"Perfect" LED current limiting challenge

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
Post Reply
ian
Posts: 251
Joined: Mon Feb 25, 2002 1:01 am
Location: toronto
Contact:

Re: "Perfect" LED current limiting challenge

Post by ian » Thu Dec 22, 2005 4:12 pm

Thanks for re-iterating Chris, but I understood you the 1st, and 2nd time. A 3rd wasn't necessary. However, if you're going ignore my points, I'm going to ignore yours.

Besides the problems with your suggestions are blatantly obvious to any designer who knows anything about batteries and LEDS.

<small>[ December 22, 2005, 04:14 PM: Message edited by: ian ]</small>

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: "Perfect" LED current limiting challenge

Post by MrAl » Thu Dec 22, 2005 5:06 pm

Hi there ian,

If you're looking for max efficiency you'll
need to use an actual dc-dc converter circuit
unless the voltage input is close to the voltage
output in which case you may get away with an
LDO regulator.
Pulse width modulation, by itself, does not
help efficiency at all. It has to be part of
a circuit that contains at least one energy storage
element and the circuit has to be arranged correctly.

Many people have argued time and time again that
PWM through a smaller value resistor saves
energy somehow, but they never include the math
to back it up because the math shows that there
is no gain in efficiency, and in fact the eff
is the same as when using a single series
resistor to set the LED current.
There is the possibility however that you only
want to PWM to keep the brighness of the LED
at a constant level, which is fine, but dont
expect eff gains because you're pulsing.

One additional point:
Current regulation based on the HFE of a
transistor is a bad idea due to the variations
in HFE over temperature and from device to
device. I guess if you're only building one
and the ambient wont vary that much maybe you'd
be ok, but i'd use a resistor in series with
the LED, unless the input voltage can fall too
far, and then i'd use a simple current regulator.
If you really do need the eff, then you're stuck
using some sort of dc-dc converter that actually
*converts* power (has probably an inductor).
From the sound of it, you'll need a buck
converter because you're LED voltage is less than
the power source.
If you are driving a 3.5v LED from 6 volts, you'll
note that the battery isnt always 6 volts but
rather has some average value like 5 volts so
if you use a regular current regulator you'll
lose some power but not as much as if the power
source was always 6 volts.
If you're drving from 9v then you want a buck
circuit.

In any case, using PWM wont save you anything
at all over a linear regulator because as mentioned
before PWM by itself does not increase efficiency.
This means you can forget about pulsing anything
unless you also intend to include an inductor in
a circuit arranged as an actual dc-dc power
converter if you expect to get eff up.

There are quite a few ic's out there that do
buck conversion with somewhat small inductors
so if you're after high eff that's your best bet.

Another trick is to drive the LED with a single
series resistor and cut the current in order
to reduce battery drain. This reduces brightness
to some degree, so get an LED with a high mcd
output to compensate. If you cant stand the
variation in output over battery life, use an
LDO current regulator.
Using a linear circuit is going to be much easier
and straight forward over a true switcher, so
buy a high output LED and lower the target current
to half what you originally planned, if possible.

I think i understand you correctly, but if not
feel free to explain...i didnt notice this thread
right away so i had to quickly read through
the five pages of arguments :)
Also, i cant help wonder what you're using this
LED for...maybe it's back in one of the other
posts?


Take care,
Al
LEDs vs Bulbs, LEDs are winning.

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Re: "Perfect" LED current limiting challenge

Post by Chris Smith » Thu Dec 22, 2005 6:28 pm

Ian, you fail to match my 94% ......

[I can achieve 80% without the help I've received here so far] {your words}

......which leads me back to believe you think your 80% is better than 94%?

And you have a victory? Not likely?

This leads us back to the “Braille" post and the little bumps that you cant seem to climb? And your post fails to actually explain much of any of your paramaters, leaving us with guessing.

4 D cells can be replace by ANY better configuration of batteries unless your dyslexic, despite your [Predjudice] blind objection to the contrary.

Been there, done that,... while you were still in diapers.

So why don’t you stop pretending about miracles that cant happen,... and listen, follow, and learn from those who have been there and done that.

Your sounding like a broken record, and we all know what that means .....Your "being off track" not us.

You also fail to mention what “efficiency?” means in your example.

