Lexing 6.5W/ 12V /600 lumen MR16 LED Spotlight

[hgg]

Hi everyone,

I have just received some 12V spotlights from DX for a solar lighting project I am working on.
Very simple, every room will have one of these lights together with a tiny motion sensor and
whenever motion is detected the specific spotlight will turn on and stay on until motion is
been detected again from a different sensor.

The spotlights are the following:  (Warm White Version)
http://www.dx.com/p/lexing-6-5w-600lm-gu5-3-mr16-48-smd-2835-warm-white-light-spotlight-12v-272970#.VFOAP1eF8VE


Lexing

Its a 12V bulb and it was listed as 6.5W / 600 lumen and so I decided to buy some.

I took some measurements and here are the results: 
(lux were measured at a distance of exactly 1 meter except the Lux/Volts graph which was at 50cm.
60 data points / during 20 minutes.)


Testing

Another typical Chinese product with false specifications...
As you can see from the first graph, the light starts with 225 lux and after 8 minutes the brightness
drops 'vertically' to 147 lux...  It starts with 7.3Watts and after 8 minutes it uses 4.2Watts.
Finally it stabilizes after 20 minutes with a brightness of 134lux and 3.9Watts which are the actual specs...

3.9 Watts / 12V / 134lux actual   or    6.5Watts / 225lux for the first 10 seconds...

Lux vs Current is linear but interestingly enough the light peaks in brightness at 11V as you can see at
the third graph, and not at its rated 12V.

The temperature of the heatsink after 20 minutes is 57 degrees Celsius with an ambient temp of 19d.
(134F/66F)

On the positive side, the bulb has a very pleasing warm color, no detectable green, completely flicker
free, very wide angle (its not a spotlight), smooth outer edge fading and of a decent construction
quality.  The packaging also was very good with a hard plastic inside, protecting the bulb. 

If you want a 4 Watt bulb... its not a bad choice.


My question is, do you know what causes this drop in brightness & current as well?
Is this common with that kind of LED bulbs?
Can you suggest a better 12V bulb with all the positives but with more lumen?

Regards,
George.

[LabSpokane]

Heat

[Rick Law]

Hi everyone,

I have just received some 12V spotlights from DX for a solar lighting project I am working on...
..Its a 12V bulb and it was listed as 6.5W / 600 lumen and so I decided to buy some...

...I took some measurements and here are the results: 
(lux were measured at a distance of exactly 1 meter except the Lux/Volts graph which was at 50cm.
60 data points / during 20 minutes.)
....
...Another typical Chinese product with false specifications...
...As you can see from the first graph, the light starts with 225 lux and after 8 minutes the brightness
drops 'vertically' to 147 lux...  It starts with 7.3Watts and after 8 minutes it uses 4.2Watts.
Finally it stabilizes after 20 minutes with a brightness of 134lux and 3.9Watts which are the actual specs...

3.9 Watts / 12V / 134lux actual   or    6.5Watts / 225lux for the first 10 seconds...
...
If you want a 4 Watt bulb... its not a bad choice.
...
Regards,
George.

[ quote with bold added ]

How did you come to that conclusion (false specs)?

You measured Lux, not Lumens.  Your distance is 1 meter away, but lux is one lumen per square meter.

The bulb went from 7W-4W.  So 6W could be some kind of average.  Some countries requires posted Watt/Amp being max it could draw(to ensure you are not exceeding the outlet's max), whereas, some countries uses average or typical watt/amp to show what it consumes.

Try remeasuring lumen instead of lux, then you can compare apples to apples. 

It would be interesting to see if it is off, and if so, how far off.

Rick

[hgg]

The bulb went from 7W-4W.  So 6W could be some kind of average

@Rick
Please look at the first graph more carefully.

The current consumption starts at 7Watts, after only 6 minutes it drops to 4.5Watts, after 20 minutes
it drops to 3.9Watts but then it stays there indefinitely.  The stated 6.5Watts is not an average value.
The average value is 4Watts.  So at least in the case of current consumption the specs are clearly false...

Concerning the brightness I have no way of accurately measuring lumens but after twenty minutes
it shines like the 40W incandescent I have in the same room.  It starts pretty bright though.

@LabSpokane
Hi, indeed heat might be the cause.  I would have to test it again but this time measure the
heat curve as well.  I will also try with a larger heatsink.

Do you know what is the mechanism that causes that effect of the current drop due to heat?
Is the cause in the semiconducting LED or could it be some kind of thermal stress on the internal
driving circuit?

Thanks.

[hgg]

You can easily unscrew the plastic base.
A small circuit is covered in kapton tape and around that a piece of double sided foam tape.

Some photos:




Which part of this circuit might get hot?  No part of the circuit is touching the heatsink.

[Rick Law]

The bulb went from 7W-4W.  So 6W could be some kind of average

@Rick
Please look at the first graph more carefully.

