Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Edge Transient Effects on Power LED Switching
Imaobong E. Obot and Richard Binns Physics and Electrical Department, Northumbria University, Newcastle upon Tyne, U.K.
Email: [email protected], [email protected]
Abstract—The variation of high-power LEDs in
performance from manufacturer to manufacturer. However,
Information provided may not be sufficient in determining
concise switching properties of these LEDs at certain biased
mode. This has caused switching problems and performance
limitations depending on the manufacture type. This paper
demonstrates the rising and falling edge transient effects for
a number of different LEDs using MATLAB CAD tool. By
biasing these LEDs with external clock pulse, effects such as
exponential decay were observed at some pulse edges as
output. Therefore, highlights via experiment of the
problems seen in edge transients and proposition as to why
they occur were documented in this paper.
Index Terms—high power LED, rise and fall time, turn-on
and turn-off voltage
I. INTRODUCTION
The commercial availability and less environmental
consequence has paved way for extensive research and
development in the area of optimization of Light Emitting
Diodes (LEDs) as switching devices. Unlike other
illumination devices (e.g. incandescent bulbs) which are
voltage-driven, LEDs are current-driven. They are
capable of handling more current than the rated value
when connected to a power source of higher voltage [1]
and can control large amounts of current passing through
them. Hence, possess the ability to turn -on and –off at a
given time, t. The speed at which LEDs switches is
dependent on factors such as edge transient effects. These
effects can be seen at the rise and fall edges of an LED
pulse when exposed to certain biased voltage(s).
Power LEDs have gone through series of development
since 1999 [1]. This happened after the advent of blue
LED, followed in 1996 by the white version which was
created by adding phosphor coating to the blue LED [2].
They are designed to operate at >100mA forward current
value and power of 1 Watt (W) or more. Super-bright
RGBs or white LEDs from illumination point of view can
be used for fast switching [3].
Ref. [4] pointed out that the performance of LEDs
from different manufacturers can be a complex problem,
especially with a key specification in mind. Typically, the
operating voltage and current limit as well as other
parameters such as wavelength are provided in the
datasheet from the manufacturer of the LEDs. However,
Information provided may not be sufficient in
determining concise switching properties of these LEDs
at certain biased mode.
Manuscript received July 19, 2015; revised January 14, 2016.
This paper seeks to experimentally address the
underlining problems caused by transient at edge-
triggered power LED pulses when forward biased. Thus,
proposed reasons to explain the results seen at the
outcome. By pulse biasing different LEDs, the turn-on
and turn-off voltage(s) were observed and analyzed.
II. EXPERIMENTAL
A. Pulse Biasing Technique – Theory
The aim of this experiment was to observe how the
LEDs behave when a pulse signal was passed through
them. For the test power LEDs (RGB, white and Infrared
Emitting Diode i.e. IED) to be suitable for optical data
transmissions then we need to know exactly how they
respond to signals. Here, a pulse generator was used as a
power supply source (pulse biasing) for the LEDs. A
clock pulse was use in exciting the LEDs by sending the
pulse across the diode’s p-n junction.
+
-PULSE
GENERATOROSCILLOSCOPE
Vin VoutIin
R2
R1
Iout
Figure 1. Experimental setup – schematic circuit diagram.
Fig. 1 shows the experimental setup schematics. The
pulse period (T), delay and transition time were initially
set to 1µs, 1µs and 2ns respectively on the pulse
generator. A resistor, R1 was used to match the
impedance across the TX-line from the pulse generator to
the LED. The R2 resistor served as current limiter in the
circuit. The operating frequency was calculated thus,
frequency, f = 1/T = 1/1e-6 = 1MHz. The output clock
signal from the pulse generator was sent to channel 2 of
the oscilloscope. Channel 1 of the scope was connected
between R2 and LED using probes. R1 and R2 = 51Ω
each.