[light output, current draw, percieved brightness]

The human Eye can only see so many pulses per second, and then it becomes a steady stream of information. It appears to be solid?

If you pulse that light at the right speed and reduce its current by the division # [and more] your are using with your time base, the light will be mostly off and have heavier pulses [more current] and thus be more efficient, brighter, ....while maintaining a same appearance or more from the LED.

You can dump 100 amps into any led [from 4 D cells] , for a few nano seconds, use 1% of the over all power and on time, and make that LED blind you at a glance.

But again, you have actually provided nothing more than a “mystic” approach for others to solve, and I suggest your encryption comes from youth?

You haven’t been there or done that, but as you ask "in encryption", you hope to find out at our expense.

<small>[ December 22, 2005, 06:52 PM: Message edited by: Chris Smith ]</small>

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: "Perfect" LED current limiting challenge

Post by MrAl » Sat Dec 24, 2005 10:42 am

Hi there Chris,


I'd really like to see some more info on
that 1% duty cycle or whatever, and 100 amps.
Sounds really interesting. Are there any web
sites around i can take a look at some circuits
or something? I know i can use a mosfet or two
to drive it, but i guess it has to turn on and
off at really carefully controlled intervals.

BTW, at 100 amps whats the forward voltage of
a typical small white LED? I've got graphs
that go up to 100 milliamps, but nothing
near as high as 100 amps.
For higher powered LED's ive got graphs that go
up to maybe 1 amp or so (because they are made
for that) but im not talking about these kinds
of LED's and i dont think you are either.


Take care,
Al
LEDs vs Bulbs, LEDs are winning.

Mike6158
Posts: 409
Joined: Fri Nov 12, 2004 1:01 am
Location: Weimar, Texas
Contact:

Re: "Perfect" LED current limiting challenge

Post by Mike6158 » Sat Dec 24, 2005 11:29 am

"If the nucleus of a sodium atom were the size of a golf ball, the outermost electrons would lie 2 miles away. Atoms, like galaxies, are cathedrals of cavernous space. Matter is energy."

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Re: "Perfect" LED current limiting challenge

Post by Chris Smith » Sat Dec 24, 2005 1:41 pm

100 amps for 1/100 of one second is equal to 1 amp flow per second. Divided by another 100 and the current flow is equal to 1/100 of an amp per second. Divide that by 200 nano seconds on time and your current flow is ...well you get the point. Some where in the .00000000 ma range.

YES, this has been done since the 60s so finding "fast switching LEDS” can be found.

The forward voltage of the led also is irrelevant [to a point] to the stored cap holding 100 plus volts or more for the same reasons. The HV cap is the only way to discharge such a current package in the time requirements and this cap can be anywhere from 100 to 300 volts or more.

Current and voltage take time to flow especially through a transistor, and when you switch them fast, Real fast, what you get is a tiny package of current flowing through the led, turning the surface of the led Die to plasma, cooling back down and doing it all over again,.... and all so fast the led doesn’t even notice it.

You do this about 1000 times per second, and your OFF time still out numbers the ON time buy a few million to one. Nothing new. You just need to learn the realm of fast and super fast switching and the physics that come with it.

One of your biggest enemies is called “ringing” which is the reverse current and voltage literally ringing in your short wire leads like a bell, should you fail to dampen them by minimizing the lead length or inserting a back flow diode of appropriate speed into the you final design.

The other enemy is called “slope” where the transistor ramps up its current during its turn on time, and this can kill any LED in a nanosecond or two as they say. You need transistors that have ramp speeds in the range of a few NS before fully becoming on. Slow to turn on transistors will simply over heat your Led and it will burn out. Sharp ON and Sharp OFF times are essential. Led frequencies can be in the GHZ range for Communications which shows that the standard parameters described by the manufacture are for “general purpose only”. All of the ratings on all of the Semi conductors are for “general purpose” only and Im not familiar with any IC device that cant outpace its specs by huge factors if its done right.

rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: "Perfect" LED current limiting challenge

Post by rshayes » Sun Dec 25, 2005 4:44 pm

Hello MrAl,

You can get a rough extrapolation of LED characteristics from some of the data sheet curves. Take for example the Stanley UW5806X data sheet, which describes a 5mm white LED.

