10 watt LED and power supply

[slburris]

More neat stuff I picked up from Sure electronics.  A 10watt LED on a heatsink
plus a 1.5amp constant current supply:

http://www.sureelectronics.net/goods.php?id=340
http://www.sureelectronics.net/goods.php?id=1125

The LED I received looks essentially like it does on their website.  No markings
on it that I could see.  I removed the screws and there was heat sink compound
between it and the heat sink.  No idea who actually makes this LED, could it be
from Seoul Semiconductor?  Looks like there 4 dice inside.

The power supply is actually two converter boards.  They are paralleled
on a carrier board to provide the 1.5amp constant current.  The carrier
board places them a little too close together.  The converters don't both
fit well, so either these boards need to be trimmed more, or the carrier board
isn't laid out correctly.  Still it's possible to smash them together as shown.

I've tested the LED with an external supply -- it is very bright.  I accidentally
looked at the LED while it was on, and I was seeing afterimages for about 10
minutes.  The heat sink didn't seem to get very warm, so that's encouraging.

Next up, I need to test the converters.  My plan is to build this into a floodlight
for some nice LED lanscape lighting.  I'll use this as a wall washer type light.
Two LEDs should be about 900 lumens -- I think that's sufficient for what I have
in mind.

Scott

[slburris]

Time for some testing!

I soldered the boards into their carrier, wired up a power supply and connected the
output to the 10w LED on a sink.

In the first picture, you can see that the supply is providing 1.238amp to the LED,
measured with my UniTrend UT70D meter.  I put a large washer with a hole over
the LED, otherwise it hurts my eyes to be this close.

Next, you can see the power supply which is feeding this.  11.98v at 0.956amp.

Things are heating up, so let's pull out the IR thermometer.

Continued in next post.....

Scott

[slburris]

In the first photo, we're measuring the temperature of the heatsink, 89F.

The second and third photos are the temperature at the middle of each
of the two buck converter boards.  They seem to be be pretty evenly matched.

The last photo is the setup with the washer removed.  Although the LED looks
bright, the iPhone camera doesn't have the dynamic range to fully capture how
bright this is compared to its surroundings.

All in all, I'm pretty happy with both the LED and the power supply.  Next I'll
be integrating it into a floodlight case and see how it looks as a landscape light.

Scott

[Zero999]

have a question... why this LED need additional power supply board to work on? cant we directly connect to battery?

Because it requires a constant current source and a battery gives constant voltage. You can use a power resistor but the efficiency will be poor and the LED will dim as the battery voltage drops. A SMPs such as the once shown above wastes little power and the LED stays the same brightness until the battery voltage drops too low for it to work properly.

[slburris]

good report, keep it coming! 1A@12V is quite a monster LED!

Note that LED sees about 8 volts at 1.2amps, which is close to 10 watts.
The forward voltage of the LED is 8 volts (I think because there are 4 LEDs
wired in series).  The switcher has a constant current output, instead of the
usual constant voltage output, which is what you want for driving LEDs.

Scott

[Zero999]

thanx for the answer. but i still dont get it. the board specified as 1.5A constant current, but slburris only measuring 1.2A. what the heck?! sorry if this is a nonsense question.

Well the minimum current is specified at 1.3A and he measured 1.238A which is still a little low so perhaps his meter isn't very good? Either way, it doesn't surprise me it doesn't match up to the specification, being cheap and Chinese.

[slburris]

thanx for the answer. but i still dont get it. the board specified as 1.5A constant current, but slburris only measuring 1.2A. what the heck?! sorry if this is a nonsense question.

Well the minimum current is specified at 1.3A and he measured 1.238A which is still a little low so perhaps his meter isn't very good? Either way, it doesn't surprise me it doesn't match up to the specification, being cheap and Chinese.

No, the problem is that at 12 volts input with an 8 volt LED the constant current supply doesn't
have enough headroom to supply the specified current.   If I increase the voltage to
around 15 voilts or higher, I get 1.5 amps. 

I haven't studied the design well enough to figure out why this is, but for my landscaping application,
1.2amps is fine.


Scott

[Kiriakos-GR]

the constant current supply doesn't have enough headroom to supply the specified current. 

That's logical and it happens all the time ,
because  we always forget to calculate the energy losses in the equitation.
I bet that even the carrier board gets hot too !! 

