Updated 5/30/2016.

Most basic general questions are handled in Craig Johnson's LED FAQ, located at

Info on using and determining dropping resistors is in my LEDs 101 File. That file also explains why it is usually not good to connect an LED directly to a fixed voltage source.

Now for the other frequently asked questions:

Where do I get blue or white LEDs with a voltage drop less than 3 volts?
How do LED flashlights work with just one battery?
How do LED keychain flashlights get away with no dropping resistor?
Where do I get infrared LEDs with oddball wavelengths such as in the 700's of nm?
Where/how do I get LED lights for motor vehicles?
Where do I buy XXXX?
How do I power LEDs from household line voltage?
Converting millicandelas to lumens!
How long is the phosphor persistence in white LEDs?
LEDs for growing plants?
660 nm or 670 nm red LEDs for skin phototherapy / healing
   Skip to the list of these 660 nm LEDs
How do I get high brightness/efficiency green that is less bluish or more yellowish than 520-530 nm?
I was sent here to look for goodies not mentioned above!

Where do I get blue or white LEDs with a voltage drop less than 3 volts?

UPDATE 5/30/2016 - Blue and white LEDs, mainly high power ones and mostly white ones, are increasingly available with voltage drops of 2.75-2.9 volts, even at currents as high as 350-700 mA. Many of these low voltage drops are specified at junction temperature of 85 C, but are only about .1 volt higher when the junction temperature is 25 C. Many of these modern LEDs have voltage drop of around 2.75 volts at 25 C when the current is 100-150 mA. Light output of these better modern LEDs at 100-150 mA is generally 40-75 lumens, which is greater than the light output of a KPR-2 / K2 krypton-enhanced flashlight lamp / bulb - and the KPR-2 uses 700 mA.

An LED flashlight bulb with a boost converter built in to utilize 3 volts is described in Craig Johnson's Starlite Flashlight and Night Pearl Flashlight Bulb Page.
UPDATE 9/23/2001 - Craig Johnson added a Night Pearl Bulb Page. There are 1.5V and 3V versions.

UPDATE 8/10/2003 - A higher power LED flashlight bulb that supposedly accepts an amazing range of voltages is now available. Craig Johnson has a web page on this one here.

UPDATE 3/13/2004 - zinc selenide white LEDs have a typical voltage drop of 2.65 volts at 20 mA. They can be made in aqua-blue. White ones have chips that emit aqua-blue light out their tops and orangish light out their sides, and the light has to be mixed inside the LED. Roithner Laser is selling a few zinc selenide models. Craig Johnson has an online review of these in one of his pages on LEDs available from Roithner Laser. Beware that ZnSe LEDs have shorter life expectancy than most other LEDs.

UPDATE 12/28/2011 - Lumex has blue and white LEDs with boost converters so as to work at 1.5 volts. Their Digi-Key with catalog numbers are 67-1876 and 67-1877 respectively. They are not very bright - they appear to be intended to have low power consumption, close to 7 mA at 1.5V in my experience, which leads me to think that the LED chip receives about or a little over 2 mA.

As of 7/31/2011, Digi-Key was not stocking these. They were in stock in 2006.

How do LED flashlights work with just one battery?

They use boost converters. There are many ways to do this.

LED flashlight "bulbs" with boost converters built in are available - see above.

One old traditional way of powering LEDs from a single 1.5 volt cell was to use National Semiconductor's LM3909. UPDATE 3/24/2008 - The LM3909 is discontinued, but there is now a published LM3909 emulator circuit at Red Circuits.

There is a circuit known as the Joule Thief for powering a white LED from a single cell of voltage 1.5 volts or even much less.

Boost converter ICs for powering white LEDs from 1.5 volts and similar low voltages are now increasingly available.

UPDATE 7/31/2011 - Two simpler, easier-to-use boost-type LED driver ICs are Diodes-Zetex ZXSC380 and ZXLD381.

Where do I get infrared LEDs with oddball wavelengths in the 700's of nm?