The forward voltage vs. Forward Current curve is plotted with semi-log axes. An ideal diode would be a straight line when plotted this way. The actual curve is far from a straight line. This is probably due to series resistance in the substrate and contacts. This series resistance will have practically no effect at low currents.

If the first decade of the graph is assumed to be the diode characteristic, the voltage appears to increase at a rate of about 150 millivolts/decade. The voltage due to the diode at 20 milliamps will be about 3.0 volts. The total voltage is actually 3.7 volts. The additional .7 voltage drop would indicate that the series resistance is about 35 ohms.

The total power into the LED at 20 milliamps will be 74 milliwatts.

If the current is increased to 1 amp, the diode voltage can be 3.25 volts and the voltage acrosss the series resistance will be 35 volts, for a total terminal voltage 38.25 volts. The power will be 38.25 watts, so the duty cycle will have to be less than .2 percent.

If you assume that the light output is proportiona to the current (actually it drops slightly at high currents), the average light output of the diode will be about 1/10 of the unpulsed light output at 20 milliamps drive.

For a 10 amp pulse, the voltage increases to 353.4 volts and the power becomes 3534 watts. The duty cycle drops to .002 percent and the average light output drops to about 1/100 of the 20 milliamp level.

A 100 amp pulse would require 3504 volts of drive and a power level of 350 kilowatts. The duty cycle would be .00002 percent, and the average light level would be 1/1000 of the original value. Actually, since the leads are less than .1 inches apart, the result is more likely to be an arc between the leads of the package.

If you need a pulsed light source, pulsing these diodes at 1 amp and a very low duty cycle might be reasonable. An example of this might be a rangefinder where a narrow, fast pulse will give better resolution. Pulsing these diodes at 10 or 100 amps is obviously total bullshit.

As you noted, the best average power an efficiency occur with continuous operation.

Mike6158
Posts: 409
Joined: Fri Nov 12, 2004 1:01 am
Location: Weimar, Texas
Contact:

Re: "Perfect" LED current limiting challenge

Post by Mike6158 » Sun Dec 25, 2005 5:05 pm

Originally posted by stephen:
Hello MrAl,

You can get a rough extrapolation of LED characteristics from some of the data sheet curves. Take for example the Stanley UW5806X data sheet, which describes a 5mm white LED.

The forward voltage vs. Forward Current curve is plotted with semi-log axes. An ideal diode would be a straight line when plotted this way. The actual curve is far from a straight line. This is probably due to series resistance in the substrate and contacts. This series resistance will have practically no effect at low currents.

If the first decade of the graph is assumed to be the diode characteristic, the voltage appears to increase at a rate of about 150 millivolts/decade. The voltage due to the diode at 20 milliamps will be about 3.0 volts. The total voltage is actually 3.7 volts. The additional .7 voltage drop would indicate that the series resistance is about 35 ohms.

The total power into the LED at 20 milliamps will be 74 milliwatts.

If the current is increased to 1 amp, the diode voltage can be 3.25 volts and the voltage acrosss the series resistance will be 35 volts, for a total terminal voltage 38.25 volts. The power will be 38.25 watts, so the duty cycle will have to be less than .2 percent.

If you assume that the light output is proportiona to the current (actually it drops slightly at high currents), the average light output of the diode will be about 1/10 of the unpulsed light output at 20 milliamps drive.

For a 10 amp pulse, the voltage increases to 353.4 volts and the power becomes 3534 watts. The duty cycle drops to .002 percent and the average light output drops to about 1/100 of the 20 milliamp level.

A 100 amp pulse would require 3504 volts of drive and a power level of 350 kilowatts. The duty cycle would be .00002 percent, and the average light level would be 1/1000 of the original value. Actually, since the leads are less than .1 inches apart, the result is more likely to be an arc between the leads of the package.

If you need a pulsed light source, pulsing these diodes at 1 amp and a very low duty cycle might be reasonable. An example of this might be a rangefinder where a narrow, fast pulse will give better resolution. Pulsing these diodes at 10 or 100 amps is obviously total bullshit.

