1. Using amount of current much less than 10-20 milliamps to power indicator lamp LEDs in battery powered equipment can allow increased battery life, and/or smaller battery size.
2. In line powered equipment, reducing LED current from 20 milliamps to 2
mA or less can save a customer 11 cents per LED per year assuming 5 volt
supply rail voltage, 80% efficiency of power supply and 11 cents per KWH
electricity cost.
The costliness of lower efficiency LEDs increases with supply rail
voltage.
3. LEDs and dropping resistors do produce heat. Reducing current through LEDs and dropping resistors by 8-20 milliamps sometimes means lowering the temperature of the entire product by .5 to 1 degree C, occaisionally 2 degrees C. By the "usual rule of thumb" for relationship between failure rate and temperature, making a product run 1 degree C cooler decreases rate of heat-affected failures by about 7%.
Such LEDs are "only a little dim" at 2 milliamps. Many of these do reasonably well at 4-5 milliamps, while being only a little brighter at higher currents due to efficiency decreasing as current increases past a couple milliamps or so.
To get such LEDs from Digi-Key: Go to their website www.digikey.com and type LED into the "search box" and click "Go". On the reulting page, find "Optoelectronics" and click the subheading "LEDs - <75mA, Discrete".
Next step is to click "Red" in the first selection column (color) and then click the button "Apply Filters".
Next step is to go to "Wavelength", the last column, scroll down to the longest wavelength numbers, and while holding down the "Shift" key on your keyboard, click both 697 and 700 nm (or everything in the range of 690 to 700 nm inclusive).
Narrow the choice down from there. I find typical is to want LEDs with wire leads, meaning clicking "Through Hole" for "Mounting Type", "Radial" for "Package/Case", and also click the "In Stock" checkbox under these selection columns, then click "Apply Filters".
Next that I would do is go to the "Viewing Angle" column and while holding down the "Shift" key, click all viewing angles 60 degrees or greater. In addition, there is a "Packaging" column, and for hobbyists and prototype developers there is usually one good choice among the 2 that remain - "Bulk".
Before clicking "Apply Filters", I would also use the "click-with-Shift" trick to rule out any that can be ruled out, since this "loop" ends with a list of items as soon as you get it down to no more than 20 items. For example, I try this to exclude sizes smaller than 3 mm and larger than 5 mm, and to exclude brightnesses less than 1 mcd.
As of late 2/16/2009, that gets the list down to 3 items, all of which are Panasonic items that are round and of size 3-3.2 mm. My favorite one of these, which I use, is Panasonic LN28RPX, with Digi-Key catalog number of P563-ND.
Panasonic LN28RPX is a red tinted diffused 3 mm / "T1" LED / "LED lamp". Although it is rated only 2 millicandela at 20 mA, I find it only mildly dim at 2 mA and quite reasonable at 5 mA.
I retry the above search with minor differences in use of search terms and I turn up 3 LEDs of this type by Lumex:
Lumex part #SSL-LX3044HD, 3 mm / "T1", very low cost, rated 6 mcd at 20 mA
Lumex part #SSL-LX5063HD, 5 mm / "T1-3/4", very low cost, rated 5 mcd at 20
mA
Lumex part #SSL-LX5093HD, 5 mm / "T1-3/4", very low cost, rated 5 mcd at 20
mA, very low cost
An older wide-angle "flat-top" "Fresnel lens" LED of this chemistry has part number of TLR-147 or TLR147. That one in my experience has served well at 3-5 mA and may remain available at a few Radio Shack stores.
More-modern brighter and more efficient LEDs with InGaAlP chemistry have similar wavelength and color specifications. The older technology of "high efficiency red" is highly available at low cost and in my experience tends to often require 3 milliamps to be "usefully bright" and to achieve "decent brightness" at 5 milliamps.
Specific LEDs of this type available from Digi-Key, round 3 mm (T1) and round 5 mm (T1-3/4) with "through hole" "radial" leads, viewing angle at least 60 degrees and lower cost include:
3 mm / "T1" ones:
Lumex part number SSL-LX3052ID
Lumex part number SSL-LX3044LID, notably lower cost
Lite-On part number LTL-10223W, noted as lacking a flange
Lite-On part number LTL-4221N, strangely costing more with inferior specs
Lite-On part number LTL-1CHE, lower cost
Kingbright part number WP132XID
Kingbright part number WP424IDT, 100 degree wide viewing angle flat-top
5 mm / "T1-3/4" ones:
Lumex part numbers SSL-LX5063ID, SSL-LX5093ID and SSL-LX5093LID
Kingbright part number WP483IDT, 100 degree wide viewing angle flat-top
Please keep in mind that flat-top / "cylindrical" / "cylinder" versions with wider viewing angles can easily require about 3.5-5 mA to achieve "good" brightness, and other versions with viewing angle closer to 60 degrees can easily appear dim if current is less than 2-3 milliamps or so. This particular LED chemistry tends to have efficiency maximized when current is at least 10 milliamps, and efficiency usually decreases significantly when current is reduced to around or under 4 millimaps.
