Wrapper idea - LED current reducer

[idlefaction]

Hey all.  as part of my ever-evolving effects wrapper, I'd like to build a very-low-current on/off LED driver.  It would involve an LED, current limiting to some high value, a switching component, and then a clock with high frequency, low duty cycle, and low current consumption.  and some way of turning it off.

i was thinking about using a BJT for the switch and driving it with a low-power 555 running at 100kHz or so.  I usually run LEDs at 1mA, i figure I can probably do it at 5% duty 5mA and get the same or brighter.

has anyone worked with anything like this and have any lessons already learned?  would the 555 be too noisy or anything?  :-)

[petemoore]

would the 555 be too noisy or anything?
 So 'they tell me. I haven't tried one on the same power rails as a signal circuit, but I think a separate supply might help.
 I don't realy know if noise [RF?] would get in the signal path because of proximity...

[Transmogrifox]

you would want 20% duty cycle at 5 mA peak current to have an average current of 1mA through the LED.

I think if you properly decouple the power supply and the 555 timer, RF noise in the signal path is probably not a problem.

100 kHz is a good frequency to choose because there aren't any broadcast radio stations near it, so multiplier circuits (ie distortion pedal) won't put any stray radio noise (like stuff picked up by your cables, and things on the common ground) into the audio range, and of course, anything about 20 kHz is definitely inaudible.   Guitar rigs are low-pass enough that even a mild 12kHz signal is hard to notice.

If RF noise stayed purely RF, it would be no problem in an audio system.  Unforturnately, guitar amps and such tend to demodulate RF signals into audio frequencies, then "RF noise" becomes audible noise.

This pulsed design won't reduce current consumption.  If you want the same amount of light out of the LED as what you have at 1 mA, you need to dissipate the same amount of power in the LED...so the lowest power/current application in this case is a simple resistor/LED and switch to Vcc.

If you want to decrease power consumption, get rid of the resistor and figure out how to regulate the current through the LED...but my guess is that the circuit to do that with would consume as much or more power than the resistor would unless you just so happened to have a voltage supply at 1.5 V or something...but you can use something that's already using current, like the circuit.  Just put in a switch that switches the LED in series with the circuit power supply. This will just reduce your headroom by a volt or so.

[Nasse]

http://www.elektor-electronics.co.uk/miniproj/frmipro.htm

scroll down april, low power led flash (pdf article)

Not exactly what you asked but nice version on subject, and ready made small pcb too

I once breadboarded some circuit from old elektor book, used one cmos ic and bs170 mosfet was controlling the led, very low rate flasher thing I planned to use on a lighthouse model for a friend who has a sailboat at sea. But it was too much components, more complicated circuits and biggish pcb (my own design so it´s my own fault)

[idlefaction]

hrm.  I'm pretty sure that there's some effect where if you pulse an LED at HF you can get the same brightness for much less average current.  when i've got one built, i shall report.  :-)

i guess the only way to see is to do it!  and a 20kHz LPF won't hurt anything.  thanks for all your input.

[R.G.]

Hey all. as part of my ever-evolving effects wrapper,

Great! That's the first time I've seen someone besides me use that, and I was beginning to think that I had simply shouted another idea off into the void with the "wrapper" concept.

I'd like to build a very-low-current on/off LED driver. It would involve an LED, current limiting to some high value, a switching component, and then a clock with high frequency, low duty cycle, and low current consumption. and some way of turning it off.

Sure - it's pretty simple. Use a single CMOS inverter package, set up two or three of the inverters as an oscillator, but use a diode and two timing resistors to "bias" the oscillator to have small on-time duty cycle. Then use the output to run a MOSFET like the BS170 or 2N7000. That gets you a good, fast, high current (1A) switch run for very low battery power. Put a small inductor in series with the LED so the LED current doesn't run to infinity instantly but rather ramps up, and a diode in anti-parallel with the LED to let the inductor ramp back down once the switch turns on.
Each cycle has the LED current ramp up to a high peak that is determined by Ipk=Vsupply*Ton/L where L is the inductance.

Nothing in there dissipates heat except the parasitics and the LED, to a first order.

[Paul Perry (Frostwave)]

If you make a pulsing indicator at say 3Hz, the fact that it is pulsing makes it much more visible & so you can use much less current. And if you are using a 555, use a cmos 7555 & that stops the nasty spike on the power line.

[idlefaction]

That's the first time I've seen someone besides me use that, and I was beginning to think that I had simply shouted another idea off into the void with the "wrapper" concept.

I'm a software developer so re-use is like second nature.  :-)  I didn't have a concrete name for it, I just thought of it as refactoring common functionality.

If you make a pulsing indicator at say 3Hz, the fact that it is pulsing makes it much more visible & so you can use much less current. And if you are using a 555, use a cmos 7555 & that stops the nasty spike on the power line.

ahhh, like bike lights.  so i guess the only reason you'd use this kind of thing normally is so you can push more current thru an LED than you'd be able to with DC without burning out the LED, bugger.  however, i was considering a 1M-1M voltage divider across B+ and a resistor-zener string on a comparator to sense when the battery went low, and use a MOSFET to switch in another timing capacitor for it; it would be pretty cool to be able to have something visual like this to tell me that my battery was near death!

yay.