B. Equations
Generally, a p-n junction diode has I-V characteristics
for small voltages [5] given by the expression:
I = Is (exp (qV
kT) -1) (1)
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 392doi: 10.18178/ijeee.4.5.392-397
For normal or actual diodes, forward and reverse
biased modes are described in (2) and (3) respectively as
follows:
𝐼 = 𝐼s (exp (𝑞𝑉
𝑛kT) − 1) (2)
and
𝐼 = 𝐼s (exp (𝑞𝑉
𝑛kT)) (3)
where:
𝐼 = forward current of diode;
𝐼s = saturation current, the diode leakage current density
in the nonappearance of light;
V = applied voltage across the terminals of the diode;
q = absolute value of electron charge;
k = Boltzmann's constant;
T = absolute temperature (K); and
n = numeric factor (ideally, the value of ‘n’ lies between
1 and 2 according to [6]).
Let 𝑟𝑑 be the LED’s resistance. From Fig. 1 applying
voltage divider rule, we can deduce the following:
𝑉out = (rd
R2 + rd) 𝑉in (4)
𝑉out is the voltage across the test power LED. Since 𝑅1 is
in parallel with (𝑅2 + 𝑟𝑑) S.I units in ohms (Ω). The
current flowing through the LED 𝐼 = 𝐼out is given by:
𝐼out = 𝑉in
R2+ rd (5)
𝑉in is the forward bias voltage. Therefore, from (3), (4)
and (5):
𝑉out = 𝐼𝑠 (exp (𝑞𝑉
𝑛kT)) × rd (6)
and
𝐼out = 𝑉out
rd (7)
C. Tools Used
An Agilent oscilloscope was used in displaying the
pulse from the LED and measurement of the rise (10%-
90%) and fall (90%-10%) times for each LED. LEDs
used were green, red, blue and white with their
specifications. The rise and fall time was measured at
different bias level for different LEDs. By adjusting the
amplitude knob of the pulse generator, the different
values for the bias voltage was set and readings were
taken from maximum to minimum adjustments. The bias
levels for the experiment i.e. input voltage were randomly
chosen. Although, for blue and white LED two particular
bias levels were intentionally selected to monitor the
transient behaviour of both LEDs at that bias point.
Offset of ± 2.5V was used in adjusting the LEDs to
forward and reverse biased conditions for measurements.
The forward bias measurements were taken from 0V
(ground) while the reverse measurements were taken
from step of -0.5V to -1V depending on the LED
response. The test procedures described above which
involved linear circuit designs are straight forward and
easily implemented. However, we would expect some
level of inaccuracies in outcomes of the experiments due
to limitation on measuring devices. Analysis of the
relationship between LEDs’ rise/fall times and input or
bias voltage was carried out using the MATLAB R2012a
CAD tool. Appendix A is a list of datasheet information
on the test LEDs.
III. FORWARD PULSE BIASING MODE – EDGE
TRANSIENT EFFECTS
LEDs excitation with an external pulse from a pulse
generator yielded results which are shown in this section.
The pulses highlight the correlation existing between
applied voltages and switching time. At certain time 𝑡1
the bias voltage switches to a positive voltage, such that
the test LED is forward biased (turn-on). When time = 𝑡2
the voltage is reversed thus, turning off the LED. For
example in Fig. 3, 𝑡1 = 0.8ns while 𝑡2 = 0.2ns. The
negative sign on the x-axis is as a result of the scope
being triggered at the falling edge. It was assumed here
that the source resistance (51Ω) was large enough to
permit current flow in the forward biased mode. Due to
noise from the measuring device and environment, to get
a good capture of LEDs pulse, the scope was triggered at
both rising and falling edges. The RGB and white LEDs
were falling edge triggered while IED was triggered at
the rising edges.
Fig. 2-Fig. 6 shows the graphical representation of
results for the pulse bias experiment on test LEDs. Each
graph represents relationship between the bias or input
voltage and time taken for turn-on and turn-off. For the
purpose of this paper, maximum of two readings were
taken (pulse 1 and pulse 2) and recorded for each test
LED. The different in scaling depicts response of the
LEDs to input voltage.
Figure 2. Pulse generated from Luxeon star white LED graph.
