|
EDUCATE YOURSELF
LED FAQs:
Q: What's "magic
smoke"?
A: When an LED is blown, it
emits a small wisp of smoke. The smoke
is then jokingly said to have been the
magical element that made the LED work,
and its release renders the device
nonfunctional.
Q: Is turning an
LED on and off bad for it? What's "PWM"?
A: No. LEDs are semiconductors, which
are quite happy with transitioning
between an "on" and an "off" state. In
fact, most LED flashlights provide extra
output levels by flickering the LED at a
faster rate than the human eye can
usually detect. When the output is not
filtered and smoothed out, the flicker
can be seen when the light source,
subject, or viewer moves. This flicker
is commonly known as "PWM," or "Pulse
Width Modulation." Unfiltered PWM must
be driven to frequencies of at least
50Hz, or else the flickering will be too
pronounced and will irritate users.
Q: How can an LED be more
efficiently driven?
A: LEDs are more
efficient at lower drive levels. This
means that slow, unfiltered PWM is less
efficient than filtered and smoothed
output, known as "current regulation."
This is because an ordinary PWM light is
active part of the time and completely
off at other times, but during the
active portion, the LED is being driven
at full blast, which is comparatively
inefficient. Therefore, a multi-emitter,
current-regulated light is more
efficient than a single-emitter, PWM
light, as the LEDs in a multi-emitter
light will operate more efficiently (and
use less power) than the LED in a
single-emitter light when pushed to the
same amount of output.
Q: What's
the "Luxeon Lottery"?
A: Different
samples of "white" Luxeon emitters, even
ones of the exact same classification
(known as "bin"), can vary in their
light output and tint at a particular
drive level. Some are slightly green, or
purple, or other colors, while others
have less noticeable coloring. This term
can apply to any high-powered LED with
natural variations in tint and output.
Q: What's "Vf"?
A: Vf is the term
for an LED's forward voltage. Forward
voltage is the voltage required to
activate an LED. For example, if you
want to power a Cree XR-E at 150mA, it
needs about 3V. If you want to power it
at 700mA or thereabouts, the Vf will
rise to about 3.8V. It's like a minimum
activation voltage for particular
currents.
Q: Why is Vf important,
and how do I measure it?
A: To
determine a particular LED's Vf, you'll
have to look at its datasheet. For
example, Cree's XR-E datasheet is found
at
http://www.cree.com/products/pdf/XLamp7090XR-E.pdf.
In page 6 of that PDF, you'll see a
graph of Vf. This means that the Vf of a
particular emitter is not fixed, but
varies, with the available voltage
determining the emitter's "natural"
drive level. For example, if you look at
where the graph reads "3V," you'll see
that it corresponds to a "Forward
Current" drive level of about 160mA.
This means that if you solder an XR-E to
the appropriate terminals in an ordinary
Minimag, which operates at 3V, you'll
get about 30 lumens of floody, white
output.
The downside to this is
that small changes in voltage can have a
huge effect on forward current. For this
reason, most drivers supply a constant
current instead of just a constant
voltage, as current is really what
determines an emitter's output.
Therefore, if you want to power an XR-E
at 700mA from a pair of AA NiMHs (a
common task), you'll need a driver that
can boost the voltage from 2.4V to
something closer to 3.6V as well as
supply a constant 700mA. If you tried
direct drive with this setup, simply
connecting the emitter to the battery,
the emitter would barely light up at
all.
It is important to note that
even emitters of the same type and "bin"
(sorting system based on several
characteristics) can have small
variations of Vf. For example, you may
have an emitter that has a Vf of 3.7V
when driven at 1000mA (or 1A), while
someone else has an emitter of the exact
same type and bin with a Vf of only 3.5V
at the same 1A current. This means that
their flashlight will consume only 3.5W
of power while yours consumes 3.7W,
although both would have the same output
given the identical drive level.
You can find the Vf of your emitters at
particular drive levels with a basic
DMM. If you are using a driver and you
don't know the current it provides, or
if you're not using a driver (a DD
setup), put the DMM in the circuit next
to the emitter and check the current.
Next, complete the circuit without the
DMM in it, and put the DMM's + and -
probes on the emitter's + and -
contacts, respectively. This is the Vf
at the current level you measured for
this particular emitter.
Most
emitters can handle a bit of current at
reverse voltage, so if you wire one
backwards but don't activate it for more
than a second or two (at most), it
should be okay. Just don't make a habit
of it.
Q: What's "bin"?
A:
Bin codes are used to sort LEDs by
luminous flux (lumen output at a
specified drive level), color, tint, and
Vf. For example, a U-bin Lux III will
have more output than a T-bin Lux III at
a set drive level. Commonly desirable
flux bins (at the time of writing) are:
-Lux I: R or S
-Lux III: T or U
-Lux
V: W or X
-XR-E: Q2 through R2
-SSC P4: U or V
-Rebel: 0080 or 0100
Q: What's the deal with LEDs and
heat?A: Forget anything you learned
about LEDs from the National Geographic
Channel's "Manmade" show focusing on
flashlights, especially from the
interview with the Philips employee.
