Blake said:
Not having any personal experience in the field, I'll stick to the "many
people believe" line. But the idea does have some credence among the
experts.
The following is an excerpt from On Semi ap note AND8067D
It would have been more helpful to the discussion if you provided the
link:
http://www.onsemi.com/pub/Collateral/AND8067-D.PDF
"There are two main reasons why LEDs are brighter when pulsed. First, the
human eye functions as both a peak detector and an integrator; therefore,
the eye perceives a pulsed LED's brightness somewhere between the peak and
the average brightness [4].
I would have to read the referenced document, "4. Smith, George,
“Multiplexing LED Displays: Appnote 3,” Siemens Semiconductor." before
I was willing to give this any credence. I think they are overusing
the reference.
Here is the closest I could find to this reference:
http://digchip.com/datasheets/parts/datasheet/000/APP03.php
(an excerpt)
"The luminous intensity, or the luminance of GaAsP LEDs, is
essentially proportional to forward current over a wide range, but
certain phenomena modify this condition. At low currents, the presence
of nonradiative recombination processes results in less light output
than the linear relationship would predict. This effect is noticeable
just below 5 mA per segment (for 1/ 4 inch characters). The result is
that noticeable difference in luminance from segment to segment can
occur at low currents. At high currents, the power dissipation in the
chip causes substantial temperature rise, and this reduces the
dissipation efficiency Figure 4.
As a result, the light output versus forward current curve falls below
the straight line, at high currents (Figure 5). It should be
emphasized that this latter effect is entirely due to self heating. If
the power dissipation is limited, by running short pulses at low duty
cycle, the output follows the straight line up to very high current
densities. Whereas 100 A/cm2 may be used in DC operation, as much as
104 A/cm2 can be used under pulsed conditions, with a proportionate
increase in peak intensity. (If this did not occur, GaAsP lasers could
not be built.) Gallium Phosphide, however, has an inherent saturation
mechanism that causes a drastic reduction in efficiency at high
current densities even if the junction temperature remains constant.
This effect is due to competing non-radiative recombination mechanisms
at high current density. As a first approximation the brightness of a
pulsed LED will be similar to being operated at a DC forward current
equal to the average pulsed current."
I think such peak response effect of the eye is weak to the point of
nonexistence for pulse frequencies that produce no visible flicker.
Sight is a photochemical process that produces output proportional to
the rate of photon arrival.
Thus, an LED driven by a high intensity low duty
cycle light looks brighter in a pulsed circuit compared to a DC drive
circuit that is equal to the average of the pulsed signal. The second factor
controlling the improved brightness is shown in the relative efficiency
versus peak current curves of an LED."
The curve shown is incorrect, if efficiency refers to light produced
per watt of electrical power. The curves show light output versus LED
current, not efficiency. Since all curves roll off to the right,
actual efficiency goes down above some current not much above the
normal 20 mA for all examples. This is science by salesman.
I suggest you set up the experiment. Design a circuit that alternates
a second at a fixed current with a second at a pulsed drive, with
the period, duty cycle and peak current settable. For many
combinations of period and duty cycle, visually find the pulse drive
current that produces no visible brightness changes on alternate
seconds. The pulse current setting would be unknown till you conclude
the no flicker condition, to reduce experimenter bias. I would like
very much to see your experimental results.
Then we can refer such discussions to your paper.