They do exist and someone has to sell them. One supplier is Roithner Lasertechnik,
Roithner Laser is a supplier of oddball, especially higher power LEDs as well as some lasers.

Craig Johnsom reviews some of these at

Another supplier of oddball wavelength IR LEDs is Epigap, They have the odd wavelengths 700, 740, 770, 810, 840, and 870 nm and some less-common longer wavelengths.

Another is Epitex,
Another link: Epitex Home Page.
They have 700, 735, 750, 760, and 810 nm and a few longer wavelengths.

Other suppliers will be added here when I find out about them.

How do LED keychain flashlights get away with no dropping resistor?

The main reason is that the "coin" cells these flashlights use have enough internal resistance to limit the current to a safe or nearly-safe value.
And most of the time the battery's internal resistance usually rises after a few seconds during use, then recovers after giving the battery a break - this will usually keep the LED from overheating. Also the battery can heatsink the LED, permitting operation with current a little above the LED's rated limit.

I have seen some units where the current reached levels that I would call adventurous - especially in Photon models with blue, blue-green or white LEDs. However, you need really heavy use with constantly fresh batteries to damage the LEDs - and any significant LED damage if such currents are sustained will probably take hundreds or thousands of operating hours. I consider it a safe bet that few users of these lights will log 500 hours of use with highly fresh batteries in a lifetime. Normally, the battery can only provide current in excess of the LED's maximum rating for a few minutes. Also, it is not easy to notice if the LED deteriorates to even half its original performance. But if you do notice any fading, chances are something like 99 percent that it will be due to the condition of the battery rather than the condition of the LED.

Where/how do I get LED lights for motor vehicles?

Truck and bus lights are available from Dialight's Transportation Products. More likely, from a Dialight distributor.

Lights for cars, especially other than center high-mount brake lights, are more customized and specialized. Availability is mainly as a replacement part for a specific car, all too often at a very high price through a dealer for that make of car. If the light is not at an edge or a corner of the vehicle, you may get away with a truck light but I do not guarantee this and I discourage replacing any light unit on your vehicle with something else unless it is approved by DOT (in the USA that is, or whatever authority has jurisdiction in your country) and the manufacturer intended it to be used in your vehicle.
Replacing an automotive incandescent bulb with an LED "bulb" of the same base style and of the appropriate color will appear to work, but light output will often fall short of requirements. Expect hype in claims of light output from LED "bulbs" that fit where incandescent bulbs normally go. In addition, the light distribution pattern will be different and even in the highly unlikely event you have adequate total light output the amount of light output into some, probably even most directions will almost certainly not be in the allowable range.
Homebrewing a vehicle light requires knowing the lower limit and upper limit for amount of light into something like 40 different directions so I do not recomend this.

Where do I buy XXXX?

Try these links:

My LED Top Page, links to manufacturers and suppliers.

Craig Johnson's Where To Buy Page,

My Bright/Efficient LED Page, has a bit of supplier info.

Craig Johnson's flashlight review page,, mentions more LED flashlights than perhaps even he could remember! Plus a few items other than just flashlights.

Craig Johnson's "What's New" page, linking to pages of his added or updated roughly within the past month.

How do I power LEDs from household line voltage?

For most hobby, experimental and science project purposes, the best way is to use a "wall wart" transformer with DC output. Use appropriate resistors. For more details on determing what resistors to use, go to My LED Resistor Page.

Persons who are properly qualified and experienced at building line voltage powered electronic devices can:

a) Hack an LED night light.

b) Build a circuit to power an LED from line voltage, which would typically for low power LEDs (up to 30 mA) consist of:

* A brigde rectifier, with the LED connected to the DC output leads. The voltage rating of the bridge rectifier is not important.

* A capacitor in series with one of the AC leads of the bridge rectifier to limit the current. Use .47 microfarads to get approximately 19 milliamps of LED current (assuming 120 volts 60 Hz). The capacitor must have an actual AC voltage rating well above the line voltage. A DC rating well above 1.414 times the line voltage is not sufficient - an actual AC rating is required.

