As current flows through light emitting diodes (LEDs), LEDs emit light in response to the current. But that is basic knowledge.

If we define that the light output of an LED is measured in foot lamberts (fL), the number of fL obtained versus the milliamperes supplied can vary greatly between individual devices. For example, consider two actual LEDs that I tested using a Model 1980A photometer made by Photo Research Company. Specifically, the instrument used was Serial Number C1066.

Note that three featured charts for the two diodes are actually all from the same data. They differ only in the degree of y-axis scale compression, which was done to make it easier to visualise the differences between the two LEDs over a wide dynamic range.

Chart LED fL light per mA Figure 1: One LED produces more fL of light/mA of current flow than the other.  

Now however, if we normalise these two diodes' curves, in this case by normalising the brighter LED to the dimmer one, we find close tracking between the pair of fL versus mA curves.

LED #3 intrinsically brighter than LED #2.

Chart LED curves normalised Figure 2: Normalised curves from the previous chart.  

It should be noted that although this apparent relationship is somewhat seductive and perhaps even suggestive of particular methods for designing a variable brightness, multiple-LED display, it is in fact a useless relationship for that purpose. We need look at that last statement a little more closely.

Although individual LEDs differ in their operational curves as seen above, they can be purchased in matched sets thus obviating the above differences. Having been thus purchased, their apparent brightness is best controlled by pulse width modulating their operating currents at the particular peak current levels for which their fL outputs are defined. This can give us very precise control of apparent fL outputs versus current pulse widths.

An LED's light output in fL can be demonstrated as being linearly proportional to the duty cycle by which its current pulse width is modulated for a specific but constant peak value of diode mA. To do this, output light was examined visually and examined objectively using the same photometer as before.

The LED pulsing setup was constructed as seen below.

Chart LED brightness duty cycle drive waveform Figure 3: LED brightness varies due to the duty cycle of the drive waveform.  

Pulsed current and DC current were applied to matched LEDs which were then viewed side by side. When their light outputs were the same, the DC current and the average of the pulsed currents were compared to each other. They were always equal.

Chart Human eye and photometer Figure 4: Photometer set can be used to measure the LED output in fL.  

A Model 1980A photometer set for a viewing angle of one degree can be used to measure the LED output in fL in response to the LED’s DC current in mA in the DC configuration shown above. Then, a pulse generator can be configured to deliver as many peak mA to the LED as there were mA of DC above.

It is then found by fL measurements vs pulse width that the fL seen by both the human eye and by the photometer is in direct, linear proportion to the pulse width duty cycle.

Originally published on EDN.