limiting, or if it hasn't put a fuse inline.

Check your psu with an array of cheap resistors.
12v and 2 amps should be about 6 ohms and 24 watts.

If it can supply to the resistors then your psu is ok and the LED is the problem.

[Kilrah]

Sorry, my understanding of a constant voltage would be that regardless of loaded/unloaded it would output the same voltage with no droop. My mistake if this isn't correct.

Constant voltage is any normal power supply, that ideally would always supply the same voltage regardless of the load but in practice will droop more or less depending on the design because the real world isn't perfect. Regardless of the imperfections keeping a constant voltage is what it tries to do, which is what matters to describe it.

Well, I checked the I/V graph and at 12V it should draw around 1.5A. Upping the voltage to just over 12.5V does push the current draw to 1A. I think this is my annoyance is that it's not drawing the right current in relation to the characteristic graph.

Firstly that graph is specced at a 85°C, and will shift tepending on temp. Secondly forward voltage on an LED is usually approximate in the first place.

Which is why the proper way to drive an LED is using a constant current source. Current is what determines brightness, and the critical design factor on the power side of things that decides whether you'll burn your LED or not is also current. Voltage and electrical power are not set, they are a varying result of your particular sample's characteristics and conditions. You can see in the datasheet that nowhere is the LED rated for 30W electrical power, the only set value is the 2.4A current, which will give around 30W.

So if you want to use it properly you need a constant current source that will pump 2.4A into the LED, regardless of the voltage it needs to do so and resulting electrical power.

[james2k2]

Brilliant responses. Thank you!

The PSU's I've been trying are capable of 2.5A, and 10A, and then one that came in the product is 3A. The 2.5A PSU I've managed to get 2.5A from using a different load (a 12V light source that's not LED's).

The problem with the PSU in the item is that it has a single trimmer for both the 12V and 24V rails. The 24V rail is powering stepper motors, and I don't want to burn them out.

Based on your comments, I can safely push the voltage (whilst monitoring the current draw) until I'm getting the kind of power I'm after?

I'm not worried about the cooling. It has a heatsink and fan bigger than that you'd find on a CPU with a TDP of 80W, and the fan's airflow is easily higher than said cpu h/s and fan. Along with the Sinkpad II Copper base.

I did have another thought. As I'm driving the LED directly with no resistors to limit the current, is it somehow triggering some kind of short circuit protection in these PSU's and lowering the current output?

[james2k2]

Well, just to update all. I've come across this website: http://pct.cree.com/dt/index.html

It details close to what I'm experiencing with voltage/current. Looks like the datasheet is wrong. For the kind of power draw I'm expecting, I need close to 13V. I have ordered the aforementioned part, and will run it from the 24V rail and then dial in the voltage and current I want. It should give me an extra 50% light. Then will need to measure the temperature and make sure it's stable at 25W.