Re: Strange problem with low energy light bulb

[default]

Experience is the best teacher.

Yes. Had someone told you there would be pocket sized radio telephone
equipment in the 50's you might not have believed it.

Agreed LEDs aren't quite there yet, but they are almost there, and
that is pretty exciting stuff. (or at least it is compared to the
LEDs we were using in the early 70's)

Experience with electronics suggests things will get smaller lighter
cheaper and work better over time.

You'd condemn the whole industry, and research effort, because some
charlatan wants to make a quick buck on TV?

 

[Kevin McMurtrie]

Don't believe everything you read.

Experience is the best teacher.

In other words, LEDs suck for bright light sources.

However, the hype is good for selling them to fools over the
TV/Internet. Like that one guy selling LED light bulbs on TV (USA). He
doesn't explicitly say that they're bright as a lightbulb, but they
are in a lightbulb package and used in the commercial as if they were
a lightbulb. Some people probably fall for that.


Right, but pointless except for low light applications.

http://www.lumileds.com/products/line.cfm?lineId=19

145 Typical lumens from 700mA @ 3.6V, or 58 lumens per watt. The LED is
a surface mount chip 3.1 x 4.6 x 2.1 millimeters in size. A thin strip
of 12 of these chips puts out light comparable to a 100W tungsten lamp
but only uses 30W.

I'm using 8 of the older Luxeon K2 LEDs in a bicycle light. The output
is nothing short of impressive. It's brilliant at just 5W of input
power. Crank it up to 40W and it puts car headlights to shame.

 

[Eeyore]

http://www.lumileds.com/products/line.cfm?lineId=19

145 Typical lumens

No. NOT TYPICAL. That's the highest output (most expensive) grade.
LXML-PWC1-0080
http://www.lumileds.com/pdfs/DS56.pdf

from 700mA @ 3.6V, or 58 lumens per watt.

In 'cool white' i.e, the very blue light with a 6500K colour temp. Not very
suitable for domestic lighting.

More like 95 lumens for a warm white, which works out as 43 lumens/W
http://www.lumileds.com/pdfs/DS56.pdf

Graham

 

[Eeyore]

http://www.lumileds.com/products/line.cfm?lineId=19

145 Typical lumens from 700mA @ 3.6V, or 58 lumens per watt. The LED is
a surface mount chip 3.1 x 4.6 x 2.1 millimeters in size. A thin strip
of 12 of these chips puts out light comparable to a 100W tungsten lamp
but only uses 30W.

I'm using 8 of the older Luxeon K2 LEDs in a bicycle light. The output
is nothing short of impressive. It's brilliant at just 5W of input
power. Crank it up to 40W and it puts car headlights to shame.

 

[Dave Plowman (News)]

http://www.lumileds.com/products/line.cfm?lineId=19

145 Typical lumens from 700mA @ 3.6V, or 58 lumens per watt. The LED is
a surface mount chip 3.1 x 4.6 x 2.1 millimeters in size. A thin strip
of 12 of these chips puts out light comparable to a 100W tungsten lamp
but only uses 30W.

That's poorer efficiency than a CFL can manage. Chances are the colour
temperature is poor too - if tungsten is your norm.

 

[Eeyore]

http://www.lumileds.com/products/line.cfm?lineId=19

145 Typical lumens from 700mA @ 3.6V, or 58 lumens per watt. The LED is
a surface mount chip 3.1 x 4.6 x 2.1 millimeters in size. A thin strip
of 12 of these chips puts out light comparable to a 100W tungsten lamp
but only uses 30W.

The colour temperature won't be anywhere near comparable.

In any case don't 'white leds' use the same phosphor method of producing light
that CFLs do ?

Graham

 

[Eeyore]

With a bit of work, I'm sure that they [LEDs] will
also get to the point where they can replace a standard filament bulb, in
the same sized package, unlike a CFL which has to accommodate the ballast.

To run LEDs efficiently from 240V AC will also require some active electronics.
I see no-one ever factors in the power losses that'll be associated with that.

Graham

 

[Eeyore]

As for actual light output, newer LED's put out 176 lumens or more.

Cite please ?

Graham

 

[GregS]

With a bit of work, I'm sure that they [LEDs] will
also get to the point where they can replace a standard filament bulb, in
the same sized package, unlike a CFL which has to accommodate the ballast.

