Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Driving XHP70 32W POWER LED near max brightness?

Status
Not open for further replies.

Triode

Well-Known Member
**broken link removed**to the page on the LED I'm talking about. I've ordered one, along with a copper sinkpad. I plan to mount it on a large CPU heatsink and put it in a projector. I've already modded the projector to not detect that it's mercury lamp is missing. I think I have the idea, but since it's a $17 component, and I need to reflow mount it, I want to be as sure as possible that I don't toast it.

here's a snippet of the specification:

Maximum Drive Current 4.8 A (6 V) 2.4 A (12 V)
Maximum Power 32 W
Light Output 4022 lm @ 32 W

I see a few options for driving this. There are very inexpensive current controlled power supplies on Ebay which say they are meant for LEDs, but I don't know how much I trust them. I do have access to a programmable load, so I could certainly get a $10 30W "LED driver" from china and see if it's ok, that would be a long wait though.

Here's the plan I would like a second set of eyes on to tell me if there is a problem:

I'm considering this basic current limiting circuit, set to 2.4A with a 19V 10A laptop power supply:


Because I have them on hand I'm thinking of using:
26K4623 for Q1
(NPN, 300V, 50MHz,
https://www.newark.com/on-semiconductor/mpsa42rlrmg/high-voltage-transistor-npn-300v/dp/26K4623

NR4941 for Q2
(50V, 30A, 75W)
https://www.digikey.com/product-detail/en/BUZ11_NR4941/BUZ11_NR4941-ND/1954150

Does this sound like a good way to do it?

By the way, am I correct in assuming that under voltage would probably not damage this device? I plan to run it in the 12V configuration but when testing my circuit I might connect it to a variable power supply and slowly bring up the voltage while monitoring the heat on the mosfet and LED, and the current across the LED. If I understand correctly, until there is enough voltage to reach this circuits target current, it will run full on.

If I just connected the LED to a 12V power supply would it probably run fine anyway? Is this kind of circuit absolutely essential, or more of an extra precaution?
 
At this power level, a constant-current driver is ABSOLUTELY essential!!!
 
Thanks, I wasn't going to skip it anyway, since I want to make sure this LED lasts, but I was curious about that.
 
Here is a sim of your proposed driver.

CreeDrv.gif


All plots are vs temperature (most important thing is not to allow the Cree to go into thermal runaway)
Green is Cree voltage.
Red is Cree current.
Violet is Cree power dissipation.
Lt. Blue is M2 dissipation.
Dk. Blue is R1 dissipation.

Comments:

Take advantage of this circuit's rather poor thermal stability to automatically reduce the current through the Cree as the Cree heatsink temperature increases by attaching Q1 to the Cree heatsink (good thermal contact).

You need a 15V supply to provide enough headroom for the constant-current circuit to work. More than 15V will increase the dissipation in M2.

Almost any power NFET that meets the power requirements will work for M2.

Most critical component is the R1; it must be 0.56Ω +- 5%

I didn't have a model of the Cree, so the one I created is approximate. I started from this: cree.gif
 
Last edited:
Wow, thanks for doing all that. I'm somewhat familiar with SPICE, but I didn't think of using it that way, by modeling the LED off the curve on the datasheet. So on top of doing the analysis for me you've taught me something I can use for future designs. Attaching Q1 to the heatsink is a cool idea, kind of a feedback loop. I'm also thinking I'll put in a thermal fuse just in case, not sure what temperature. It says the minimum solder temp is 100ºc, so probably something below that.

I just noticed there is a SPICE file on their page it contains the following:
* Cree XLAMP XHP70 12V White LED
* Model valid for 350mA to 2400mA & Tj=25C
.MODEL XHP70 12V WHT D
+ IS=5.27242E-17
+ N=11.79269
+ RS=0.54779
+ XTI=406.02879
+ EG=2.5000

I think I'll download spice and repeat what you did to refresh on it, and see if I can figure out how to import their model data. I'm sure it will yield the same results, but it will be a good exercise and I can use it to try out variants. Since I mainly work with MCUs I might later make a smart controller and use this to test what the PWM would do.

It sounds like 19V would probably waste some power and get Q2 warmer than needed. I'll see if I can get something closer to 15V.
 
Here is a running .asc file with the new model included. If you change the 2N4124 to something else, you need to resim...
 

Attachments

  • 2TLEDdriver.asc
    1.5 KB · Views: 222
If this were simply for illumination, I would recommend pulsing current through the LED as this will lower the total dissipation enormously, but if using it for a projector then pulsing would probably be inappropriate.
 
