I need a cheap way to take a 3v nominal coin cell and power a 3.3 forward voltage blue led. It needs to be cheap in production on a large scale, so a rudimentary joule theif won't work.

Ideally it needs to be under $1 for my entire circuit, including pcb. I could go with two cells, but there's not many cheap 2 cell holder options. It also needs to be quite small, so that is also a constraint.

  • \$\begingroup\$ Why not just use an LED with a lower forward voltage? I found one quite easily on digikey with a V_f of 2.8V. \$\endgroup\$ – Hearth Oct 13 '18 at 2:08
  • \$\begingroup\$ What is production quantity? Through hole or SMT? Do you have a part number for the 3.3V blue LED? \$\endgroup\$ – Misunderstood Oct 13 '18 at 4:57
  • \$\begingroup\$ Direct connection or LED driver (look up Joule thief circuit). \$\endgroup\$ – KalleMP Oct 13 '18 at 6:37
  • \$\begingroup\$ @Felthry I assume you mean: LTST-C193TBKT-5A? I was planning on trying that, but if you check the datasheet, the max Vf is 3.05, and coin cells drop to under 2.5v when low. \$\endgroup\$ – Kyle Hunter Oct 13 '18 at 11:59
  • 1
    \$\begingroup\$ @KyleHunter commercial led booster ICs like the yx8018, the 5252F, sc8183, etc, basically simple boost switching regulators/joule thieves, are done on large scales. Source: every solar led lawn/path light at any target/walmart/dollar store. \$\endgroup\$ – Passerby Oct 13 '18 at 23:46

Connect the led to the coin cell. Nothing simpler than that. The high Equivilant Series resistance of the coin cell, as well as a lower voltage than the nominal 3.3-3.6V @ 20 ma, means that it's self regulating.

enter image description here

This is how led throwies work, and any super cheap single coin cell led "flashlight" works, with some plastic to bend the led pins onto the coin cell.

The physics behind it is that leds are like any other diode. They have a current/voltage curve that goes quickly up, so they have a small range of voltages where they are both visible and not dying. A typical blue led will be visible from a few hundred microamps/fractions of a million (pin point light) to 50 milliamps (bright pop of light as it explodes). This translates to a voltage below it's nominal to a voltage above it's nominal, maybe 2 volts of a difference. And it's nominal voltage at a given current varies. Two of the same brand leds could have 3.35V at 20mA while the other 3.43V at 20mA. And heat etc. The nominal forward voltage is an average and not required to light it. You can do it at a much lower current/voltage.

| improve this answer | |
  • 2
    \$\begingroup\$ Source of picture, my hand. Right now. \$\endgroup\$ – Passerby Oct 13 '18 at 2:17
  • \$\begingroup\$ But the 3v coin cell won't have a hold of gh enough voltage to power it? How does that work since the 3v coin cell is less than the 3.6 forward voltage? \$\endgroup\$ – Kyle Hunter Oct 13 '18 at 2:24
  • 1
    \$\begingroup\$ @Kyle leds work on a scale, from 0 current to too much current it explodes. The nominal voltage is where the led works at nominal current for n number of hours of life. By giving it less than the nominal voltage, it uses less current, and glows less brightly. That's it. \$\endgroup\$ – Passerby Oct 13 '18 at 2:27
  • \$\begingroup\$ @Kyle think of it like a car. You can give it half pedel of gas, it goes x miles per hour. Give it less gas, it goes slower. Give it too much gas, and it tops out and won't last as long as a car thats driven slowly/not at it's maximum. \$\endgroup\$ – Passerby Oct 13 '18 at 2:29
  • 1
    \$\begingroup\$ @Kyle, ill show you the Voltage vs Current graph of a led data sheet soon. \$\endgroup\$ – Passerby Oct 13 '18 at 2:46

Your most important decision is which LED.
You want the most photons per watt at the lowest cost.
Comparison of LEDs in the selection process is not as easy as it may appear.

Low power blue LEDs luminous intensity (lm/steradian) ranges from 5 mcd to over 5,000 mcd
View angles range from 15° to 140°
View angle has a huge effect on luminous flux measured in lumens (lm=mcd/steradian).
mcd is the photon intensity within a steradian (sr) cone measured in lm/sr.
The cone has an angle.

enter image description here

If we were to choose a minimum intensity of 100 mcd at 30° that is 0.021 lumens (lm).
Lumens includes every photon exiting the LED at any angle.

If the LED view angle is spec'd at 4x 30° (120°) 100 mcd @ 120° = 0.314 lm about 15x the flux of 0.021 lm @ 30°.