To run LEDs efficiently from 240V AC will also require some active electronics.
I see no-one ever factors in the power losses that'll be associated with that.

When you use up the watts, you get heat. The lamp has to efficiently get
rid of it. May even need a fan.

In my experimental 45 watt, 9- 5 watt Luxeon array, I use copper, aluminum, and a fan.
DC drive is nice.

greg

 

[Eeyore]

With a bit of work, I'm sure that they [LEDs] will
also get to the point where they can replace a standard filament bulb, in
the same sized package, unlike a CFL which has to accommodate the ballast.

To run LEDs efficiently from 240V AC will also require some active electronics.
I see no-one ever factors in the power losses that'll be associated with that.

When you use up the watts, you get heat. The lamp has to efficiently get
rid of it. May even need a fan.

Quite possibly so. How many watts does the fan need ?

In my experimental 45 watt, 9- 5 watt Luxeon array, I use copper, aluminum, and a fan.
DC drive is nice.

The normal home doesn't have DC. What do you use for current limiting and how much power does that
dissipate ?

Graham

 

[GregS]

Arfa Daily wrote:

With a bit of work, I'm sure that they [LEDs] will
also get to the point where they can replace a standard filament bulb, in
the same sized package, unlike a CFL which has to accommodate the ballast.

To run LEDs efficiently from 240V AC will also require some active electronics.
I see no-one ever factors in the power losses that'll be associated with

that.

When you use up the watts, you get heat. The lamp has to efficiently get
rid of it. May even need a fan.

Quite possibly so. How many watts does the fan need ?

Just uses a small CPU fan. I'm actually using a CPU sink/fan combo, but the LED's are epoxied to a copper
plate. I even have diamond dust as a buffer/insulator. To close space the LED's
I needed the best thermal transfer. Did I forget to mention the Peltier device.
I am also using unmounted LED's.

The normal home doesn't have DC. What do you use for current limiting and how
much power does that
dissipate ?

Nothing is compact. I use a large variable supply. There is 45 watts going to the LED's.
About 20 [email protected] amps. 3 series sets of LED's. You cannot look at the light.
The device was intended to be flashed in the final form. peaking at about 60 watts.

greg

 

[Lostgallifreyan]

Cite please ?

Graham

Table: Flux Characteristics (Tj = 25°C)
Cree XR-E type P4, 80 lumens at 350 mA
Cree XR-E type Q4, 100 lumens at 350 mA
Graph: "Relative Intensity vs. Current Tj = 25°C)"
Line rising through 100% output at 350 mA to >220% at Imax of 1A.
Source: http://www.cree.com/products/pdf/XLamp7090XR-E.pdf

Logical inference is that the P4 can output >176 at 1 amp, and the Q4 can
do >220 lumens at 1 amp.

NIST tested an XR-E Xlamp at 350 mA and found outputs exceeding that
claimed in the data sheet. They did not measure the output at 1A but they
did certify the lumens per watt.
Source:
http://www.cree.com/products/pdf/NIST XLamp LED Document.pdf

They did not test the output at 1A, or at least did not certify it, but
either the LED's die before you manage to push 1A through them, or they can
do what is claimed. Even if there is a fall-off with current, the Q4 type
will definitely exceed 176 lumens at 1A.

 

[Lostgallifreyan]

And there is newer XR-E Q5 Xlamp too. Not sure what its spec is though.

 

[Lostgallifreyan]

To run LEDs efficiently from 240V AC will also require some active
electronics. I see no-one ever factors in the power losses that'll be
associated with that.

95% efficient power conversion from 5 to 32V and I think I saw similar
claims for a small module that can power a series chain of LED's from a
mains input. Can't cite a source right now, I just got you one already for
the 176 lumens claim. If I can find the other I'll cite it.

 

[Lostgallifreyan]

95% efficient power conversion from 5 to 32V and I think I saw similar
claims for a small module that can power a series chain of LED's from
a mains input. Can't cite a source right now, I just got you one
already for the 176 lumens claim. If I can find the other I'll cite
it.

Here's one at 82%: http://www.magtechind.com/__LED_Lighting_PSU/Anz107.pdf
I saw better, but this was one I found while looking for a specific item.
If I find something significantly better on 240 VAC I'll post again.