LEDs are quite efficient relative to
other light sources, but they do produce
heat. In fact, most aren't even 30%
efficient! This means that the more
power you pump through them, the more
light and heat they will produce. Unlike
incandescent bulbs, however, LEDs are
actually damaged by heat. It's common
for a well-driven power LED to exceed
120F (quite hot to the touch). Too much
heat for prolonged periods can decrease
the life of an LED or even kill it. The
efficiency (and therefore output) of an
LED suffers with heat as well, meaning
that with most lights, there is a
certain drive level above which the
increased heat will actually result in
less output than a more moderate drive
level. To combat this, well-designed
flashlights provide a method to get the
heat away from their LED. The most basic
(and by far the most common) method is
the heat sink. This is nothing but a
chunk of metal that contacts the LED and
is heated by it, leeching the damaging
heat away from it. The next step is to
somehow transfer that heat to the
environment, where it can dissipate.
This means that the heat must have a
"thermal path" which leads from the LED
to the heat sink to the surrounding
flashlight to the environment. Some
flashlights benefit from being held by
someone's hand so that their bloodstream
can act as a heat pump (the blood near
the flashlight is heated, moves away,
and cools, and the cycle continues).
Other lights have fins that increase the
surface area which contacts the outside
air. LED dive lights don't have much of
a problem here, since the surrounding
water is like an enormous heat sink.
Bike lights benefit from the cool night
air rushing past them.
Q: What's
the deal with LEDs wired in series or
parallel? What about "current hogging"
and "thermal runaway"?
A: If your
input voltage and current are
acceptable, LEDs can be run in series or
parallel. The issue is how well the
setup will work, and how reliably. LEDs
experience "Vf shift" when they heat up,
with Vf dropping as temperature rises.
With a constant-current (CC) source,
this is no problem, as the only effects
will be less light (because of the heat)
and less power consumed (a lower Vf
affects the V*A=W formula). Otherwise,
there won't be much to worry about. With
a constant-voltage (CV) source, however,
the Vf shift comes into play. If Vf for
a particular current goes down, but
you're only keeping the VOLTAGE
constant, the current at that voltage
will go up. This is a problem because it
will cause more heat and further Vf
shift, drawing more current, leading
into a feedback loop where the LED(s)
eventually pop from the stress. This is
what is called "thermal runaway." As you
might guess, this is an issue when
deciding between a series or parallel
LED setup. Recall that devices in series
will all draw the same current (at
whatever voltage each device needs for
it), while devices in parallel all see
the same voltage while having
potentially different current levels
going through each one. If you have a CC
driver running a series LED string (a
"string," or "leg," is a group of
devices, like LEDs, connected in
series), you won't have thermal runaway
problems. However, LEDs in parallel,
even driven by a CC source, can
experience "current hogging." As the
devices are in parallel, they won't
necessarily have the same current. Each
string in a parallel array driven by a
CV source can experience thermal
runaway. If driven by a CC source, one
string can still undergo thermal
runaway, but another string sharing the
CC source will simply get less of the
total current - the current was "hogged'
by the other string. If your CC source
in such a setup is set to drive LEDs
near their limits, then having one
string bear the load intended for two or
more strings can be problematic.
Remember that when LEDs pop, the circuit
breaks, leaving the other strings in a
parallel setup with a kind of surplus of
current, which leads to even more rapid
thermal runaway, which will eventually
leave you in the dark. This is why it's
generally recommended to wire LEDs in
series, as well as using a CC source if
possible. If you don't have a CC source,
use an adequate resistor to limit
current.
Q: LEDs are all the
same, right?
A: Wrong. See LED Types.
LED Types:
5mm/Nichia:
This
is the ultra-common low-power LED that
gives off a wide spot of directed light.
They come in many colors, and the
"white" version is usually bluish, to
some degree. They cost about $0.15 each.
The Vf of these LEDs varies with color,
with red 5mm LEDs having a Vf of a
little over 2V and white 5mm LEDs
needing about 3.5V. An ordinary white
5mm LED will produce from 5-10 lumens at
a safe drive level of about 20mA.
Because of this very low current
requirement, these LEDs are ideally
suited to keychain flashlights that run
on small button or coin cells, as these
cells can't handle much current anyway.
Luxeon I/III:
These have been the
standard range of high-power LEDs for
years. They cost between $8 and $15,
depending on bin. Prices dropped along
with demand upon the release of the new
high-efficiency power LEDs (explained
below). The only difference between
Luxeon I and Luxeon III is their speced
drive currents. Lux Is are rated for a
particular light output level at 350mA,
and Lux IIIs are rated for a particular
light output level at 700mA. This has
led to the common marketing practice of
calling Lux Is "1-watt" LEDs and Lux
IIIs "3-watt" LEDs. This is misleading,
because these wattages only describe the
speced capabilities of these emitters (Vf
of 3.4V * 350mA = 1.2W, and Vf of 3.7V *
700mA = 2.6W) and not the actual power
consumed by the LED in any particular
flashlight. Thus, a "3W LED flashlight"
could be running at any actual drive
level.*
A common Luxeon I driven
at a decent 1W will provide between 30
and 60 lumens, depending on bin. A
common Luxeon III driven at a decent 3W
will provide between 60 and 90 lumens,
depending on bin. They will emit those
lumens in some sort of pattern:
Lambertians emit more of their light
straight forward, creating a nice
180-degree flood. Batwings are similar,
but emit slightly less light in the
center. Side-emitters emit most of their
light to the sides.