* A resistor in series with either AC lead of the bridge rectifier to limit peak current. If you have a capacitor across the LED (recommended around 220 microfarads), then this resistor only needs to be perhaps 33 to 100 ohms (for 120 volts AC). Otherwise this resistor needs to be at least 1.5 Kohms unless the LED is known to reliably repeatedly withstand current peaks well over 100 mA. This resistor should be a composition or wirewound type rather than a film type or otherwise known to reliably handle the current inrush through the current-limiting capacitor when power is applied.

* A fuse in case things go wrong.

* A resistor across either the current-limiting capacitor or the line terminals of this circuit, so that the line terminals do not present a shock hazard from charge of the current-limiting capacitor after this circuit is disconnected. The resistor needs to discharge the current-limiting capacitor from the peak line voltage (typically 169 volts) to no more than 28 volts (almost 2 time constants) within a fraction of a second. A 220K resistor discharges a .47 microfarad capacitor from 169 volts to 28 volts in about .19 second (max. of .23 second with 10% tolerances of resistor and capacitor) and will dissipate approx. .065 watt at 120 volts.

Converting millicandelas to lumens!

I have this Here towards the bottom of my bright/efficient LED page.

How long is the phosphor persistence in white LEDs?

A fan of this website e-mailed me that he tried an experiment to determine the risetime and falltime of a white LED, and found these to be less than a microsecond.
I do not guarantee that all white LEDs turn on and off this fast - I consider it possible that some may take a few microseconds. The phosphor may be slower at reduced power - I have seen some phosphors (other than LED phosphors) have faster response at higher illumination due to stimulated emission and quenching by photons. Sorry, this is all the info I have for now on this!

LEDs for growing plants?

UPDATED 4/30/2012

Go here to jump to red/blue ratios and specific LEDs.

As of April 2012, the most efficient LEDs suitable for growing plants are about twice as efficient at producing the desired light as fluorescent lamps of types made for this. LEDs cost so much more than fluorescent lamps of the same output that fluorescent lamps are usually more economical, at least on large scale.

Now, for wavelengths and how much light is needed:

Chlorophyll has two spectral ranges of utilization - red and blue.

Plants also have two kinds of chlorophyll - A and B.

Basic photosynthesis will proceed if only one chlorophyll is utilizing light. Chlorophyll A is the basic one for making chemical energy from light energy. Chlorophyll B is the main "accessory pigment" in green algae and "higher plants", and serves to utilize light at wavelengths not absorbed well by Chlorophyll A and transfer the energy from such wavelengths to the process that uses Chlorophyll A. Other accessory pigments are mainly carotenoids, which mainly utilize blue and violet light.

Although stimulating only one photopigment is sufficient for photosynthesis, many plants have some requirement of stimulating more than one photopigment for proper growth regulation, flowering and fruiting.

Most published spectral curves do not show well actual utilization, and a few show well ratio of absorption to transmission, or ratio of light utilized to light not utilized. If these curves are redrawn to show ratio of light utilized (or absorbed) to incident light, the peaks become wider and flattened. Many of these curves are also inaccurate, showing the peaks as more symmetric than they actually are. The peaks are asymmetric, with wavelengths shorter than peak being used better than wavelengths longer than peak.

The red region has a utilization peak around 660-670 nm for Chlorophyll A and around 635-645 nm for Chlorophyll B, depending on the source of information. Plants generally make good use of all red wavelengths except for ones much longer than 670 nm. 700 nm is close to useless for plants. Most plants actually make good use of orange and even yellow-orange light, to such an extent that high pressure sodium lamps have been used for growing plants.

The blue utilization peaks of chlorophyll are mostly reported to be around 430-440 nm for Chlorophyll A and around 453-470 nm for Chlorophyll B. Unlike the red peaks, the blue peaks are usually shown with substantial asymmetry indicating that wavelengths shorter than peak are used well while wavelengths longer than peak are not used well. Beta carotene is a major blue-absorbing accessory pigment, with a double peak at about 450 and 480 nm.