The radiant beam has a maximum distance. The distance of a flashlight beam is the distance where and illuminance is measured at 0.25 lux. 0.25 lux is about the amount of illuminance from a full moon.

Distance the light will travel is a factor.

100 mcd @ 1 meter = 100 lux.

At the given angle and distance the amount of area covered is the luminance (lm/sr/m²).


The intensity of the area (illuminance) measured in lux (lm/m²/s) add the element of time or brightness.

enter image description here

Bottom Line

The point being your purpose determines the importance of the view angle (i.e. steradian). And view angle is a very important property in the comparison of LED intensity.

You must understand the various measurement geometries of

  • Luminous Flux (lm)
  • Luminous Intensity (mcd)
  • Luminance (lm/m²)
  • Illuminance (lm/m²/s)

How to power the LED depends a lot upon the battery discharge curve. The good thing about a CR2032 is its flat discharge curve. The flat curve is favorable for a cheap current limiting resistor.

Notice the green (internal resistance) curve on the right image changes with capacity and the load remains constant. It is a myth that a CR2032's internal resistance is the same as a current limiting resistor.
For more about internal resistance and CR2932 current see this Texas Instruments app note: Coin cells and peak current draw

enter image description here

The Forward Voltage.

The forward voltage is a function of the current. This means you do not connect an LED to a constant voltage source without current regulation.

3.3 forward voltage blue led.

3.3 Vf is ambiguous. Most important is at what current. Most low power LEDs are spec'd at typical and or max current. The idea here is to get a Vf below the battery voltage.


Below is the Brightek QBLP674-IB
Brightek QBLP674-IB

Brightek QBLP674-IB datasheet
Brightek QBLP674-IB IV curve

The max Vf of this LED is 3.7V which is reflected in the graph. So it may be safe to say the graph represents the maximum Vf. But the 3.7V is at 100 mA. The LED can be powered at 100 mA with a 10%, 1khz duty cycle.

So if the target minimum luminous intensity at 120° is 10 mcd (10% of minimum 100 mcd). The current can be set between 500 µA and 1 mA. At this current the Vf is about 2.5V.

So given a battery voltage of 2.9V and Vf of 2.5, a current limiting resistor of about 400 Ω would suffice for 10-20 mcd @ 120° or 0.031 - 0.062 lumens.

And alternative would be a Broadcom ALMD-CB1E-VW002 with 5700 mcd @ 15° and 20 mA.
Broadcom ALMD-CB1E-VW002
Below 1 mA the Vf is about 2.6V. If the target intensity is 100 mcd (0.306 lm) the current would be about 350 µA (≈850 Ω resistor).

enter image description here

This graph shows the Vf at 20 mA is 3.2V.
This means this is a graph of typical values.
To be safe I would add 0.4V to the minimum.
When calculating the resistor value I would use 2.8V (min)
I would not use this LED unless I sampled a significant number of theses LED and measured the maximum Vf at 1 mA.

enter image description here

To convert from mcd to lumens I used the Rapid Tables online calculator

enter image description here
enter image description here

My Recommendation

My Recommendation is the Everlight EAPL2835BA3

Luminous Flux: 2.5 lm to 5.0 lm
View Angle 120°
Test current 50 mA
DigiKey Qty 4000 cost 6.182¢

enter image description here

How does this compare with the 9.360¢ (qty 2K) Brightek QBLP674-IB?
Luminous Flux: 100 mcd (min) to 210 mcd (typ)
View Angle 120°
Test current 20 mA

This means the Everlight is 66% cheaper, and its intensity is about 4x the Brightek.

Both LEDs spec'd a minimum luminous output and the same view angle of 120°.

The minimum intensity of the Bightek LED is 100 mcd.
The minimum flux of the Everlight is 2.5 lm.

enter image description here

2.5 lm converts to 795 mcd @ 120° that is 8x not 4x as I stated above.

When LED output is in flux (lumens) the total light emission is measured.
When LED output is measured in mcd (intensity), it is measured at 50%.

Also we must further reduce the Everlight intensity based on the "test current". The Everlight is spec'd at 60 mA and the Brightek at 20 mA.

View angle intensity is measured in radiometric. So if the wavelengths are not equal then you must do a photometric to radiometric conversion.

A 450 nm blue radiometric is 3.6% of photometric. 1 lux = 0.036 W/m²
A 470 nm blue radiometric is 1.6% of photometric. 1 lux = 0.016 W/m²

When an LED is rated in lumens (flux) and another is rated in mcd (intensity) is is not so easy to compare. Therefore you must look at the spacial distribution (view angle). View angle is measured at 50% relative intensity.

enter image description here

I find the non-polar easier for comparison.

enter image description here

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.