-1 -0.5 0 0.5 1 1.5
x 10-6
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Time (ns)
Bia
s v
olta
ge
(V
)
Falling Edge Triggered Pulses
Pulse 1 biased at 2.41V
Pulse 2 biased at 2.65V
Exponential decay
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 393
Figure 3. Pulse generated from Luxeon star/o royal blue batwing LED graph.
Figure 4. Pulse generated from N74KG 5mm super-bright green LED graph.
Figure 5. Pulse generated from N73KG 5mm super-bright red LED graph.
Figure 6. Pulse generated from 5mm High IED graph.
-1 -0.5 0 0.5 1 1.5
x 10-6
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Time (ns)
Bia
s v
olta
ge
(V
)Falling Edge Triggered Pulses
Pulse 1 biased at 2.65V
Pulse 2 biased at 2.93V
Exponential decay
Spike before LED's turn-off
-1 -0.5 0 0.5 1 1.5
x 10-6
-1
0
1
2
3
4
5
Time (ns)
Bia
s v
olta
ge
(V
)
Falling Edge Triggered Pulses
Pulse 1 biased at 2.65V
Pulse 2 biased at 3.26VOscillating effect
Exponential decay
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
x 10-6
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Time (ns)
Bia
s v
olta
ge
(V
)
Falling Edge Triggered Pulses
Pulse 1 biased at 1.89V
Pulse 2 biased at 2.01V
Staggered steps before LED's turn-off
Oscillating effect
-1 -0.5 0 0.5 1 1.5 2
x 10-6
-1
-0.5
0
0.5
1
1.5
2
2.5
3
Time (ns)
Bia
s v
olta
ge
(V
)
Rising Edge Triggered Pulses
Pulse 1 biased at 1.21V
Pulse 2 biased at 1.29V
Energy
region at
constant
voltage
Oscillating effect
Exponential decay
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 394
For the red LED, bias voltage values were 1.89V, and
2.01V. Oscillating effects (also known as current ripple)
and were noticeable here. Pronounced oscillating effects
were noticed at 0V after LED’s turn-off.
The respective applied voltages for green LED biasing
were 2.65V and 3.26V. The same effects seen in IED and
red LED is noted here. Ripples with spikes were not seen
at turn-off although, it was observed at the tail-end of the
falling edge.
Applied bias voltages for the blue LED were 2.65V
and 2.93V. Interesting huge spikes were seen here before
diode actual turn-on (exponential rise in voltage). These
spikes were the cause of the negative voltage scale (see
Fig. 3). At 2.65V oscillating effect was absent at turn-on
whereas it appeared at 2.93V. Other huge spikes were
noticeable at the diode’s turn-off before an exponential
decay. The oscillating effect was absent at the tail-end of
the falling edges of the LED’s optical pulses.
The bias voltages for the white LED were 2.41V and
2.65V. Similar effects seen in the blue LED was noticed
here too except for the absent of ripples at the rise edges
of the white LED pulses.
On the other hand, bias voltage values for IED were
1.21V and 1.29V (maximum). IED’s response to voltage
changes was so sensitive that higher voltage readings
reading could not be obtained with the pulse generator.
Some oscillating effects with huge spikes were
observed here. These ripples occurred around the left and
right (just before turn-off event) part of peak voltage
forming sort of an exponential decay.
From the pulses graphs below, a linear rise in voltage
was observed in red and green LEDs while an
exponential rise was witnessed in the blue and white
LEDs. However, almost linear rise in bias voltage was
observed at high bias level for the blue and white LEDs.
Oscillating Effects (Current Ripple) Explanation
When a pulse is loaded down a waveguide, if
impedance is not perfectly matched, some of the energy
gets absorbed while some reflects back as distortion(s) on
the pulse. Oscillations or current ripples were seen as
such distortions on white, green, red LEDs as well as IED.
Although not verified, it was believed to have been
caused by parasitic elements as results of impedance
mismatch along the TX-line. Such elements included
probes’ capacitance, inductance and resistance.
Additionally, parasitic elements could be as a results of
packaging processes.
