Pulse driving vs Normal Constant Current Driving Leds

[richon]

is Pulse driving better at "lower consumption" and "higher lumens" vs Normal Constant Current Driving a led Leds?


recently I build some simple Mosfet constant current drivers for some leds in my house...

but as I can't keep steady... I read that some use a 555 IC to drive the leds with pulses of higher current but lower times oscilation to avoid burning/overheating the led and it would make a led get more lumens than with constant current driving.




Ricardo.

 

[Pöbel]

if the brightness is the same:

Constant current: more efficient (lower vf), tint might shift at low current
PWM: no tint shift, less efficient (higher vf because higher current), possible flickering

 

[richon]

for example:

a 3 array of 3Watts leds with 0,8A consumption with constant current driver would use more or less current with PMW? would it be more lumens pero watt with PMW? would it damage anyhow the led with PMW

 

[sn0wBLiND]

You'll still need to limit the maximum current, PWM alone isn't enough to limit it.
What LEDs are you using and how many? What is your input voltage?

CREE has a datasheet that could give you a few tips: http://www.cree.com/products/pdf/XLamp-Pulsed-Current.pdf

 

[Curt R]

The major problem with PWM is that most circuits do not ramp up
the current drive to the LED with each pulse. This results in a high
current shock to the junction point of the bonding wires in the LED.
Sputtering of the wires at that point cause the wires to erode over
time causing an increase of resistance, that causes heat build up and
failure of the wires at that point. In an application that is on for a
considerable amount of time the LED life is shortened. With a flashlight
that is used normally and not in a job related application there should be
no problem.

Curt

 

[xul]

The eye sort-of responds to peak brightness so you can save w-h by pulsing.

 

[SemiMan]

The major problem with PWM is that most circuits do not ramp up
the current drive to the LED with each pulse. This results in a high
current shock to the junction point of the bonding wires in the LED.
Sputtering of the wires at that point cause the wires to erode over
time causing an increase of resistance, that causes heat build up and
failure of the wires at that point. In an application that is on for a
considerable amount of time the LED life is shortened. With a flashlight
that is used normally and not in a job related application there should be
no problem.

Curt

Is this fact or assumption? If fact, can you please post a link to the article that describes this mechanism with hopefully real world testing.

I have serious doubts as to how much of a factor this would be considering MOSFETs in a power supply switch amps (or 10's of amps) repetitively, 100KHz or more, super fast, and repeatedly for 10's of thousands of hours. Or maybe even more similar, the diode in a buck regulator is hit with amps similar to the above FETs.

Semiman

 

[IMSabbel]

if the brightness is the same:

Constant current: more efficient (lower vf), tint might shift at low current
PWM: no tint shift, less efficient (higher vf because higher current), possible flickering

An exception would be _really, really_ low light levels on large LEDs.

If you want to get something like 0.1 Lumen on a SST90, you could be better off using a low duty cycle PWM with 300mA than trying to CC drive it at 0.25mA, due to intrinsic losses, etc.

 

[Kinnza]

Two reference threads on this topic with lots of good info:

http://www.candlepowerforums.com/vb/showthread.php?70073-Constant-Current-vs.-PWM-dimming-Revealed

http://www.candlepowerforums.com/vb...-How-does-it-work-and-how-to-detect-it./page2

 

[Kinnza]

The eye sort-of responds to peak brightness so you can save w-h by pulsing.

Eye is a good integrator of light, so in general sense, it not respond to peak brightness but averaged one. Eye fails doing it on some special conditions: short light pulses (below 0,1s, effect most noticeable at 50ms pulses or so), good lux levels (more properly, high luminance levels). So this trick is used often for signaling lights (as for planes and airports), but it is way less useful on lighting tasks and for sure it can't be generalized. And this effect on increased perceived brightness is always accompanied by sensing flickering.

 

[xul]

always accompanied by sensing flickering.

http://en.wikipedia.org/wiki/Persistence_of_vision
?

 

[Kinnza]

I referred to enhanced brightness perception always being linked to flickering. When frequency is high enough such there is no flickering sensation, meaning eye is integrating light perfectly, there is no enhanced brightness sensation. Enhanced effective brightness is often used at frequencies at where there is no direct flicker sensation, but still there is stroboscopic effect, which, at the end, is a kind of flicker effect.

 

[Curt R]

Is this fact or assumption? If fact, can you please post a link to the article that describes this mechanism with hopefully real world testing.

I have serious doubts as to how much of a factor this would be considering MOSFETs in a power supply switch amps (or 10's of amps) repetitively, 100KHz or more, super fast, and repeatedly for 10's of thousands of hours. Or maybe even more similar, the diode in a buck regulator is hit with amps similar to the above FETs.

Semiman

INTRODUCTION
This application note describes electrical overstress (EOS) events, their effect on Cree XLamp LEDs and some simple methods of protecting XLamp LEDs against EOS. Electrical overstress is simply exposing an LED to any current greater than the maximum cur*rent specified in that LED's data sheet. The number or length of EOS events is irrelevant because any single EOS event can cause damage to the LED. This dam*age can be exhibited either in an immediate failure or a failure many hours after the EOS event.

2. Transient over-current events Transient over-current events are events that subject the LED to current that is higher than the maximum rated cur*rent on the LED data sheet, either directly through high current or indirectly through high voltage. These events are transient, meaning they happen for a short period of time – typically less than one second. These events are sometimes referred to as "spikes," as in "current spike" or "voltage spike." If the over-current event occurs immediately when the LED is turned on or plugged into a energized power supply (also called "hot plugging"), this over-current event is called "in-rush current."