The current consumption starts at 7Watts, after only 6 minutes it drops to 4.5Watts, after 20 minutes
it drops to 3.9Watts but then it stays there indefinitely.  The stated 6.5Watts is not an average value.
The average value is 4Watts.  So at least in the case of current consumption the specs are clearly false...

Concerning the brightness I have no way of accurately measuring lumens but after twenty minutes
it shines like the 40W incandescent I have in the same room.  It starts pretty bright though.

...

Not that I have any relationship with the manufacturer and want to defend them, your data just doesn't support your conclusion of bad spec (yet) and that turns on the argumentive side of me.

Your first graph alone is meaningless, however, with the second graph together  show it works less well when it got hot...  It said nothing about how it compares to the 600lm spec.  Besides, watts drawn is not the important part, it is the 600 lumens output that is important, but you were measuring lux instead of lumens.

To measure the lumens with a lux meter, I would try to project the light to one square meter area - looking at the shape of your light, it would probably be a circular projection.  If so, your lighted circle should be 1.13 meter in diameter for one meter square area.  Assuming the light is rather even, your lux should approximately equal to your lumens.

Try it and see how close that measurement comes to 600lm (at the start before it heats up) - I too am curious.  If that is "not even close", I would say the spec is wrong.  I suspect the lumens quoted is when "best case" which from your data means when the light is cool.  If the spec is "good", it likely would be within 20% of 600lm given all the measuring error that could occur.

As to wattage consumed, I would not even care - It would be like going to the car dealer and say: it is suppose to burn 1 liter of gas per 10km at 100kph, and I went 10km at 100kph but it consumed only 2/3 of a liter of gas -- now fix my car so it consumes 1 liter instead of 2/3 liter; that makes no sense.  While output is correlates to power consumed, power consumed doesn't equal output.   I would focus on what it can output instead of how much power it consumes.

Rick

[hgg]

@Rick

This is like wanting to buy a car that can output 200hp but instead of that it outputs 120hp after
you drive it for a while.  Your dealer is trying to convince you that despite that, the car outputs
indeed 200hp at the start of the ride.  I agree.

I can see the point you are trying to make by saying that I did not measure lumens but by
inference it looks like that it outputs a lot less.   The beam pattern is so wide and with no
detectable edges that it is almost impossible to even approximate the lumen from the lux.

- The first graph shows clearly that its output drops dramatically when it gets hot.
- 600 lumen @ 4 watts & 5 euros would be quite an achievement!
- A typical 40watt incandescent is rated at approximately 400 lumen
- After 20 minutes both bulbs have the same brightness.
- Chinese products in general have a history of overrated specs.

So I think its safe to assume that the maximum output of this led bulb is around 400 lumen
and that this is another Chinese product with misleading specifications.

Maybe you get a 600 lumen bulb for a few minutes but then it gets disappointing.

@3roomlab
Maybe the component is a bridge rectifier because the bulb has no polarity.
100 pieces for a $1?...

And here is its temperature graph:



George.

[hgg]

AcHmed99 Hi,

Thank you for the info.

The current should be constant your graph shows almost a 50% drop over temperature, maybe some thermal foldback is tripping.You have no temperature axis on your graph the maximum case temperature is 85C if its hitting that at say 25c ambient then something is FU

I have.  The vertical axis represents both mA and degrees Celsius.
The temperature curve is the bottom one but I've created the purple one as well which is Temp x 10
in order to be more visible.  (I forgot to name the axis)

Do you say the controller is lowering the current on purpose because of over heating?
The circuit inside the bulb was free floating and no part was attached to the heatsink.
Overheating of the LEDs, some part of the circuit or both?

Thanks.

[hgg]

I think the bulbs were indeed rated for Canada.   
Initial current is 600mA and temperature 18.5 degrees C.

If it starts at 600mA and drops to 350mA then something is going on.

It does exactly that.  All the bulbs behave the same.
I am measuring current draw from my power supply which is accurate to 1mA.
The horizontal line of the graph represents the number of seconds passed.
In order for the bulb to start I have to set the current limit of the power supply pretty high like 1.5A

I have just tried again but with a small fan blowing at the bulb and after the 20 minute period
the current draw stabilizes at 400mA instead of the 323 without the fan.  More lumen as well.
So its definitely a temperature related problem but I not yet sure if its the controller of the
LEDs that are causing it.

I will try again with the bulb disassembled and see if I can measure the circuit components temperature.
Maybe try heat-sinking some of them as well.

[LaurenceW]

Hmmm, worth bearing in mind that the current "state of the art" is around 100 lumens (not lux! - different) per watt, and that's only for top-flight branded stuff. I VERY much doubt a no-name lamp made of many small individual chips and selling for six bucks would get ANTWHERE close.

Love to be proved wrong, though.