[Transmogrifox]

If you want the same average current, there's no difference between pulsing the LED at a higher magnitude current, or just keeping it on at a lower magnitude.

Your goal is lower average current...and hopefully you would perceive the same brightness, though the total average optical power output is less.  

Your brain is low-pass, so are LED's.  There is a speed where you would simply perceive the average brightness instead of the peak brightness, so you wouldn't be accomplishing anything by using too high of a frequency.

Maybe try 100 Hz.  This way you aren't challenging either the frequency response of the LED or your eyes and brain.  The only downside to 100Hz is that one must be more careful about isolating the power supply.  Make a varable oscillator from 100 Hz to 100kHz and see where you notice the brightness begin to decrease.  That would be about the maximum frequency you should use.   I know of a company that sells LED display road work/ caution type signs and they dim their signs by strobing them at a higher frequency and I don't believe that they vary the PW (though it could be easily implemented this way).

[Nasse]

So you can make the "lighted area" bigger by multiplexing more than one leds faster than eye can see, one at time. Quess the rate has to be more than 16 times per second, 24 or more is better. Two 4017´s and some more components...

[Paul Perry (Frostwave)]

Ther are plenty of references to making leds brighter by pulsing them, but when yuo look closely it turns out that pulsing just allows a higher average current, by some non-linear cooling characteristics.
eg
". However you can increase the "apparent" brightness of LEDs for  use as an anchor light by pulsing them at a rapid rate, usually about 12 to > 16 pps. This is at or just above the flicker fusion rate for most normal adults. The bulbs can temporarily exceed maximum rated steady state light utput limits when pulsed because of the interpulse cooling. The perceptual  effect of the pulsing light is that it is much brighter than a steady state light of the same average output. It is also mildly disturbing to view - a  perfect attention getter. "
If you want a LARGER led for the same power, put it inside a half ping-pong ball  :?

[Transmogrifox]

Ther are plenty of references to making leds brighter by pulsing them, but when yuo look closely it turns out that pulsing just allows a higher average current, by some non-linear cooling characteristics.

Not to be anal, but the original post was not about merely getting maximum total brightness out of the LED, rather, to get the same perceived brightness with less average current; that is, to make a low power device that doesn't eat up your batteries so quickly, but still outputs the same perceived brighness.

The premise of idlefaction's theory about this is that an LED puts out the same brightness with 1 mA  going through it as it does for a moment when it is pulsed with a peak of 1 mA.  Assuming there is not low-pass tendency in the LED, and you exceed the flicker perception of the eye, one would think that a person would percieve the same brightness whether it was pulsed for a short duration (narrow duty cycle) at a high frequency, or if the same current was kept constant.  Obviously, when pulsed at a short duty cycle with the same magnitude current as the steady state case, the overall power consumption is less.

However, due to the charge storage capacity in an LED, there is a certain frequency where the LED fails to reach the desired peak brightness during the cycle, therefore, the current is averaged and the optical output is dim.

If it's pulsed too slowly, the pulsing is visible.

My conclusion is that it will take more than experimental design to actually reduce power consumption and still produced the same perceived brighness, especially considering the driver circuit dissipates a certain amount of energy in itself.[/b]

[Paul Perry (Frostwave)]

Not to be anal, but the original post was not about merely getting maximum total brightness out of the LED, rather, to get the same perceived brightness with less average current; that is, to make a low power device that doesn't eat up your batteries so quickly, but still outputs the same perceived brighness.

Absolutely, I'm just pointing out that if you want ot have an easily seen indicator using as little power as possible, then make the damn thing blink at 3Hz or so. How brigfht would cop car lights etc have to be if they were not flashing :roll:
I don't want people to think that circuits designed to givbe "brighter" leds are necessarily giving MORE EFFICIENT leds!

[Transmogrifox]

good point  :lol:

[mikeb]

Considering parts count, solder connections and complexity, why not just pay a little extra for super-super-super-bright LEDs?

Mike

[idlefaction]

that's pretty much where this started actually, good thinking ;-)

i use 3mm blue LEDs and the brightest one i've been able to find is 450mcd which is enough at 1mA, but dim.  plus, if it worked out that i could run these things at 100uA and get the same brightness, great!  i'm all for saving batteries.

that said while i was typing this i found a 3600mcd blue LED in the mouser catalogue so i guess that solves that one  

[mikeb]

This might be better for you...

http://cgi.ebay.com.au/ws/eBayISAPI.dll?ViewItem&category=26207&item=3811977626

I've bought a couple of lots of LEDs from these people before. They sell direct from Hong Kong, have a nice lady working the email contact counter. Note shipping is included -> ~ 21c / LED for 3000 mcd.

Mike

[idlefaction]

ok, that is so, so cool.  thankyou so much!!!  wow.  i should be looking on ebay more.

[Hal]

but if an LED flashes rapidly - you don't see it flashing, but you see it as dimmer than if it was on constantly....as far as i know....

[Paul Perry (Frostwave)]

Considering parts count, solder connections and complexity, why not just pay a little extra for super-super-super-bright LEDs?

Well, can't argue with that! And something else to consider is the angle that you want to view the led at, the distribution of light varies A LOT, so a led that you wnat to look 'down' on, will be different from the one you want ot be able to see from a low angle across the stage.