The blue response bands of both chlorophyll A and chlorophyll B and also beta carotene are served well by Lumileds royal blue LEDs, which typically have peak wavelength near 450 nm.

Photosynthesis works most fundamentally from red light and secondarily from blue light. However, many plants have some need for blue light for proper growth regulation and/or flowering and/or fruiting.

As for how much light is needed:

This page at Wisconsin Center for Space Automation and Robotics describes a plant growth unit illuminated by 670 nm red and 470 nm blue LEDs. Illumination intensity is adjustable, with maximum values (my translation to watts of light per square meter) of about 98 watts per square meter for red and about 18 watts per square meter for blue. If you need to get by on much less than this, I would make these figures more equal in case your plants have specific blue light requirements that need to be satisfied. I suspect plants that do not have especially high needs can fare reasonably well with 25 watts per square meter of red and 9 watts per square meter of blue.

UPDATE 11/28/2009: This 11/25/2009 "LED Professional Magazine" item mentions some plant-growing LED lamps with mixtures of red and blue LEDs, in 3 different wattages. The two higher wattages are available with red/blue ratio 8/1, and the lowest wattage is available in red/blue ratios of 8/1 and 2/1.

Update 4/30/2012 The top-rank red Luxeon "Rebel", LXM2-PD01-0050, is typically about 43-46% efficient when the junction temperature is 25 degrees C, and typically about 38-41% efficient in the more reasonable situation of heatsink temperature of 35 degrees C. This is at 350 mA.

Update 3/22/2011 The most efficient royal blue Lumileds Luxeons currently available are LXML-PR01-0500 (of the Rebel series), which are typically 45-48% efficient at 350 mA with a 25 degree C junction temperature. Thankfully, blue LEDs have a much lower temperature sensitivity than red ones have.

Update 7/23/2011 Osram LD W5AM-3T3U-35-Z royal blue is typically 40% efficient at 350 mA according to its datasheet.

Update 4/30/2012Cree XTEARY-00-0000-000000K01 is typically 50% efficient at 350 mA, with a typical peak wavelength around 455 nm.

UPDATE 4/30/2012 Usually, the best red LEDs for growing plants are InGaAlP ones with peak wavelength in the 630's of nm, since the best of those are usually more efficient than the best 660 nm LEDs. Plants tend to do almost as well with the shorter red wavelength as with the longer one.

Now, LEDEngin, Osram and Philips-Lumileds offer high efficiency high power LEDs with nominal peak wavelength of 660 nm.

Osram now offers their LH W5AM-1T3T-1-L-Z, available from Digi-Key.(Not always in stock.) Typical dominant wavelength is 645 nm, typical peak wavelength is 660 nm, typical "centroid wavelength" is 656 nm. Efficiency is typically 37% under datasheet-stated conditions including current of 400 mA. The datasheet states typical efficiency even higher at 44% at 100 mA. Efficiency is typically 93% as great at junction temperature of 60 C as at 25 C, and decreasing in an accelerating manner as temperature increases, according to its datasheet.

Lumileds now offers their LXM3-PR01, and the one currently available at Future Electronics appears to me to be the LXM3-PD01-0260. This is typically 37% efficient at 350 mA with a junction temperature of 25 C, and likely at typically least 34% efficient at 350 mA at a heatsinkable surface temperature of 50 degrees C. Multiply these by .9 for efficiency at 700 mA. Its active material is said to be AlInGaP.

Keep in mind that plant-growing fluorescents are about 30% efficient at producing desirable wavelengths of light for F32T8, 25% efficient for F40T12, and 20% efficient for the F20T12.

Keep in mind that blue LEDs operate less efficiently with more power (current more than 350 mA, or power greater than roughly 1 watt per LED), even if the amount of power and current is well within their ratings.