Observed Spikes Explanation
An explanation for the spikes before turn-on was not
concluded at the time of this report. But, possible reasons
could be linked to the dopants and other materials added
by manufacturers to increase efficiency and reduce
‘droop effect’ LED’s internal structure [7], [8]. Droop
can be simply defined as the loss of efficiency of LED
when operating at high power [7]. Also, an attempt could
have been made to charge the parasitic capacitor(s) (from
breadboard or probes), before charging the actual LED’s
inherent capacitor.
Rise and Fall Times of LEDs
The rise time and fall time of a pulse is simply defined
as the change in time from the 10% to 90% and 90% to
10% points of the voltage respectively [9]. As displayed
in the results, at different bias levels, the rise and fall
times of output pulse was greatly affected, resulting in
different values. This is explained as follows:
Assuming a step-function input pulse was asserted into
the LEDs.
From Fig. 7 mathematically, there was an exponential
change in the rising and falling edges of the pulse. The
output voltage, 𝑉out as a function of time, t, using the first
order of differential equation is summarized below:
𝑑𝑉out
𝑑𝑡= −∝ 𝑉out ⇒ 𝑉out (𝑡) = 𝑉o [exp(−𝑡
𝜏⁄ )] (8)
where:
𝜏 = 1∝⁄ = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = RC time constant;
𝑉out = output voltage;
𝑉o = bias voltage.
Figure 7. Rise time measurement for white LED.
Fig. 7 depicts a pulse from an RC circuit. It is believed
that when a diode is turned on, charges are built up. Let τ1
and τ2 be the time taken to charge and discharge the
capacitor respectively. Applying (8) with appropriate
substitution, an increase in the bias voltage causes LED
to turn on (i.e. charge). Therefore:
𝑉out (𝑡) = 𝑉o[1 − exp ( −𝑡𝜏1
⁄ )] (9)
The reverse is true at turn-off (i.e. discharge) resulting
in:
𝑉out (𝑡) = 𝑉o [exp(−𝑡𝜏2
⁄ )] (10)
Equation (9) occurred at 0V bias voltage. According to
[9] τ1 and τ2 are time constant of an RC circuit and are
related to:
rise time, 𝜏r = (ln 9)𝜏1 (11)
and
fall time, 𝜏f = (ln 9)𝜏2 (12)
The fall time determines how fast an LED can switch
off. Theoretically, from equation (5): 𝜏1 = 𝜏2 for RC
circuits. This means that, for an LED with RC step
function, the rise and fall times are meant to be equal.
Apparently from data obtained, this is not so. The reasons
for this behaviour were as follows:
1) The LEDs p-n junction does not emit light when
the bias voltage is not sufficient to turn it on.
-1 -0.5 0 0.5 1 1.5
x 10-6
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Time (ns)
Bia
s v
oltage (
V)
100%
90%
10%
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 395
A.
B.
C.
2) Due to the rapid discharge effect, the fall time of
LEDs decreases as the voltage increases although,
at a slow decay, the fall time increases. This can
be linked to the carriers sweep-out from the active
region [9].
The falling edges of the LEDs showed an exponential
decay due to abrupt discharge of charges. For a decay
constant 𝜆, the decay equation is:
𝑉out (𝑡) = 𝑉0𝑒−𝜆𝑡 (13)
Carefully considering Fig. 8, the falling edges reflects
a constant voltage right after the current ripples and
before turn-off. The IED seems to have two responses
just before discharge i.e. at 8ns and 12.5ns. This region
appears to store energy at a constant voltage (also refer to
Fig. 6). Energy stored is based on the principle that when
a diode is turned-off, the charges generated by previously
applied bias voltage are stored as the diode’s internal
capacitor discharges. The same effect was noticed as
‘staggered step’ behaviour in the red LED (Fig. 5).
Figure 8. Fall time measurement for IED.
As the bias voltage was increased, the extra energy
region increased as well. From Fig. 8 the falling edge of
IED was non-linear although, a linear calculation for fall
times of these LEDs was assumed in this work. The
constant or linear portion of the stepped edge was the
charge decaying linearly from the capacitance in the IED
of a more regulated bandgap drop. Consequently,
constant current decay then the exponential decay occurs
at the end following a capacitance general discharge.
