# Compare white LED radiated optical power: lumens vs milliwatts

I'm trying to compare the brightness of various white LEDs, some of which are listed in lumens and others in milliWatts (this is milliwatts of optical output, not watts of electrical power consumption). Is there an easy way to do this that doesn't involve integrating the spectrums?

For example, Mouser 997-L1502270500600S0 is a 2200K white LED with 690mW of radiant flux. Mouser 997-L15027705060 is a 2700K white LED with 600lm of luminous flux. Which is brighter?

The lumens to watts conversion always assumes one wavelength, which white LEDs don't have, so I assume that calculation can't apply here.

Some background: 1) Lumens to mW conversion for a single wavelength 2) A few other SE questions on Lumens and Watts.

• LEDs, generally, don't produce "white." (Okay. Unless they are heated to "white hot" in an oven, after which they generally become useless when cooled.) However, they do produce (broadly speaking) a range of wavelengths. A 365 nm LED might produce from about 355 nm to 380 nm with a peak at 365 nm. This may be allowed to charge up phosphors (not just one) which will "Stokes-shift" the light down into visible wavelengths to produce "whitishness" of some kind as seen by a human. Without knowing the specific phosphors, efficiencies, etc., it would be "difficult" to make such comparisons.
– jonk
Nov 1, 2019 at 3:42
• The color temperature might be a clue about the phosphors being used. And you might use that information to seek out and look for phosphors associated with those color temperatures, along with the more efficient means used to drive those phosphors, to develop an a priori approach that may work "well enough for horseshoes," so to speak. But the only thing I'd trust is an Ocean Optics (cheap) spectrophotometer with various gratings and a well-designed setup, plus software used to compute CIE x-bar and y-bar from x, y, and z coords. (The setup will account for standardizing the steradians, too.)
– jonk
Nov 1, 2019 at 3:45
• @jonk The white LEDs I reference above clearly use down-converting phosphors and the spectral output is given, in relative percentage intensity only, in the datasheets. I definitely agree that buying the LEDs and testing them would be one way to determine which one is brighter. But, I'd like to pick which one to buy based on their specifications. So, given a 2200K white LEDs that gives 690mW and a 2700K white LED that gives 600lm (and all of the details in each of their data sheets), which one is brighter? Nov 1, 2019 at 4:09
• (I decided to delete some of my last comments as unhelpful. I actually don't know what you are trying to achieve and so I don't have much to add.) Why don't you buy some and make some decisions based upon what you can decide through test and observation?
– jonk
Nov 1, 2019 at 5:21

Both parts are listed in the same datasheet

The amount of flux is easy to compare. The color will be a little different, but the 227 (2200k color) part will be brighter.

The main problem here is that Mouser have included a rubbish specification from an unknown source - nobody quotes mW flux for white LEDs. mW are quoted for very short wavelength ble end of spectrum LEDs that are used for applications where their short wavelength results in super-low lumen figures but gives high energy per quanta - eg applications such as dental adhesive hardening.

In this case the REAL specs for the Lumileds product can be found here - link is on Mouser page.

However, if you HAD to compare lumen with mW.

A good modern LED outputs maybe half it's total input energy as light and will have a high lumen/Watt rating - maybe 150-200 l/W.
So a 690 mW flux out LED may have around 1.4W in and at 150 l/w gives about 200 lumen.
That's inexact (and then some) but the 600 lumen LED has enough greater output that it is almost certain to have more flux ouput.

BUT you asked about "brighter" so you want lux = lumen/m^2 so you care about both cone angle (radiation angle at half brightness" AND wavelength due to lumen variation due to eye sensitivity response curve. AND you MAY care about CRI (colour rendering index) as while that's not directly related to brightness per se, the differences in perception with CRI will affect how people feel about brightness, or may.
And .... . It's complex :-).

Generally data sheets do give lumen and lux and radiation curves - so you generally do not have to get too too arcane.

• Thanks for the link to the datasheet and also the "rules of thumb" calculation. I should have been more clear about "brightness" I'm really interested in radiant power. I guess that the "real/arcane" way to do this would be to actually do the integration against the luminosity function to convert lumens (and the relative spectral output curve) into radiant power in mW... but I'm not going to do that as it would involve lots of grabbing graphed data from the datasheets and importing it into analysis software. Nov 4, 2019 at 4:31
• ... unless you know of a better/ easier way? Nov 4, 2019 at 4:40
• I've looked at similar to what you want with other products by looking for one of their blue LEDs that had a good characteristic match with a white LED indicating that it was at least similar to the white LED's "engine" and then playing some games. || How much effort is required depends on what you are "really" trying to do and how much you want to spend and/or how much effort you are willing to make. Along the way I bought a lovely (clunky looking) Motorola SLR spot luminance meter. $US5000 ish new but it cost me a few hundred. You could ...create something functionally luminance Nov 4, 2019 at 9:53 • make something functionally nearly as good for a limited set of conditions with some lenses and a cheap lux meter. Add a lightly diffuse not very large sphere to the LED and take a uitable number of measurements. || As brightness and radiant power correlate wellish you can probably do minimal test work by taking the radiation pattern, measuring the brightness in a selected area around the major axis, seeing what that is as a total of the radiation pattern area and then comparing that with total radiant flux. Or something :-). I'm sort of 'thinking as I go' here. Cllar someone ... Nov 4, 2019 at 9:58 • who understand what you are trying to do, get a white board and a supply of Coffee (or Pepsi or .... - non alcoholic probably wise) and explain what you are trying to do with diagrams and hand wavium as required. As long as they UNDERSTAND what you are explaining it opens up (for me at least) the creative processes and allows ideas and thoughts to trickle out. | For me this works really well. Nov 4, 2019 at 10:00 The cone angle 2θ =65.65° where the cone 2x half-angle projects an area onto the sphere defined by $$\A=r^2\$$. Thus scaling the total beam angle over the std. cone angle , you can estimate the total lumens. This is a good estimate for this type with the angle given always at half peak power. Lumens = luminous Flux/steradian * beamwidth / 65.65 deg/steradian The specs are; • L150-22705006000S0: 800mA * 6.1V = Pd= 4.88W ,BW = 116° Lum. Flux= 690 mW • 690 mW * 116° / 65.65° = 1219 lumen$1.96(1)
• L150-2770500600000: 640mA * 6.1V = Pd= 3.90W , ,BW = 116° 600 lumen \$1.59 (1)

But you must realize the 1st one runs hotter so comparing at the same current, 640/800 * 1219 lumen = 780 lumen is only 30% brighter and you need a really good heat sink to meet these specs.

• Stewart: Is this calculation correct? It looks like you are converting between luminous flux (measured in lumens) and luminous intensity (measured in candelas). The power number of 690mW from the data sheet is radiant flux, which relates to the luminous flux through the Luminosity Function and not through the solid angle... Granted, I'm the one who's confused and asking questions here, so please set me straight. Nov 4, 2019 at 4:39
• I don’t do this every day, so I made some errors but Luminous Flux is used by Plant Scientists who don’t use the CIE curve for Lumens so it must be measured like a Spectrum Analyzer with a narrow BW and swept across and numerically accumulated. Plants do not see the same as humans and in many cases just the opposite, sensitive to IR, deep Red , Blue and UV. Humans are most sensitive to true green then normalized to a flat response in the brain. Nov 5, 2019 at 13:48