I mention more, especially on spectral response of chlorophyll, in my Photosynthesis Page.

660-670 nm red LEDs for skin phototherapy / healing

BIG UPDATE 11/11/2008!

I get asked a fair amount about getting 660 nm or 670 nm red LEDs, especially high power ones, for skin healing / wound healing / phototherapy purposes.

I am skeptical about this. I am aware of a fan or two of this e-mailing me an impressive list of studies supposedly supporting this.

The Big Update: I finally ran into something showing a photochemical mechanism for beneficial effect of intense 660 or 670 nm radiation as well as claimed before/after photos indicating that such light mitigates skin wrinkles. Prior to 11/11/2008, fans of red LED phototherapy had yet to point out to me a reference to a suspected or known specific photochemical mechanism.

The article to see: From Microtornadoes to Facial Rejuvenation: Implication of Interfacial Water Layers, by Andrei P. Sommer* and Dan Zhu, Institute of Micro and Nanomaterials, University of Ulm, Germany, 2008 - published in Cryst. Growth Des., 8 (11), 3889-3892, 2008. 10.1021/cg8000703

That article does mention the WARP 10 device, peak wavelength 670 nm FWHM 20 nm. Claimed to be successful was irradiation at intensity of 728 watts per square meter (72.8 mW per square cm) for duration amounting to a dose of 40 kilojoules per square meter - working out to 55 seconds. A different device providing 660 nm peak wavelength at an intensity of 70 w/m^2 (7 mw/cm^2) was not found to be effective.

I am still fairly skeptical on this, but less skeptical than I was before. This article was touched on by LEDs Magazine.

And I am aware of at least one acne treatment light using 660 nm and 410-415-or-so nm light sources - with the 660 nm source appearing to me to maybe be LEDs and the 410-415-or-so nm source appearing to me to usually be fluorescent lamps of the "03" code actinic type.

As it turns out, the shorter of these two wavelengths has a known anti-acne effect and the mechanism for this is known. A waste product of acne bacteria is changed by light of wavelengths around 410-420 nm into a chemical that is toxic to acne bacteria. However, except for one article noted above, I have yet to hear of specific mechanisms or chemicals or photochemical processes making any use of 660 or 670 nm light in the human body - with the exception of photopigments in the human eye, all known ones of which have peak response at wavelengths shorter than 600 nm.

As for LEDs to produce 660 nm:

Please be aware that the main LED chemistry for high output / high efficiency LEDs of this wavelength is used almost entirely for making low power LED chips. This chemistry is "GaAlAsP", and a minor variant is "AlGaAs".

Since this LED chemistry is so much a low power one, phototherapy lights based on 660 nm tend to use low power LEDs. Two easily available better candidates are Stanley H-3000L and the much lower cost Lumex SSL-LX5093SRC/E, available from Digi-Key. Almost as good are Lumex SSL-LX5093XRC/4 and Panasonic LN261CALUR, both available from DigiKey.

UPDATE 7/23/2011: Jump above to here for some high power, truly high efficiency 660 nm LEDs.

How do I get high brightness/efficiency green that is less bluish or more yellowish than 520-530 nm?

This is handled in my LED Color Adjusting/Modification Page.

I was sent here to look for goodies not mentioned above!

If you got to this point as a result of following a Usenet posting or other referral and did not find your answer by now nor anywhere above in this web page nor in pages linked therefrom:

First, explore My LED Top Page, and all links therefrom even if it takes several hours.

If this completely fails so badly that you think you are better off e-mailing a sort-of webmaster with a 50-hour-per-week unrelated fulltime job, then try e-mailing me, at Spammers beware - my sysadmin has a hobby of punishing spammers and I encourage and cooperate with him! Also beware that my response rate for non-spam e-mail from strangers is down to about 60%, maybe even less.

Back Up to My LED Top Page.
Back Up to My Home Page
Written by Don Klipstein.

Please read my Copyright and authorship info.
Please read my Disclaimer.