Digi-Key has some offerings by Avago and Kingbright in this area. (through-hole, radial, test current 2 mA, nominal wavelength in the 568 to 635 nm range). It appears to me that Kingbright makes the "viewing angle" of these somewhat on the narrow side. I have only tested in this area one and it was the 5 mm yellow one by Avago, (HLMP-4719) and found it noticeably dim at 2 mA, but I found it "up to par" at 3.5 mA. High efficiency red ones (nominal wavelength 623 to 635 nm) should do well at 2 mA.
The Avago 2 mA ones have a rated maximum current of 7 mA.
Some of these are even characterized for performance at 1 or 2 mA. Ones characterized for performance at 2 mA should do very well at 1 mA.
Ones characterized for performance at either 1 or 2 mA and that are diffused red tinted 3 mm or 5 mm types include:
Avago HLMP-K150 and HLMP-D150 (60-65 degree viewing angle)
Kingbright WP7104LSRD and WP7113LSRD (30-40 degree viewing angle, and
with impressively low cost)
Panasonic LN28RALXU does do well at low currents, does somewhat well at 1 mA,
is characterized at 5 mA (3.8 mcd at 5 mA) and has a 70 degree viewing angle.
Not characterized at low current but also likely good are Lumex SSL-LX5093SRD (60 degree 5 mm), SSL-LX2573SRD (110 degree rectangular), and SSL-LX25783SRD (a larger rectangular 110 degree one). The Lumex ones have fairly low cost.
All of these have maximum continuous current of 30 mA, including the ones characterized at lower currents as low as 1 mA.
My favorites here are currently Nichia NSPG-F50AS (rectangular) and the less costly, available-from-DigiKey TT Electronics / Optek Technology OVLLG8C7. Nichia NSPG520AS and NSPG320C are excellent with extreme brightness and are likely to look good at .1-.25 milliamp, although they only have a 45 degree nominal viewing angle.
The OVLLG8C7 supposedly produces 670 millicandela with a viewing angle of 85 degrees. I find closer to 1500 mcd at 20 mA and 20-25 mcd at .55 mA. Expect it to be useful as an indicator lamp at current at least as low as .35-.4 mA.
CAUTION - InGaN LEDs are static-sensitive.
UPDATE 3/17/2009: Some (maybe most) recent Nichia green LEDs such as at least some lots of NSPG520AS and NSPG320C have efficiency remaining sufficiently high at currents down to .05 milliamp to have little benefit from pulsing.
I have developed a "4049 abuser" oscillator, which I show below a version for 9 volts supply voltage a green InGaN LED.
(This circuit will be moved to a separate page by the time I draw any more.)
This circuit as shown draws about 5-6 microamps for the oscillator and about 40-50 microamps total, with 9 volt supply voltage and an InGaN LED.
It is shown to switch a MOSFET such as International Rectifier IRLML2803PBF.
With this, some InGaN green LEDs get "sufficiently bright" (reliably 3-plus millicandela even when viewing angle is rated to be at least 85 degrees) at 45-50 microamps average current, including the current used by the "4049 abuser" oscillator.
Please keep in mind that deviating from shown component values in the oscillator circuit may impair efficiency.
Increasing C1 from 330 to 390 or 470 pF is unlikely to impair efficiency, but pulsing rate will decrease from about 165 Hz (with 330 pF) as C1 is increased. The 115 Hz likely with 470 pF is unlikely to be problematic, though strobing effect increases as frequency decreases. Much lower frequencies below 100 Hz are not recommended, due to strobing effect becoming more severe and flicker often being visible at 60 Hz and possibly having minor subconscious irritation effects through somewhat higher frequencies.
With 5 volt or 4.5 volt supply voltage, the MOSFET can be substituted with one of the unused inverters of the 4049. Also at 5 volts, the value of R3 should be decreased from 1 megohm to 330 K - and decreased to 220K for 4.5 volt supply voltage, 270K for 4.8 volts or 5 volts that may run a little low.
RLED should be decreased from 2.7 K to 1.0 K for 5 volts, 680 ohms for 4.5 volts, and 820 ohms for 4.8 volts or 5V that runs a bit low.
This circuit can be used with white LEDs and yellow/orange/red high brightness ones, though RLED will likely need to have value decreased to extent of average current getting to 100-125 mA or so. This means 1-1.5 K at 9 volts, maybe 470 ohms for white and 680 ohms for yellow, orange and red at 5 volts, maybe 330-390 ohms for white and 470-560 ohms for yellow, orange and red at 4.5 volts.
"Direct coupled oscillators" also work, but my experience suggests that these are better with extreme low duty cycles and are better suited for when the pulse rate is low enough for the LED to visibly blink. It may be possible to achieve flash rate of a few Hz with suitable brightness and average current draw 30 microamps or less with a variant of this LED flasher at DiscoverCircuits.
Also suitable would be a variant of this ultra low power LED flasher using a low power comparator IC, at DiscoverCircuits.
Copyright (C) 2009 Donald L. Klipstein, don@misty.com.
Please read my Copyright and authorship info.
Please read my Disclaimer.