LED chip performance alone does not account for that current drop. Sounds like some electronics is throttling the current back, as heat rises. Which only means that overal light output will be even MORE reduced, overall.

But it's DealExtreme, so EVERY word on their web site must be true, right? Like I'm rideing this bike, it is.

[ConKbot]

Hmmm, worth bearing in mind that the current "state of the art" is around 100 lumens (not lux! - different) per watt, and that's only for top-flight branded stuff. I VERY much doubt a no-name lamp made of many small individual chips and selling for six bucks would get ANTWHERE close.

Love to be proved wrong, though.

LED chip performance alone does not account for that current drop. Sounds like some electronics is throttling the current back, as heat rises. Which only means that overal light output will be even MORE reduced, overall.

But it's DealExtreme, so EVERY word on their web site must be true, right? Like I'm rideing this bike, it is.

100 Lumens/watt may be cutting edge for efficiency from mains to light coming out of optics (if there are any... )  However, cutting edge LEDs can get > 200 Lm/W, and even cheap chinese LEDs can get 100Lm/W, so I'd suspect they are quoting LED efficiency, not the whole system efficiency (driver, LED, and optics efficiency)

Or just putting whatever looks good down on the paper and calling it done.

[Rick Law]

... so I'd suspect they are quoting LED efficiency, not the whole system efficiency (driver, LED, and optics efficiency)

Or just putting whatever looks good down on the paper and calling it done.

I was thinking along that line too (that the lumens spec is from the LED) and I looked up their site.  It uses 48x2835smd, and it lists "theoretical" 750-900lm, "actual" 500-600lm.

With two sets numbers, I think the 500lm-600lm probably is a measured number or guess work.  500lm settled, 600lm max.  Whereas, the "theoretical" 750-900lm is probably the LED's lumens from the chip's spec (18.75lm-15.625 per  LED).

Too bad the OP cannot measure the actual lumens.   The lux at fix distance is not meaningful to compare to lumens.  But since there is such a significant current drop as it warms up, there may be other more important issues then spec-accuracy...

Rick
Rick

[hgg]

Ok, I think I found the problem.
Its not that the LEDs are overheating, its the circuit components...

I had the light running for 20 minutes with the little board outside the heatsink.
Have a look at the temperature of each of the components.

Bridge Rectifier : 109C !!!
CC Controller : 90C
Schottky Diode : 95C
Inductor : 69C
Finally the heatsink with the LEDs was at 53C instead of the 58C when the circuit was encased inside.

Imagine the temperatures with the board enclosed inside the bulb...

During that period the current drawn was 515mA instead of the 323mA.  This translates to 6.18 Watts
which is what was supposed to be.  So the specs are theoretical but not actual.  That's what
they are always doing... 

"The LED's can deliver 600 lumen but we cannot do it"  should have been the actual DX listing. 

Another interesting thing are the following measurements when I played with the voltage:
  13V / 526mA
  12V / 584mA
  11V / 664mA
  10V / 770mA
  09V / 935mA
So the controller increases current draw when the voltage is decreased. 
(To keep lumen constant maybe?)

When I used a very small fan blowing only on the circuit board the temperature of all parts dropped
down to a very summery 35 degrees C and current draw was 584mA or exactly 7 Watts. 
Now we are talking!

If I can only find now a way of dissipating the circuit component heat to the heatsink.

Is there any type of putty or non conducting semi liquid that I can pour in the heatsink and then
insert the circuit permanently inside?  Would that be a good idea?  Any other suggestions/ideas?

Thanks.

[bktemp]

The efficiency of the LED driver circuit must be really bad if it gets so hot. Could you try to measure the voltage and current going into the LEDs and the input power at the same time?
Since the efficiency decreases at low voltage, the circuit is probably designed for higher input voltages (12Vac -> 16Vdc).

[hgg]

Voltage going to the LEDs is a constant 17.8V even if I change the input voltage.

This is the Schottky diode : http://diodes.com/datasheets/products_inactive_data/ds13006_R8.pdf and
this is the bridge rectifier : http://www.topdiode.com/pdf/MB2F-MB10F.pdf  (MB6F)
which can operate up to 150C but with a very large drop in current.


MB6F

Could this be the only component that has to be properly heatsinked ??

[bktemp]

Voltage going to the LEDs is a constant 17.8V even if I change the input voltage.

A boost converter explains why the input power increases at lower input voltages and then decreases at a certain point:
It trays to maintain the output current (probably 0.3A), but it has to use a higher current at lower voltages. This reduces the efficiency. At one point it reaches the current limit of the IC and the output voltage and therefore input power decreases.

this is the bridge rectifier : http://www.topdiode.com/pdf/MB2F-MB10F.pdf  (MB6F)
which can operate up to 150C but with a very large drop in current.

