"Perfect" LED current limiting challenge

[Carl Pugh]

If you are trying to design a circuit with the highest efficiency then a resistor (or transistor biased as a resistor) cannot be used.
An inductor is probably the best approach with a transistor duty cycle modulated to limit the current.
You could search the web for boost or buck convertors. The circuits used for the boost and buck converters have a lot in common with what you are trying to do.
Good Luck
Carl Pugh

[Joseph]

You will definitely need reactance as the power attenuator instead of resistance for efficiency. In this regard you have a few options.

Use a capacitor to send pulses to the LED and regulate the power through frequency change--lower for less power. The capacitor needs to be repetitively switched high and low on the end not connected to the LED and a diode needs go from the LED to ground to shunt the reverse voltage as the capacitor voltage is reset. It is just a charge pump.

Use an inductor as Stephen mentioned with PWM to control the average current. A reverse shunting diode can protect the LED.

Or use both to take better advantage of semi resonance to get more efficiency than either alone can easily provide. If you place an inductor in series with the capacitor, it will lower losses when the current to the LED is switched on. The frequency is still altered to adjust the current through the LED. Best Wishes.

[Dimbulb]

If a series resistor used as a current limiter in a simple design gave energy in the form of light then the led current could be reduced.
If a new battery design had current limiting charecteristics and voltage for leds then the scheme should be new and useful.

[Edd]

As all of the others had also inquired, I also was desirous of further precise information as to the end application of the LED. Initially, I might have surmised it to be some light reference source. Then as the tale unwound it seems to be merely used in an indicator function, with max consideration towards optimum conservation of its sourcing battery.

I think what I might throw at you for consideration is my K.I.S.S. circuit...and heavy on those S's....please. Let it be a criterion/benchmark for its performance vs utter simplicity comparison, should you proceed further towards the definitively more involved complexities of DC/DC converter utilization, u/p incorporation, adjunct light output detection and its introduction into a feedback controller loop for further PWM modulation of the sourced power fed on to the LED.

This is something that I built up waaaaaay back, circa '68, to fulfill a definite shortcoming being eternally confronted in common flashlights. That, of course, being its quite rapid, ever dimming performance; from a cluster of cells initial 1.5 VDC fresh threshold, on down to its 1.0 VDC level...and the resultant weak, yellow beam output down towards and below that level.
I retrofitted the flashlights stock lamp with a #49 lamp with its common voltage level of 2.0 VDC and its miniscule current requirement of only 60 ma. Yet, this still provided the close use, typical illumination level that I required, but with yet ever more stringent current demands than the typical ~ 500 ma conventional flashlight bulb behemoth. Then I could use a set of four cells placed within the confines of flashlight housings that were initially intended for the 6 VDC common ignition cell….. or later on… the actual availability of designed units that were actually made for accommodating 4 individual "D" cells. The 6 V battery with either its spring terminals or terminal stud configurations has now skyrocketed in its pricing (low demand).
For a final comparison of performance , had the conventional 5.95 VDC lamp at its .5A drain been used, the initial battery cluster's 30watt power capability would have been drained down to the 1 VDC per cell (yellow light onset threshold) with only a provision of ~3.5 hours of use and a final reserve/yet untappable, 20 watt power level being retained in the battery stack, at that point where the dimmed light output condition had onset. What a waste!
With the utilization of my depicted circuit inline, from the battery to the lamp, a constant power level was sent to the lamp all the way down to the point where the battery cluster had tapered to its 3 VDC level of output. Now, that battery pack gets bled fully all the way down, with only its final 1.5 watts of capacity left to be wasted !
Now I can pick up that flashlight and expect a constant white light performance for upwards of ~125 hours of use ! Problem solved.
Eventually, I even used a conventional stick style 5 "D" cell flashlight and incorporated the circuit housed within a top located dummy cell, its same size housing being made up from poly plumbing pipe.

I re-plotted its circuit for your referencing, wherein I merely scaled down the initial circuits designed power level from producing ~2.0 VDC @ 60 ma down to the more conservative 10-20-30 ma levels to be used for a LED....all in accordance to the heftiness of the particular LED(s) used, with the typical 20 ma medium power level being more of the norm.

WOW …. that LED will just hang in there at a uniform output. Should you not have access to an adjustable power supply , it would be permissible to use a string of typical 9 VDC NEDA batteries; start with an aged unit that might be down to that depleted 6 VDC level and then start series polarized / top connector clipping units in order to progress on up to the higher voltage levels.

Now, in a repeat evaluation, should a LED have been otherwise solely utilizing a series c/l resistor, in your testing, somewhere approaching the increased voltage threshold of ~12VDC, along with the increased ramping up of its brilliance, that LED's emitter die should have reached a hyper-bright level along with a then abrupt emissivity cessation and its transformation into an intense black "fly speck"at its positioning within the locus of its reflective parabolic housing .
However, using the circuit in line, it is immediately receptive of any voltage/change within the previously specified ranges, with the LED merely holding at its initially established bright level.
Accordingly, with no discrete capacitive elements incorporated, it is receptive to quick turn on-off or supply voltage level transitions.

The selection of the ultra-reliability qualified 2N2905's was merely due to stock and familiarity with it in NASA/Aerospace/Military design and its companion lower level 2N2907 and their NPN counterparts. Here, it is loafing at 50% or less of its BVcbo and current/power capabilities specs. The selection of another equivalent electrical spec'd industrial/commercial/consumer grade unit is certainly permissible.

In your further experimentations, you might still want to incorporate the inline circuit, if solely for protection of the LED from damage.


