Which uses more juice? Flashing or Static LED?

It’s that time of year again, when cyclists are breaking out our lights. Which got me to wondering this morning, will my battery last longer if I set my lights to flashing, or static?My first guess is flashing means the light is on for less time, therefor it uses less. But is extra power needed to make the light flash?

Your LED is flashing all the time even when it appears static. It just flashes so quickly that persistence of vision makes it appear to be continuous.

So, while I have no solid evidence; I’m still thinking slower flashing uses less juice than standard rapid-flashing-that-looks-static.

All the LED lights I use specify much longer run times for flashing compared to static.

It will depend on the circuit causing the flashing. But in general flashing will require much less current. A simple circuit with a battery, transistor, resistor, and capacitor can make an LED flash. With the circuit tuned right the battery can last for years. Without flashing, maybe a day.

That depends entirely on the circuit driving the LED. If it’s being powered by a battery, it will be on continuously - not flashing quickly at all.

When optical mice were new, Logitech’s packaging had a flashing LED.

A typical LED doesn’t have much resistance. If it was powered continuously straight off a battery it will overheat and die. The two most common solutions are to: A. put a resistor in series with it (some LEDs are packaged with one), which wastes power, or B. have it pulse off and on quickly. You don’t usually notice this unless you do something like wave it around in the dark. This is far less wasteful and so better designed battery powered products will usually go this way.

So, it might already be flashing to save power, and by adding a slower power cycling on top of that, if saves even more.

About 10 years ago I read a paper that discussed the efficient driving of LEDs. The conclusion of the paper was that it is more efficient for average lumens per watt to drive the LEDs at high current in pulses than to have a steady lower current drive of the LED.

That may be true for overall system efficiency, but LED efficiency generally falls off as current increases, like so.

One point of the paper was that the graphs like you show were made with steady currents. If you pulse the current in the LED quickly you get a different curve.

There is no “extra power needed to make it flash”.

The main issue is how bright the LED is in steady mode vs. flash mode. On most lights, it’s the same (i.e. the peak brightness during the flash cycle is the same brightness as the steady mode). So if the flashing consists of 50% on / 50% off, it would use half as much power. All other factors (e.g. whether the LED itself is more efficient at high current) have very little effect.

There’s some debate on whether flashing mode is safer. There’s little doubt that flashing lights attract more attention. But there are some who claim that it’s harder to gauge the distance to a flashing light, and also some say inattentive (sleepy, DUI, etc) drivers tend to be attracted by flashing lights and steer towards them.

Sure there is.
The control circuitry takes some non-zero amount of power, but it’s obviously a win to flash the LED rather than leave it on all the time.

There are two types of “flashing” being discussed here. One is the relatively low frequency flashing often done for bicycle and other lights, which is slow enough that your eyes can tell it is flashing, and the other is the much higher frequency flashing which is commonly done with alarm clocks and other LED devices, and is so fast that your eyes can’t detect it. In either case, the LED almost always draws significantly more power than the control circuit and the flashing is therefore much easier on the batteries.

Your eyes and brain don’t see the average light coming from the LEDs. Instead, they tend to see the peak value. The battery however gets drained based on the average current value going through the LED. So if you have one LED that is drawing 20 mA continuously and a second LED that is pulsed at 30 mA but is flashing on and off so half of the time it is on and half of the time it is off, the second LED will not only appear to be brighter, but it will drain less current from the battery in the long run (15 mA average compared to 20 mA average for the first LED).

It’s the same circuit that’s used to control the LED in the steady mode

My bike light lasts about 12-15 hours not flashing and well over a hundred when flashing.

I have experimented with this myself, and it is generally true, but not all LEDs behave the same way; some LEDs will have a significant degradation in efficiency even when pulsed, to the point where they will appear noticeably dimmer when pulsed (for example) at 40 mA peak vs. 20 mA continuous (20 mA average for both). Some LEDs also show a shift in color when pulsed due to a higher voltage drop (in an extreme case, I have found red LEDs that turn yellow or even greenish at moderately high currents, but not enough to overheat them, which will also shift the color, including pulsed current).

One of the easiest ways to save power is to simply use brighter LEDs and run them at reduced current; for example, I have used ultrabright LEDs as indicators in battery-powered circuits running on a few hundred microamps and they are still plenty bright enough (roughly as bright as a standard LED running at 10-20 mA, 100x more power). Of course, this increases cost, and the aforementioned graph shows that LEDs can degrade significantly in efficiency at low currents (pulsing them could solve that though; e.g. 10 mA at 10% duty cycle for 1 mA average, but a simple indicator can get away with it).

AFAIK, the reason e.g. automotive lights use PWM to control LED brightness is because LEDs shift color as you reduce the forward current/voltage. It’s at the cost of perceptible flicker, but I think the color of taillamps/turn signal lamps is either a NHTSA or FMVSS requirement so color-shift isn’t permissible.

I’ll have to look that up.

PWM is used for LEDs because it’s much more energy-efficient than a linear regulator. And because LEDs aren’t harmed by short pulses at full voltage - so there’s no need for DC/DC converter with a clean output (constant output voltage).

I don’t think the color shift is a common or noticeable issue. I just looked at an arbitrary data sheet and it doesn’t mention any dependency of wavelength on voltage. It does mention a temperature dependency, but it’s very small - on the order of 0.05nm/C. Even a 100-degree rise in temperature would only cause a 5nm shift, which would be imperceptible to the eye.