I'm not sure if pulsing would be a problem. I looked it up and these color wheels spin at 3600 to 6000 RPM, from what I can find. It has 7 segments, so that works out to changing segments at 420Hz max. So PWM at 10 times that speed, or even nearly 50x would be doable for both a PWM source and the MOSFET. I would imagine that at 10 pulses per segment pass you wouldn't see much aberration. If I really wanted to get fancy I could tie it into the color wheel encoder and sync them, though I don't think that would really be necessary.

For now I'm going to give the current limiting circuit here a try.

By the way, thanks for the updated file Mike! I'm going to give it a run as soon as I get a chance.
 
It would make a lot of sense to synchronize it if you're going to the trouble of pulsing it at all. I bet its not that hard.
 
Probably pretty easy, I work with microcontrollers to produce PWM signals very often, and the 4 pins from the encoder are exposed and easy to identify. Infact it may be possible to have the encoder pulses clock the PWM without even using an MCU, depending on the number of counts per rev.
 
...I would recommend pulsing current through the LED as this will lower the total dissipation enormously, ....

How?

If you are making the case for using a switched-mode power supply (feedback wired for constant-current) instead of a linear current regulator like was posted above, then I will agree with you. However, if you do that, because of the output filter on the SMPS, the LED doesn't see anything different than it would from a linear regulator.

How does simply "pulsing" the LED accomplish anything?
 
I built this test circuit on a breadboard. The wires from the bottom are the inputs, those going up are outputs. The blue resistor is 0.56 Ohm (marked as .47, but I measured several and found a .56), the other is the 100K.

The .56 goes from leg 1 of the BJT and Vin- to leg 3 of the MOSFET and leg 2 of the BJT. I know the circuit isn't readable from this picture, but it follows the schematic above.

To test it out without risking my LED, which I don't have yet anyway, I connected it to a 6.6 ohm 10W resistor for a load, and a variable power supply.

Results: It did provide a linear range. As I turned up the input voltage the current started to flow when I hit 4V at about 0.1A, it rose linearly till it hit 1.2A at 9V then leveled at close to 1.2A up to 15V. I didn't go higher than that because I didn't want to fry the MOSFET.

I see there's another response so I'm going to look then ask about the level of the current limit
currentlimit.jpg
 
How?

If you are making the case for using a switched-mode power supply (feedback wired for constant-current) instead of a linear current regulator like was posted above, then I will agree with you. However, if you do that, because of the output filter on the SMPS, the LED doesn't see anything different than it would from a linear regulator.

How does simply "pulsing" the LED accomplish anything?

I see your point. I was thinking LEDs can switch at very high speed. I know what high speed PWM through a motor just gets smoothed out but I figured maybe not with an LED, but of course you would put a filter on it. Which (I think) could just be a big capacitor. I may try making a switching current controller, I have about a week before my LED arrives anyway.
 
I ended up just buying a switching controller from a seller in NY, though the unit appears to be from China. I figure for $5 if it tests out ok I can hardly go wrong.
This is the unit:**broken link removed**

I still need to set up a good heat sink, I'm thinking I'll mill out a spot for the positive leads on a piece of copper and attach wires, then solder the negative and sink areas of the LED to the pad and attach the whole thing to a CPU heat sink.
 
I ended up just buying a switching controller from a seller in NY, though the unit appears to be from China. I figure for $5 if it tests out ok I can hardly go wrong.
This is the unit:**broken link removed**

I still need to set up a good heat sink, I'm thinking I'll mill out a spot for the positive leads on a piece of copper and attach wires, then solder the negative and sink areas of the LED to the pad and attach the whole thing to a CPU heat sink.
If coplanarity of heatsink interface is poor, expect worse results, but cheap CPU surplus sink& fan is the way to go.

I highly recommend you buy a sinkpad board to interface to heatsink, otherwise, you will be dissapointed with junction temp rise, aging and blue shift. Then drill tap sink for screws.
 
Last edited:
What a coincidence, I just ordered a sinkpad board before I came to check this forum. Good to have reassurance that it's a good way to go though. I have fairly limited tool bench these days, but I do have a drill and taps, so like you said I think I'll screw it down to a CPU heatsink and try to orient it so the projector fans flow through it, possibly with an additional fan if needed.
 
I hope you got 4500~5000K CCT
You still need a power source for the CC limiter.
 
Yeah, I checked what wavelength the DLP was made for. And yes, I have a switching 12V 2A power supply.
 
Are you planning to use the CC board? you can't put 12V 2A in and have 12V out with 2A CC (unless buck boost), but 16V is ok.
 
Well, I have a few I could try, I have a 19V 6A, a 15V 3A and a few 12V 2A ones.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top