It is rated for 0.5A @ 30°C. But inside the lamp it will get much warmer. Therefore the design is a fail. A proper design uses schottky diodes for the bridge rectifier because if the lower voltage drop.

[hgg]

@bktemp

It is rated for 0.5A @ 30°C. But inside the lamp it will get much warmer. Therefore the design is a fail. A proper design uses schottky diodes for the bridge rectifier because if the lower voltage drop.

The design is indeed a complete fail.
What amazes me is how is it possible that somebody with even a basic knowledge of electronics would
design such a thing?  This bulb is designed to fail.

I don't want to leave it like that.  Maybe I can replace the rectifier with an appropriately sized one
and find a way to heatsink it as well. 

@3roomlab

i think you could try and see if you dip the whole thing in epoxy (or special thermal epoxy)

That's what I am trying to find out know.  Maybe the cheapest solution is to fill it with silicon but I am
looking for one with high thermal conductivity.

for experiment purpose, can you dip the whole fixture in oil?

I've read that somewhere, it will definitely work but the whole thing has to be tightly sealed.
Some suggest a mixture of toothpaste and petroleum jelly.   

- One option is to replace the rectifier with a much higher rated one
- and the other to fill the fixture with a thermal putty of some sort.

What do you think that would be the best solution?
Thank you.

[hgg]

Those are case temperature measurements in open air at about 20C?

Yep.
I bought a few of them and they all behave at exactly the same way.  I will use them inside the
house, one in each room and  they will be solar powered.  Maybe its a good idea to replace the
drivers with proper ones but if I can find an easy solution like proper heatsinking or just replacing
the bridge rectifier with a higher rated one, I prefer that solution.

What about if I make a small bridge rectifier with through hole Schottky diodes on a small board
and then connect it with small wires on the smd pads? 

[bktemp]

If you connect the lamps to DC, you can omit the bridge rectifier completely. But it makes sense to use a single schottky diode as reverse polarity protection.

Try to apply the 12V after the rectifier. If the board then still gets very hot, I would replace it completely.

[hgg]

Very good idea!
Thanks.

[LaurenceW]

Well, conkbot, I guess BOTH you and I could be right. This from the always-reliable Wikipedia site:

Nichia Corporation has developed a white LED with luminous efficacy of 150 lm/W at a forward current of 20 mA.[49] Cree's XLamp XM-L LEDs, commercially available in 2011, produce 100 lm/W at their full power of 10 W, and up to 160 lm/W at around 2 W input power. In 2012, Cree announced a white LED giving 254 lm/W,[50] and 303 lm/W in March 2014 .[51] Practical general lighting needs high-power LEDs, of one watt or more. Typical operating currents for such devices begin at 350 mA.

Note that these efficiencies are for the LED chip only, held at low temperature in a lab. Lighting works at higher temperature and with drive circuit losses, so efficiencies are much lower. United States Department of Energy (DOE) testing of commercial LED lamps designed to replace incandescent lamps or CFLs showed that average efficacy was still about 46 lm/W in 2009 (tested performance ranged from 17 lm/W to 79 lm/W).[52]

Yes, higher lumens per watt is possible. But in practice? From China? For six bucks? Nah.

And if this light is getting HOT, that suggests a good few less that 6.5Watts are finding their way to the LED chips.

[hgg]

Ok, I removed the bridge rectifier and after 20 minutes I took some temperature measurements with the
circuit off the housing.  One set with no rectification and one with a protection diode on the + input.

                                 No Rectification   with Protection Diode
                                 ===========================
            Inductor :               47.5 C                   62.0 C
  Schottky Diode :               44.0 C                   60.0 C
Boost Controller :               53.0 C                   65.0 C
           Heat sink :               62.0 C                   60.0 C
Protection Diode :                  -                         75.0 C

Conclusion:  Diodes get hot...   You can see that the protection diode went up to 75 C in free air
and through its legs heated up all the other components as well.  But now, the current lowering
problem was solved.  It was working at 515mA which is 6.2 Watts.  Close to the theoretical.

I left the diode in place and I filled the housing with a transparent silicone sealant I had around.
Then I took some measurements again.  Here is a comparison with a before and after:


Comparison

(Numbers on the vertical axis of the graphs are the same for mA and LUX.)

I made the mistake of not checking the distance of the lux meter and it was more than a meter,
so values are a bit lower but still the final measurement is higher than the one before the mod.
It starts with 6.5 Watts and after 30 minutes it consumes 5.7 Watts.

The relative brightness before the mod dropped 40.5% after 20 minutes
and after the mod only 13.3%  Not bad !

The only negative thing now is that the temperature of the heatsink went up to 72.2 degrees C!
This is 3 degrees lower from the temperature of the protection diode.  I think that if I had not used
the protection diode, the temperature would have been even lower.  I will try it on a different bulb
without a protection diode and without silicone.

bktemp thank you for the idea.