 

[Lostgallifreyan]

[email protected] (GregS) wrote in

Just uses a small CPU fan. I'm actually using a CPU sink/fan combo,
but the LED's are epoxied to a copper plate. I even have diamond dust
as a buffer/insulator. To close space the LED's I needed the best
thermal transfer. Did I forget to mention the Peltier device.

I can see why diamond dust might help, but why a TEC? The LED's can operate
at temperatures great enough to pour heat through a decent thermal coupling
to a heatsink, no need to pump it through by force, that just makes work
for work's sake unless you are forced to drop the temperature in a confined
space, which is doubtful given the size of heatsink you need to dissipate
LED heat + TEC work heat.

 

[Lostgallifreyan]

In any case don't 'white leds' use the same phosphor method of
producing light that CFLs do ?

Higher intensity. Might be nonlinear, as in greater efficiency if you pump
them harder. Might be different materials. I don't know for sure though.

Definitely smaller size, so if you're going to be a stickler for full
context such as analysis of lumens per watt of actual mains input, you must
take all of the context. People have alreay said (rightly) that LED lamps
won't have the trouble that CFL's have in fitting most current luminaries.
That's obviously important regarding watse and expense.

More: LED's are growing more efficient all the time. It might be that in
future these lamps might be directly driven by encapsulated laser diodes
emitting near UV to pump phosphors. Laser diodes have efficiencies beyond
low pressure sodium, they leave it in the dust. It's likely that this
technology will quickly make CFL's look barbarous.

 

[GregS]

[email protected] (GregS) wrote in


I can see why diamond dust might help, but why a TEC? The LED's can operate
at temperatures great enough to pour heat through a decent thermal coupling
to a heatsink, no need to pump it through by force, that just makes work
for work's sake unless you are forced to drop the temperature in a confined
space, which is doubtful given the size of heatsink you need to dissipate
LED heat + TEC work heat.

The LED's maximum rating essentially calls for room temperature. Even with room temperature,
the LED die will be much hotter. As the LED heats up, the light output sags.
Normally, the Luxeons are allready mounted on a piece of aluminum, which looks like
a star, or Luxeon Star. With my mounting arrangement, the Luxeon Emitters are back to back,
thus need faster heat transfer. The actual size of the heat transfer piece off the LED die is very
small, and is electrically insulated with epoxy, so the heat transfer could be improved.

greg

 

[Eeyore]

Eeyore wrote


95% efficient power conversion from 5 to 32V

But that's not mains. Nor do LEDs run off a voltage source. You need to control
the current.

and I think I saw similar claims for a small module that can power a series
chain of LED's from a
mains input.

I very much doubt it. One very important point however is that these need to be
driven with a constant *current* not a constant voltage and that's typically a
lossier method too.

Graham

 

[Lostgallifreyan]

[email protected] (GregS) wrote in

The LED's maximum rating essentially calls for room temperature. Even
with room temperature, the LED die will be much hotter. As the LED
heats up, the light output sags. Normally, the Luxeons are allready
mounted on a piece of aluminum, which looks like a star, or Luxeon
Star. With my mounting arrangement, the Luxeon Emitters are back to
back, thus need faster heat transfer. The actual size of the heat
transfer piece off the LED die is very small, and is electrically
insulated with epoxy, so the heat transfer could be improved.

greg

Good point re output drop with heat. It's not so bad though, linear drop to
90% at 70°C. I guess if you have enough of them densely packed it becomes a
specialised lamp, but I imagine a 486 heatsink with no fan should take care
of up to four XR-E's and still operate well below 70°C. K2's are less
efficient. Were anyway, I think newer ones that match Cree's efficiencies
are being made, according to some posts I read on Photonlexicon.

Even at >120°C the chips put out 70% of 25°C rated output, and I doubt even
ten closely packed emitters will get that hot if convection can keep the
mounts below 85°C. Ten emitters in a lightbulb's space couldn't even get
that hot unless they had less than twice the efficiency of 100W tungsten
lamps, and they're a lot more efficient than that, >80 l/W with a claimed
industry best thermal resistance of 8°/W.

Any tips for mounting? I'll look deeper at the Cree site for guidance,
but for now I'm contemplating an anodised heatsink (such as used for 486
CPU) and using a thin smear of epoxy to fix three of them in place along
with a small variable-output driver.. That's crude, so I'm keen to hear
some better ideas for improvised assemblies.