[Zero999]

That makes sense.

To be honest, I thought it could've been that but I thought a drop out voltage of over 4V was too poor.

I can't think why, the data sheet for the MBI6651 implies it uses a low side driver, had it been an N-MOSFET switching the high side it would've made sense. It's probably something obvious I've missed.

[Mechatrommer]

That's logical and it happens all the time ,
because  we always forget to calculate the energy losses in the equitation.
I bet that even the carrier board gets hot too !!  

yes the energy loses, but..... when it says 1.5A out, then it should be 1.5A out, i dont care, they should already calculated it for us. but the cheapo unbranded reason is acceptable.

[Kiriakos-GR]

yes the energy loses, but..... when it says 1.5A out, then it should be 1.5A out, i dont care, they should already calculated it for us. but the cheapo unbranded reason is acceptable.

Blame the white papers , and the reference boards ....         

[slburris]

Sure Electronics has this little graph on their website:



Which seems to match my experience pretty closely.

Scott

[Kiriakos-GR]

Wait a minute Scott , just noticed that you are using testing cables with crocks !!

Well that's an mistake, causing extra losses ...  Use proper cable at list 0.75mm diameter ..

[Zero999]

Sure Electronics has this little graph on their website:



Which seems to match my experience pretty closely.

Scott

You say little graph, I can hardly read that.

Apart from being too small it's saved in low quality JPG format which is fuzzy, especially when zoomed in, which might be deliberate.

[Simon]

66% efficiency does not sound very good, considering the controllers are targeted at a specific current they will always operate at they should have nailed it at 80-90% for the target output

[Kiriakos-GR]

With out the use of proper wires - cables ..  what ever , any measurements are inaccurate .

For 1.5 A constant juice , 1mm or 1,5mm car cables are the minimum diameter , that will keep them cold enough, so to not have an high resistance or thermal losses ..   

[slburris]

66% efficiency does not sound very good, considering the controllers are targeted at a specific current they will always operate at they should have nailed it at 80-90% for the target output

I'm not sure I follow your calculations.

The input is 11.98v @ 0.956a, which is 11.54 watts

Output is 8v (if the datasheet on Vf for the LED is right) @ 1.238a, which is 9.90 watts

Dividing 9.90 by 11.54, we get 0.857 or 85.7%

With alligator clips and less than ideal wire sizes, this is a lower bound on
efficiency.  Wired into the landscape system, which is 16gauge wire --
1 mm for the metric world (everyone but the US it seems),
I don't expect any issues.

Scott

[Kiriakos-GR]

With alligator clips and less than ideal wire sizes, this is a lower bound on
efficiency.  Wired into the landscape system, which is 16gauge wire --
1 mm for the metric world (everyone but the US it seems),
I don't expect any issues.

Scott

I do not see why, just the words are stopping you from testing it , with better cables.

[Simon]

66% efficiency does not sound very good, considering the controllers are targeted at a specific current they will always operate at they should have nailed it at 80-90% for the target output

I'm not sure I follow your calculations.

The input is 11.98v @ 0.956a, which is 11.54 watts

Output is 8v (if the datasheet on Vf for the LED is right) @ 1.238a, which is 9.90 watts

Dividing 9.90 by 11.54, we get 0.857 or 85.7%

With alligator clips and less than ideal wire sizes, this is a lower bound on
efficiency.  Wired into the landscape system, which is 16gauge wire --
1 mm for the metric world (everyone but the US it seems),
I don't expect any issues.

Scott

yes sorry your right I forgot to consider the lower current in, another silly late night post

[Zero999]

With out the use of proper wires - cables ..  what ever , any measurements are inaccurate .

For 1.5 A constant juice , 1mm or 1,5mm car cables are the minimum diameter , that will keep them cold enough, so to not have an high resistance or thermal losses ..   

It depends on the length of the cables.

Have you actually done any calculations or are you just guessing?

0.5mm² should be more than enough to satisfy the thermal requirements and shouldn't effect measurements too much for short runs (under a meter).

The resistance of a two core 0.5mm² piece of cable is 93m? per meter, if the current is 1.5A the loss will be 209mW, if the power dissipated by the LED is 10W the inaccuracy due to the cable is just under 2.1%.

Even better, measure the voltages directly at the board (which is what he's doing according to the photograph), that way the cables will be taken out of the equation.