EFFECTS OF ELECTRICAL OVERSTRESS ON XLAMP LEDS
It is impossible to predict the failure mode of every LED exposed to electrical overstress, but Cree has seen two common symptoms of XLamp LEDs that have had an EOS event cause a catastrophic LED failure. Damage to Bond Wires One common failure mode from EOS is damage to the bond wires inside the LED package, as illustrated in Figure 1 below. This damage usually occurs as a burned wire or a broken wire. In addition, the EOS event can cause damage to other materials in close proximity to the bond wires, such as the encapsulant or phosphor. Damage Near Bond Pads Another common failure mode from EOS is damage to the LED chip itself near the bond pads, as shown in Figure 2 below.

As described in the Causes of Electrical Overstress section, one common form of over-current event occurs when LEDs are connected to an energized power supply or when the power supply is first turned on. This event is called in-rush current.

Cree strongly recommends adding some level of protection to LED modules that do not include an on-board power supply to minimize the risk of an in-rush current event from the separate power supply.

This is only from one LED manufacturer and is only a part of their application note. I have seen another that showed photographs of the actual sputtering of the bond wires.

Curt

 

[SemiMan]

INTRODUCTION
This application note describes electrical overstress (EOS) events, their effect on Cree XLamp LEDs and some simple methods of protecting XLamp LEDs against EOS. Electrical overstress is simply exposing an LED to any current greater than the maximum cur*rent specified in that LED's data sheet. The number or length of EOS events is irrelevant because any single EOS event can cause damage to the LED. This dam*age can be exhibited either in an immediate failure or a failure many hours after the EOS event.

2. Transient over-current events Transient over-current events are events that subject the LED to current that is higher than the maximum rated cur*rent on the LED data sheet, either directly through high current or indirectly through high voltage. These events are transient, meaning they happen for a short period of time – typically less than one second. These events are sometimes referred to as "spikes," as in "current spike" or "voltage spike." If the over-current event occurs immediately when the LED is turned on or plugged into a energized power supply (also called "hot plugging"), this over-current event is called "in-rush current."

EFFECTS OF ELECTRICAL OVERSTRESS ON XLAMP LEDS
It is impossible to predict the failure mode of every LED exposed to electrical overstress, but Cree has seen two common symptoms of XLamp LEDs that have had an EOS event cause a catastrophic LED failure. Damage to Bond Wires One common failure mode from EOS is damage to the bond wires inside the LED package, as illustrated in Figure 1 below. This damage usually occurs as a burned wire or a broken wire. In addition, the EOS event can cause damage to other materials in close proximity to the bond wires, such as the encapsulant or phosphor. Damage Near Bond Pads Another common failure mode from EOS is damage to the LED chip itself near the bond pads, as shown in Figure 2 below.

As described in the Causes of Electrical Overstress section, one common form of over-current event occurs when LEDs are connected to an energized power supply or when the power supply is first turned on. This event is called in-rush current.

Cree strongly recommends adding some level of protection to LED modules that do not include an on-board power supply to minimize the risk of an in-rush current event from the separate power supply.

This is only from one LED manufacturer and is only a part of their application note. I have seen another that showed photographs of the actual sputtering of the bond wires.

Curt

Big difference between pushing the part beyond its rated maximum current and simply hitting it with a fast high current pulse as a PWM signal would do. PWM on its own is not an issue at all. Most flashlight circuits with PWM are battery resistor combinations that pretty much limit the current to below the maximum the LED is rated for.

That said, with very short pulses, the LED can likely take well over its rated current without damage and for long long periods of time. Burning off a bond wire is more a result of average current (over a short period of time), not a short over current event unless it is way over the rated current. Burning off a bond wire is no different from blowing a fuse for all intents and purposes.


Semiman

 

[yifu]

PWM is efficient at the driver while linear regulation is efficient at the LED. If you want the most efficient driving method, just use a 1 ohm resistor in the circuit, like Electrolumens does it.

 

[wquiles]

PWM is efficient at the driver while linear regulation is efficient at the LED. If you want the most efficient driving method, just use a 1 ohm resistor in the circuit, like Electrolumens does it.

If you have 1 amp going through that 1 ohm resistor that is 1 watt of wasted energy, so no, that is "not" the most efficient driving method. If you were driving a 3 watt LED at 1Amp, that would be close to 25% of your battery power "lost" as heat. Definitely not efficient at all, and the higher the current the larger is the amount of energy wasted in that resistor.

The most efficient driving method is direct drive from battery to LED, which (assuming normal wires, nothing too thin) right at 100% efficient. Of course, you can only direct drive where the LED and cell(s) are matched, otherwise you run the risk of killing the LED, shortening its useful life, etc..

Will

 

[MikeAusC]

If you have 1 amp going through that 1 ohm resistor that is 1 watt of wasted energy, so no, that is "not" the most efficient driving method. If you were driving a 3 watt LED at 1Amp, that would be close to 25% of your battery power "lost" as heat. Definitely not efficient at all, . . . .

So true - no-one would bother advertising a driver if it delivered 75% efficiency !

 

[yifu]

If you have 1 amp going through that 1 ohm resistor that is 1 watt of wasted energy, so no, that is "not" the most efficient driving method. If you were driving a 3 watt LED at 1Amp, that would be close to 25% of your battery power "lost" as heat. Definitely not efficient at all, and the higher the current the larger is the amount of energy wasted in that resistor.

The most efficient driving method is direct drive from battery to LED, which (assuming normal wires, nothing too thin) right at 100% efficient. Of course, you can only direct drive where the LED and cell(s) are matched, otherwise you run the risk of killing the LED, shortening its useful life, etc..

Will

With real direct drive, you gotta really watch out for the Vf of the LED if not it delivers way too much current and kill the LED. Edit, i believe he uses a 0.01 ohm resistor, so there's very little power loss from I^2 R.