As you noted, the best average power an efficiency occur with continuous operation.
:eek:

<img src="http://images12.fotki.com/v253/photos/4 ... orn-vi.gif" alt=" - " />

lot of vitriole in this thread... now I know what the E in WWE stands for... electronics

<small>[ December 25, 2005, 05:06 PM: Message edited by: NE5U ]</small>
"If the nucleus of a sodium atom were the size of a golf ball, the outermost electrons would lie 2 miles away. Atoms, like galaxies, are cathedrals of cavernous space. Matter is energy."

User avatar
Chris Smith
Posts: 4325
Joined: Tue Dec 04, 2001 1:01 am
Location: Bieber Ca.

Re: "Perfect" LED current limiting challenge

Post by Chris Smith » Sun Dec 25, 2005 5:27 pm

Page 416 of Rudolf Graf’s book shows a 10 amp 20 NS pulse delivered from a 15 volts source to a Led device [also known as a laser diode] can be achieved using a simple cap and FET combination.

The laser diode in Question can also only handle about 25 mils of DC current with out burning out, but when pulsed can handle currents of 1 through 100 Amps. The myth that the manufactures spec sheet is all they can handle lives on right up there with the tooth fairy.

Dream on, a saying from the 60s where it was already old news back then, but the youth just hasn’t woken up yet to these simple facts. They are still playing with their erector sets, Pics, and everything slow in the world of Semi conductors.

Your statistics are static, and don’t account for current flow down a solid conductor or discharge using a cap. They apply to power transformers and resistance there of.

I guess physics are really out of the realm here where Making candles part one and two, out of a complicated PIC seems to be the limit for the STUDENTS?

Voltages need not be in the thousands of volts to achieve a current flow of 100 amps, 10 amps or even 1 amp. A capacitor unlike voltage source of a transformer can transfer its current almost instantaneously where as a power supply from a conductor and coil can not.

There are so many factors here that appear to be next years [or next decade] project for so many that we will watch the youth crawl before they can walk.

No need to chastize the youth for thinking they can do it better, not having achieved some of the ground rules in science or electronics.

I have been laughing at the nay sayers on this score for better than two decades, and not having found one student worthy of a following, I can see it will be another decade or so before just one bright student figures out this very simple equation.

Its not even a Brain Fart moment that will solve this simple problem, and the words “Pulsing these diodes at 10 or 100 amps is obviously total bullshit” Is a reminder that mind constipation is the true blockage here.
Yes it can be done, its old news, and shame on you for your failing grade.

That’s right up there with “we didn’t really go to the moon” and all the other short minded self serving essays. The funny part is my data sheets VS your opinion, a private joke worth savoring. Remember to pull your foot out before conversing?

<small>[ December 25, 2005, 05:41 PM: Message edited by: Chris Smith ]</small>

Dimbulb
Posts: 324
Joined: Sun Oct 13, 2002 1:01 am

Re: "Perfect" LED current limiting challenge

Post by Dimbulb » Sun Dec 25, 2005 6:35 pm

The project may need to address two battery conditions:

1. Condition: The battery voltage IS adequate
lets assume We are satisfied with the
3.6 volts is 89% efficient [3 x 1.2 volt cells]

2. Condition: The battery voltage IS NOT adequate.
propose a scheme to make the batteries useable
right down to dead condition.

lets assume the operator can switch between the two
modes stated above.

Lets assume led draws 30 mA @ 3.6V
so we are on the same page.

The next logical question is :
How to effieciently get 3.6 volts out of the low batteries ?

what if 3 parallel LEDs having lower voltage Charecteristics were switched
in place of the 1 LED
to give 3 X 10mA @ 1.4V

<small>[ December 25, 2005, 07:10 PM: Message edited by: dimbulb ]</small>

JPKNHTP
Posts: 488
Joined: Wed Jun 29, 2005 1:01 am
Location: Missouri
Contact:

Re: "Perfect" LED current limiting challenge

Post by JPKNHTP » Sun Dec 25, 2005 10:17 pm

-JPKNHTP
-God Bless

rshayes
Posts: 1286
Joined: Tue Mar 04, 2003 1:01 am
Contact:

Re: "Perfect" LED current limiting challenge

Post by rshayes » Sun Dec 25, 2005 11:57 pm

An alkaline "D" cell appears to be capable of delivering about 7.7 ampere hours with a 50 milliamp drain (1.0 volt end point). There is some uncertainty since batteries are often measured with a fixed resistor load, which causes the current to vary as the cell is discharged. The average voltage will be about 1.25 volts. The total energy will be about 35,000 joules. Four cells would deliver 140,000 joules. If the LED uses 74 milliwatts, a 100 percent efficient driver should operate for about 525 hours. A 70 percent efficient driver should run for 367 hours or slightly over 2 weeks.