73's de Edd
[email protected] .........(Interstellar~~~~Warp~~~Speed)
[email protected].........(Firewalled*Spam*Cookies*Crumbs)

Hey !...I just got to thinking……….Do radioactive cats have 18 half-lives?

<small>[ December 18, 2005, 08:13 PM: Message edited by: Edd Whatley ]</small>

[jimandy]

The legacy of your circuit epiphany of 1968 has been endowed in the form of figure 2.25 of Horowitz & Hill "A of E" (2nd edition 1989, but first published in 1980). It is an optimal circuit for a constant current source "...whose current doesn't depend on supply voltage" (their words). Did you fellows perhaps collaborate?

<small>[ December 19, 2005, 09:04 AM: Message edited by: jimandy ]</small>

[ian]

Wonderful circuit, but you, like most of the others have missed the point entirely. Even you said......
"Then as the tale unwound it seems to be merely used in an indicator function, with max consideration towards optimum conservation of its sourcing battery."
Yet........ you provide a circuit with a > 50% loss of useable battery power. If I use a DC to DC controller the current from the battery would be less than HALF of what your circuit uses.
I just am not sure of the details of proper implementation.

I think the best way is to use an inductor in series with the LED, along with a reverse biased diode in parallel with the inductor and LED. Then switch the LED off and on at a high frequency letting the inductor discharge through the diode-LED combination. I'm just not sure the little 20mA current flow stored in the inductor has enough oomph to overcome the capacitance and losses of the circuit to be anywhere useful.
Conventional DC to DC converters switch at 1 AMP.

<small>[ December 19, 2005, 11:50 AM: Message edited by: ian ]</small>

[jimandy]

But then, of course, you've got additional power loss implicit in the need for the oscillator/driver circuit. Or has that been mentioned before?

[Chris Smith]

Put a end to this silly discussion and lower your battery voltage to the nearest cell level of the required voltage for the LED in question,[2.4v 3.0v 3.6v 4.5v 4.8v] and trim the current draw with a final low resistance resistor for your over all current draw.

If you can beat that with any active circuitry, you’re a magi?

[ian]

Actually I was thinking of using solar panels but..... this is not a silly discussion. Powering LEDs from a battery is an extremely common application. The batteries cannot always be chosen to suit the LED, and various LEDs have various voltage drops. I think it would be good to have a design for an efficient circuit.

What does amaze me is this very common application doesn't seem to have any applied solution, and what amazes me even more is the lack of undestanding and lack of better solutions from the very bright people who frequent this board.

<small>[ December 19, 2005, 06:16 PM: Message edited by: ian ]</small>

[jimandy]

ian -"what amazes me even more is the lack of ... better solutions from the very bright people who frequent this board."

And I'll bet there are some bright people in this forum that are amazed that you persist.

[philba]

Ian, what surpises me is how much of a rude jerk you are and yet people still try to help you. I'm sure there must be other forums around. Why don't you go find them? Don't let the door hit you on the way out.

Phil

[Chris Smith]

Name one better way,... other than choosing the closest voltage match to the LED,..... and then adjusting the current difference needed through a low value resistor,..... for minimal efficiency loss,...other than what I have suggested?

You cant because it is the ONLY way to minimize current loss through any system.

Any vibrator circuits, chokes, caps or coils will collectively drain far more current than a 3.6 volt battery and a ten ohm resistor in a 3.2 volt led. [adjust the resistor value as necessary]

You do the math and see if you truly are the Magi I mention?

Good luck. I don’t care what voltage LED you chose, a battery combo and low resistance matching resistor can not be beat for a minimum loss and steady current match.

Trying to re-invent the wheel using more parts than the original is silly and,... we call them "Rubes" for short. Especially when there is no gain to be had by this approach.

[Robert Reed]

Hey--I just got a brilliant idea for unbeatable efficiency--ELIMINATE the LED.

[ian]

Hey, I can take this crap from the cheap seats, I don't mind. You guys are the ones who can't come up with a solution so you mock my request. Have a ball! But here's what you fools don't get.....

In experimenting with this circuit I used a DC to DC converter. With 4 new D cells outputting 6.4 volts, and a white LED with a forward drop of 3.1 volts I was able to drive the LED with 20mA while drawing 12.5mA from the battery.
Thats about 80% efficient?

DC to DC converters use an inductor to store current flow, converting higher voltages at low current to lower voltages at high current. All I need is a more optimized design for use with LEDS and minimal differential voltages.

Just because you guys can't figure it out doesn't mean nobody can, or that it can't be done. In fact, it seems most of you don't know what I'm talking about, let alone be able to come up with a solution.

What I'm trying to accomplish is a very efficient battery circuit for a LED. If you aren't able to come up with a design solution, or don't understand the concept..........
Instead of suggesting I change my parameters, or rethink my design, or drop it..........
Why not just leave it to someone capable of solving the problem posted?

Oh, and don't let the door hit YOU on the way out.

[dyarker]

I just went back and reread from the beginning. You said you had a circuit (still some kind of secret) to efficiently drive an LED at 20mA. You said you wanted protection to limit LED to 30mA in case YOUR circuit had a problem. Well, unless you want a 2nd PIC and DC to DC supply set to 30mA, then you need a resistor or Edd's circuit. The voltage drop will NOT be large unless YOUR circuit fails. Which is what you asked for.

And efficiency depends on EVERYTHING that goes in, including cost of parts. For the cost of inductors, driver transistors, etc; you could buy several batteries. By the way, where does the supply voltage for the PIC come from?

Please door hit HIM on the way out.