* Some
misleading sellers advertise LED
flashlights as "10W," "15W," and so on.
Even if these figures were accurate,
they would only relate to power
delivered by the batteries, as most of
those dozen watts would have to be
burned up by a resistor if such a
flashlight were to last more than a few
minutes. The LEDs used in these
flashlights could never actually take
10-15W of power.
Luxeon V:
Starting at around $10 each, these are LEDs with four "dice," or light-emitting
chips. They still fit into a small
package, but the larger emitting surface
makes for focus and throw challenges.
It's all too easy to get a "donut hole,"
which is a beam that has a darker spot
in the center (like a donut; get it? ),
and it's much more difficult to throw a
small, tight spot with these than with
the Luxeon I and III explained above.
The benefit of these LEDs is that they
have more output, with most samples
reaching 100-140 lumens at proper drive
currents (around 700-900mA), depending
on bin. The downside with this package
is the higher Vf of 6-8V, which
necessitates specialized driver circuits
and consumes more battery power. Luxeon
V LEDs are available in Lambertian and
Side-Emitting packages (explained
above).
Luxeon K2:
These came out
a couple years ago, and they bore the
hopes of flashaholics everywhere.
Functionally, they were like Lux IIIs
speced for drive currents in the
neighborhood of 1.5A, with increased
lumen output and heat tolerance.
However, multiple delays, limited
availability, and competition with the
new high-efficiency power LEDs
(explained below) made it a
disappointment. It wasn't a leap forward
so much as a shuffle in a general
direction. For these reasons, they never
met with much success.
Cree 7090
XR-E: These came out in the fall of '06,
generating a flashaholic frenzy. They
could be driven at voltages and currents
similar to Luxeon IIIs, but they were
twice as efficient, meaning that in two
flashlights identical except for the
emitter, the XR-E flashlight would have
twice the output! LED flashlights
producing 150 lumens became a reality.
Better bins released over the coming
months provided even more efficiency and
output. Unfortunately, they are not
direct replacements for Luxeon lights
due to a unique package (explained
below). They cost between $6 and $15,
depending on bin and seller.
SSC
P4: Released a few months after the XR-E, these use the XR-E's EZ1000 die,
giving them the same high efficiency and
output as the XR-E. However, they have a
significantly different package, bearing
more resemblance to the old Luxeons.
They are approximately $7 to $15,
depending on bin and seller.
Edison Opto KLC8:
These are
high-efficiency power LEDs also using
the EZ1000 chip in a package similar to
the Lux III, but with some differences
(explained below). They are around
$4-$6.
Luxeon Rebel:
These came
out in mid '07. They have the same
efficiency characteristics as the XR-E
and Seoul, but the package (and intended
application) is very different from the
other high-power LEDs. They are ideally
suited to surface-mount applications
assembled with reflow soldering.
Luxeon K2 TFFC:
This is an updated
version of the K2, with new "Thin Film,
Flip Chip" technology. It can compete
with dice like Cree's EZ1000 (used in
the XR-E and SSC P4).
SSC P7:
Released in early '08, this is basically
Seoul's version of a Luxeon V, using
their P4 as a base. It has four dice all
wired in parallel (compared to the
LuxV's 2S2P), with the expected output
of four SSC P4s.
Practical
considerations:
Lux Is and IIIs have a
hard acrylic dome, smooth 180-degree
radiation pattern, negative slug
(underside), and easily soldered "legs."
Lux Vs are similar, but with higher Vf
and larger emitting surface. Luxeon K2s
are also similar, but can be driven at
higher power levels. XR-Es have a narrow
radiation pattern (about 120 degrees)
that projects a very faint gridlike
pattern (bare, without optics) which
lends itself well to collimation by
aspheric lenses, a floating acrylic
dome, and tiny pads that are very
difficult to solder. The SSC P4 has a
soft gummy dome to which dust sticks,
very similar appearance to a Lux III
except for tiny metallic portions
visible surrounding the dome, smooth
180-degree radiation pattern, positive
slug, easily soldered legs, and the
tendency to tint-shift toward a cool
blue when it heats up (if improperly
heatsinked). The KLC8 has a positive
slug, 180-degree radiation pattern,
acrylic dome, legs, slightly less
luminous flux when compared to the other
EZ1000 packages, and collimation by a
reflector results in a yellow-green ring
around the hotspot. The Luxeon Rebel has
a rectangular board with the emitter
surface located off to one side, a
neutral slug, and tiny pads that are
very difficult to solder. The Luxeon K2
TFFC is similar to the ordinary K2. The
Seoul P7 has quite a large overall
package, and the parallel dice make the
quest for suitable drivers a major
challenge, but it has excellent output,
especially for its size compared to
single-die emitters.
NEXT CHAPTER>>
|