Interestingly, the green LEDs showed less of this
behaviour as seen in Fig. 4. This was attributed to the
GaP (Gallium Phosphate) structure of green to the
GaAlAs (Gallium Aluminium Arsenide) structure of red
LEDs [1]. The p-n junction of the green LEDs is nitrogen
doped so at high voltage; there was colour shift where
yellow instead of green light was emitted. This colour-
shifting could have affected the falling edge, thus
reducing this behaviour in the green LEDs.
Moreover, the behaviour was not exhibited by the blue
and white LEDs since they both have InGaN (Indium
Gallium Nitride) structure [10].
IV. ENERGY STORED IN HIGH POWER LEDS
In LEDs, capacitor stores energy and gives it back to
the device when needed. For more energy to be stored
more bias voltage is required thus, more energy stored in
an LED implies that the LED can stay on for a longer
period. Energy stored is expressed as:
Ec = 12⁄ 𝐶𝑉2 S.I unit is Joules (J) (14)
Considering the capacitance value for blue LED at
2.65V, with current equal to 52mA (quite small), from
equation (16), the energy stored by the 14nF capacitor
will then be ~49E-9 J. The same value would be true for
white LED at the same input voltage. At a higher voltage
say 2.93V, more energy (~34.1E-9 J) was pushed into the
LED. Therefore, a greater voltage is required to push
more energy thus, increasing the switching speed of
LEDs. Another interesting point noted here about the
green LED is that, at 2.65V, the energy (~1.8E-9 J)
served to the green LED was much higher than that of the
blue and white LEDs at the same bias.
V. CONCLUSIONS
Although different manufacturers provide
specifications for a LED, it was noticed that these
information was insufficient. Hence, this work followed
the fundamental principles guiding LEDs via
experimentation. Thus, biasing the LEDs with pulses was
carried out to study the response time of LEDs to applied
voltage or current. For instance, we would expect high
fall time values in red and green LEDs as they were
labeled High Power LEDs. From results, the reverse was
seen implying that, it takes a short time for these LEDs to
turn off.
The current-voltage relationship is paramount in
choosing the right bias level thus, determining right
response time(s) for LEDs. This response time in effects
creates the rise and fall times needed for LEDs fast
switching. In general, this tells us that as the bias voltage
is increased, the energy level also increases.
TABLE I. DATASHEET SPECIFICATIONS [10]
Colour of
LEDs
Red Green Blue White Infrared
Forward
voltage (V)
1.85 at IF = 20mA
(max)
3.3 at IF = 20mA
(max)
3.99 at IF = 350mA
(max)
3.99 at IF = 350mA
(max)
1.2 at IF = 20mA
(max)
DC forward
current
(mA)
30 25 350 350 100
Average
forward
current
(mA)
- - 350 350
Reverse
voltage
5 5 - - 5
Peak
wavelength
(nm)
660 515 460 (max) 10000
(max)
940
Viewing
angle
34 34 10 110
Typical
intensity
3390 at IF = 20mA
(mcd)
8990 at IF = 20mA
(mcd)
120 (cd)
LED
junction
temperature
(°C)
- - 135 135
-2 0 2 4 6 8 10 12 14 16 18
x 10-7
-0.5
0
0.5
1
1.5
2
2.5
Time (ns)
Bia
s V
olta
ge
(V
)
Not to scale
90%
100%
10%
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 396
Analysis of results of the experiment using CAD tools,
unusual transient behaviours were noticed in the tests
LEDs. Biasing the LEDs with external clock pulse, the
output optical pulses exhibited some current ripples or
oscillating effects which was linked to impedance
mismatch along the TX-line. In addition, huge spikes
were observed before actual turn-on and –off in the blue
and white LEDs. These spikes, although not verified were
linked to the effects of pre-charging and discharging of
parasitic capacitors along the transmission line before the
actual LED turn-on or -off respectively. These effects
could be a major problem in switching of the LEDs to
obtain desired output(s). Therefore, it is advised that
before an LED is used for any application a test should be
carried out to ascertain LED’s specification to real time
application. For instance, Fig. 2 shows a turn-on at 2.5V
contrary to specification from manufacturer (see Table I).