Three "AA" cells under the same conditions should last about 40 hours. This is not too far off the reference cited by JPKNHTP.

A linear constant current type of regulator with four "D" cells would run for about 150 hours. A switching driver, even if only 70 percent efficient, would provide over double the operating time of the linear constant current controller.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: "Perfect" LED current limiting challenge

Post by MrAl » Mon Dec 26, 2005 5:19 am

Hello again, mostly Stephen and Chris...


Stephen:
I'm certainly well aware of how to extrapolate
part specs, but i didnt think it was possible to
extrapolate a 20ma LED up to 100 amps :-)
One other thing i'd like to extrapolate is
the efficiency at 100 amps. The efficiency of
a regular 20ma white LED drops as the current
increases, so im wondering how little is left
by the time we get to 100 amps.


Chris:
This certainly sounds interesting.
So what you're saying is that if i can rig up
a circuit to pulse my little 20ma white LED
at 100 amps i'll get blindingly bright white
light out of a tiny little LED with very high
efficiency?
Some questions come to mind...say a Nichia
20ma white LED...

1. What is the max voltage across the LED while
the pulse is 100 amps?
2. What is the total light output (ratio to
light output at 20ma is fine)?
3. What is the efficiency of the LED at 100 amps
pulsed?
4. What is the color of the light output at
100 amps (assuming a white LED)?
5. Are there any products on the market today
that utilize this technology (flashlight, strobe,
etc)?

Very interesting stuff...


Thanks to both of you, and take care and
happy holidays,
Al
LEDs vs Bulbs, LEDs are winning.

ian
Posts: 251
Joined: Mon Feb 25, 2002 1:01 am
Location: toronto
Contact:

Re: "Perfect" LED current limiting challenge

Post by ian » Mon Dec 26, 2005 7:41 am

Guys, I've "discussed" topics with Chris before. Frankly he's a bit of a bully. Chris thinks that because of his "age" and "experience" what he says is gospel. In other threads Chris was actually supplying bogus information and calling that info "easter eggs" for me to find. How can I, or you, take him seriously?

In this topic, rather than help solve the origional problem stated, he has redefined the problem to suit his solution. In addition his solution doesn't work.

Chris does not take into account the varying voltage of the battery as it discharges, the varying forward drop of a LED depending on the type or lot, or the realistic instantaneous power capapability of a (typical)LED. Or the stated concern for not overdriving the LED if the micro fails and maximum current is applied to the LED. All of these parameters were/are defined in my origional question.

The origional question was about getting maximum efficiency of power from a battery to a LED. My specific application uses 4 D batteries and a white or red LED. But I didn't want to get into specifics. It should be possible to come up with a universal, simple way of efficiently driving ANY LED from ANY battery. In MY specific application I could not drive the LED with a DC to DC converter because there wasn't enough of a voltage differential after the batteries dropped in voltage.

In consideration of this I see 2 ways of which I am not that familiar........

1) Direct inductor
In this way the LED would be driven through an inductor and switched with a FET. When the curent reaches a reasonable maximum of say 30mA the FET would switch off and the inductor would discharge through a Schottky diode and the LED. It should be possible to use a large inductor since the current is only 30mA. But does 30mA have enough "oomph" from the inductor to push through the circuit losses?

Charge pump........
I'm not that sure at all how charge pumps work or their efficiency, I'll need to look into this.

<small>[ December 26, 2005, 08:35 AM: Message edited by: ian ]</small>

josmith
Posts: 340
Joined: Tue Dec 04, 2001 1:01 am
Contact:

Re: "Perfect" LED current limiting challenge

Post by josmith » Mon Dec 26, 2005 11:31 am

;)

<small>[ December 28, 2005, 08:53 AM: Message edited by: josmith ]</small>

Post Reply

Who is online

Users browsing this forum: No registered users and 31 guests