Light source efficiency math?

A watt of emitted optical power, focussed to a reasonably small spot, should be feelable, if not painful. 1 watt is a lot of light.

1 microwatt is perfectly visible from a good LED, and 1 mW can be too bright some time. I've seen under 1 nanowatt, dark adapted.

I have an LCD clock that glows constantly from internal blue LEDs. I'm guessing the LEDs are run at a few microamps. Almost too bright.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin
Loading thread data ...

I read that a dark-adapted human eye can sometimes see a single photon - but you have to look really hard!

Reply to
bitrex

No it won't. A heat sink made of a magic material with infinite heat capacity would stop there being any temperature rise, but an infinite heat sink of copper would equilibrate to a situation where the thermal resistance of a large solid block of copper would be very close.

"Carry away" is colloquially conduction - though convection is also important for real LED units which is why they die so young in glass globes. Thermal radiation losses are negligible until T > ambient+30.

About 80% is wasted in resistive heating in the matrix and another chunk in heating up the phosphor that turns blue light into broad yellow. Some LED show darkening of the phosphor from light flux damage. An LED die form high brightness types today is close to the luminous flux at the surface of the sun's photosphere (though only at its specific peak wavelength).

Semiconductor LED's generally produce a peaked output that is centred on a given wavelenth +/ 50nm FWHM with very low tails away from the peak.

At the very low power levels concerned it is performing as a heat pump converting electrical energy into somewhat more photons than you would expect by stealing thermal power from the crystal lattice.

--
Regards, 
Martin Brown
Reply to
Martin Brown

You could simulate an "infinite heat sink" by a cold sink (like solid CO2, liquid N2 (77 K) or liquid helium at 4 K) and a Peltier element between the case and the cold sink. Adjust the Peltier element current so that the device case is at 25 C or that the junction is at 25 C:

In practice, those manufacturers that specify the LED characteristics at Tj=25 C simply start from a cold junction (Tj 25 C and power up the LED and measure the characteristics during the firth millisecond after power on :-).

More reputable manufacturers (such as CREE) specify the characteristics at Tj=85 C and about Imax/3. From these specs, it is much easier to calculate the steady state specs for a real device conditions (doesn't require multiple iterations in a spreadsheet).

Reply to
upsidedown

Maximum value theorem for parabolic PDEs: the maximum value of any steady-state solution of a parabolic PDE must occur either on the boundary of the domain, or as part of the initial conditions.

Corollary: there can be no net energy flux into the domain if the solution is uniformly a constant.

Reply to
bitrex

Oh that's interesting. at about what delta_T above ~300K (my office is nearer 295) does radiation equal convection (STP)... Well right the shape matters. As a physics type I'm required to pick a sphere as my standard shape. Convection seems to start at ~ 1C/K delta T. Well that was for a 1/4 watt TH resistor, which seem to conduct zero heat through the leads.

(expect for maybe those solid carbon composition types. Those leads are non-magnetic at least)

George H.

Reply to
George Herold

Radiation happens in both directions, from deice to environment and from the environment (e.h. walls) to the device. Thus, the net power transfer is proportional to Tdev ^ 4 - Tenv ^4. Temperatures in Kelvins.

Reply to
upsidedown

Close but no cigar.

The difference-of-fourth-powers thing applies only when the surfaces have e missivities that are temperature-independent, and this is rarely the case. (See "greenhouse effect".) The discrepancies are easily of the order of the effect under discussion, namely the wall-plug efficiency of a LED.

There are also important corrections for the transmittance of surfaces.

Cheers

Phil Hobbs

Reply to
pcdhobbs

Yep, that would work and perhaps fit in my palatial office. I had visions of disaster trying to find a place to put an infinite heat sink.

The problem with using a Peltier cooler on a commodity LED is that the thermal conductivity of the typical flashlight style of LED is insufficient to insure that the junction temperature will be sufficiently close to the temperature of the Peltier cold junction. Note that lumens/watt is specified for the junction temperature, not the case temperature.

It's not beneath their dignity to cheat a little so that they can produce larger luminous efficacy (lumens/watt) figures by specifying that it be measured at 25C. For example:

while operating at 1W and 25 degrees Celsius (77F)..." For testing, they don't try to attach a near-infinite heat sink or complexicated cooling contrivance to cool the LED. It's easier to use short current pulses to power the LED and measure the peak light output. At a very small duty cycle, the LED doesn't have time to rise much in temperature.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

It isn't a bad ballpark guide though provided that you don't have something awkward like brilliant polished metal or an IR window.

Indeed, but for your average random opaque plastic surface without a mirror metallic finish a deltaT exceeding 10% of ambient (in Kelvin) is about the point where radiation becomes non-negligible. The fourth power dependence of radiation hides a multitude of sins in emissivity.

I once got caught out by a bright metal aluminium case on a 78xx based PSU - I liked the mirror finish but the regulator didn't and promptly went into thermal shutdown after 10 minutes running at rated power. A coat of black paint fixed it.

--
Regards, 
Martin Brown
Reply to
Martin Brown

power

I once amazed my co-workers by reducing the temperature of an FPGA a few K just by painting the *inside* of the enclosure black. (Previously it was shiny Al.) The thermal path was:

FPGA -> PCB -> metal box -> ambient.

There was no fan, and a significant fraction of heat transfer between the PCB and the metal enclosure was due to radiation.

The outside of the box was already painted black, so this change actually removed a masking step in production.

Allan

Reply to
Allan Herriman

I have built some LED retrofits for 48" fluorescent fixtures with Cree LEDs, running at about a 1 W input level (350 mA @ 2.8 V or so). I can tell you the optical output is DEFINITELY feelable on the skin, about like full noon sunlight. You can tell an immediate change in skin temperature when switching the lights on and off. If you have your hand close enough, there is a delayed change due to the warmth of the PC board/heat sink. But, the radiated power is really quite strong.

Jon

Reply to
Jon Elson

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.