APPENDIX DATASHEET SPECIFICATIONS
The test LEDs were LEDs, 5mm infrared emitting
diode, Luxeon star/o royal blue batwing, star white, 5mm
Super-bright red and green LEDs.
ACKNOWLEDGMENT
The authors wish to thank Northumbria University,
U.K, staff of the Physics and Electrical department and
Redeemer’s University, Nigeria.
This work was in partial fulfilment for Master’s
degree in Microelectronic and Communications
Engineering.
REFERENCES
[1] C. M. Bourget, “An introduction to light-emitting diodes,”
Hortscience, vol. 43, no. 7, pp. 1944-1946, Dec. 2008. [2] R. Forster, Light-Emitting Diodes: A Guide to the Technology and
Its Applications, Berkshire: BSRIA, 2005.
[3] C. G. Lee, “Visible light communication,” in Advanced Trends in Wireless Communications, M. Khatib, Ed., Croatia: INTECH,
2011, pp. 327-338.
[4]
Semiconductor Thermal Measurement and Management
Symposium, April 2009, pp. 151-158.
[5] D. Mynbaev and L. Scheiner, Fiber-Optic Communications Technology, New Jersey: Prentice Hall, 2006.
[6] Light Emitting Diodes: An
Introduction, Cambridge: Prentice-Hall, 1987. [7] U. S. Barbara. (April 19, 2011). LED efficiency puzzle solved by
theorists using quantum-mechanical calculations. [Online].
Available: http://www.sciencedaily.com/releases/2011/04/110419164211.htm
[8] R. Stevenson. (August 1, 2009). The LED’s dark secret. IEEE
Spectrum. [Online]. Available: http://spectrum.ieee.org/semiconductors/optoelectronics/the-leds-
dark-secret
[9]
[10] Quadica Developments Inc. (2013). Luxeon Star/O LED - Royal
Blue Batwing, 220mW. Technical Datasheet DS23. Power light source LUXEON® Star. [Online]. Available:
http://www.luxeonstar.com/Luxeon-Star-O-LED-Royal-Blue-
Batwing-220-mW-p/lxhl-nrr8.htm.http://www.luxeonstar.com/v/vspfiles/downloadables/DS
23.pdf
Imaobong E. Obot was born in Akwa Ibom State, Nigeria. She received the B.Sc. degree
from the Redeemer’s University (RUN), Ede,
2009 and M.Sc. degree from Northumbria University (NU), Newcastle upon Tyne, 2013,
in Physics with Electronics and
Microelectronic and Communications Engineering respectively.
Her research interests include optical sensors,
Bulk Recombination on Switching Characteristics of AlGaAs/GaAs pnpn Bistable Device, Nano-optical
devices and information theory.
Ms. Obot is a member of the IEEE and IET. She was a recipient of the Niger Delta Development Commission (NDDC) 2012/13 postgraduate
scholarship award.
Richard Binns attended Huddersfield University to study Electronics
and Information Engineering as an undergraduate and then did a PhD in Analogue Test Strategies. This involved the design of circuitry for an IC
and then deriving methods for testing the design built without direct
connection to parts of the circuits. The work basically centered on the clever application of current monitoring techniques and the simulation
of circuitry using pulses.
His interests include rebuilding servers, networks and operating systems. Dr. Binns joined Northumbria University in 1997 under a Post-Doctoral
research programme lead by Professor Phil Hallam. His research work
was on Analogue Synthesis techniques.
International Journal of Electronics and Electrical Engineering Vol. 4, No. 5, October 2016
©2016 Int. J. Electron. Electr. Eng. 397
K. Gillessen and W. Schairer,
A. Poppe and C. J. M. Lasance, “On the standardization of thermal characterisation of LEDs,” in Proc. 25th Annual IEEE
E. F. Schubert, Light-Emitting Diodes, second ed., New york: Canbridge University Press, 2006, pp. 393-396.