Version 3 Beginnings

A project log for TritiLED

Multi-year always-on LED replacements for gaseous tritium light sources

Ted YapoTed Yapo 12/22/2017 at 17:080 Comments

In honor of the Coin Cell Challenge, I thought I'd put together a V3 design.  As if by providence, DigiKey just started stocking the Oslon Signal series LEDs.  Of particular interest is the Verde color, with a 505nm output peak nicely coinciding with the scotopic (dark-adapted) peak sensitivity of  the human eye.  I re-assembled the #Automated LED/Laser Diode Analysis and Modeling system and measured the relative efficiency of this LED:

I was quite surprised to find that this LED has a peak efficiency between 3-4mA.  From the datasheet, the recommended current is 350mA, and the max 1000.  I didn't have a good heatsink solution, so I only ran the LED up to 40mA for testing.

There's enough information in the datasheet graphs to calculate the relative efficiency for currents greater than 30mA, so I digitized the If vs Vf and relative intensity vs If graphs on page 9.  Here's a plot like the above, but created from the datasheet values:

According to the graphs in the datasheet, the efficiency continues to drop with increasing current.  Between the measurements and the datasheet values, it appears that this LED is about 4.5x as efficient between 3-4mA than it is at the maximum of 1000mA.  Even compared to the recommended current of 350 mA, the LED is still 3x more efficient at around 3 mA.

To see how the curves correspond, I scaled the relative scales so that they match at 30mA:

The measurements show a slightly higher efficiency than the datasheet at 40mA, but considering the datasheet curves were calculated with values scraped from plots, I'm not overly concerned about it.

Oddly, the datasheet says not to use the LED under 30 mA - I have no idea why there's a minimum current rating for an LED.  Does anyone know?

I'll admit that I was lazy with a lot of the recent LEDs I used on V2 boards, assuming they would have their efficiency peak around 20mA.  I'm glad I took a closer look at this one before completing the V3 design.

Optimal Sawtooth Waveform

Assuming an inductor-flyback drive like the V2 design, a little math gives the optimum peak for the sawtooth current pulse:

The ideal peak current is 4 mA, although the efficiency remains above 95% for currents up to 10mA, and above 90% until around 18 ma.

Since the current will fall with the coin cell voltage, I probably want to start the current a little higher to maximize efficiency over the lifetime of the cell.