"Perfect" LED current limiting challenge

[Mike6158]

Originally posted by ROBERT REED:
Jimandy
Or that tooth you chipped and no power on earth can keep your tongue from rubbing over it!

Been doing that for a week now... I didn't chip it very bad. Just enough to keep jacking with it

<small>[ January 29, 2006, 04:32 PM: Message edited by: NE5U ]</small>

[jimandy]

Chewing gum on the chipped tooth helps -- except, of course, in the middle of the night when you inhale it into your bronchya.

<small>[ January 29, 2006, 05:18 PM: Message edited by: jimandy ]</small>

[Jarhead]

Ian,

Go check out two of my threads:

http://candlepowerforums.com/vb/showthr ... ?p=1223704

http://candlepowerforums.com/vb/showthread.php?t=42516

(yes, I am Jarhead and Newbie)

[MrAl]

Hi again Ian,

If you're driving a 3.2v LED with a 6.4v source
that's PWM'ed then if your circuit was 100%
efficient you would measure 10ma from the 6.4v
source when the LED had 20ma avg through it.

Eff=Pout/Pin

and if you're not seeing around 80% it could
mean there is something else wrong, such as the
drop across the FET when it's turned on, or
the FET rise time and fall time isnt fast
enough because the drive Z isnt low enough.

Are you sure the circuit is connected as per
my post showing the connections for a buck
regulator?

Yes, a fair amount of programming for the PIC
is required in order to set up a working buck
converter with feedback, but once done the
LED gets regulated very closely to whatever
you want it to be, even when the source voltage
drops down to 4.5 volts. You also need to
provide the PIC with feedback (such as the voltage
across the LED)
so the PIC can compensate for the changing
input source voltage (6.4v to 4.5v maybe).

Also, there's a working example on the web
somewhere of a PIC driving a Luxeon star 1 watt
LED using a buck configuration. If you're really
interested in seeing it i'll try to find it
again. Details include the program code to
regulate the brightness of the LED.


Take care,
Al

[MrAl]

Found the App note:

http://ww1.microchip.com/downloads/en/a ... 00874b.pdf


Take care,
Al

[ian]

FINAL NOTE

I've done a fair amount of testing using a simulated battery voltage of 4 'D' cells, varying the voltage from 6.4 to 5V using different inductors.

It is possible to achieve up to 80% efficiency with a well balanced circuit BUT..........

It's very hard to control, or "balance" the circuit to get ideal results. Even choosing the right inductor is a problem. It seems to be necessary to match the inductor to the frequency and duty cycle.

Even the smallest changes in frequency and duty cycle seem to have large effects on the overall current. When the voltage drops increasing the duty cycle to compensate for the current reduction also reduced efficiency.

Also, I used an "ideal" circuit with no overcurrent protection, adding something for that will also reduce efficiency.

The PIC micro I used drew about 3 mA so I needed at least that to break even, but again the characteristics seemed to work against me. It seems most of the power of a 'D' battery is in the 1.3 voltage range. There's significant current savings when the battery is new at 1.6V but that doesn't last long. The savings of the circuit cancels out when the battery is at 1.3V so the savings is only around half of the life of the battery.

I guess the bottom line is the concept is possible but not that great for just one white led.

In my case, using 4 dry cells to 1 white LED just didn't give me significant enough results.
More LEDs, higher power LEDs or other loads would definitely see an advantage if you want to figure it all out.

Thanks to all the brilliant minds who offered suggestions.

Ian

[MrAl]

Hi there again Ian,


There is also the possibility the efficiency
appears to change (i'm assuming you mean it
changes by quite a bit...not just 5 perceont?)
because of the measurement method. When dealing
with pulses, average current times average
voltage is not equal to power, in the way that
works with sine waves.
Perhaps you can post a few numbers here so we
can take a look and try to determine if it's
the measurement method that is making the
efficiency 'appear' to change, rather than it
actually changing by a whole lot.
You might be more happy with the circuit if it
turned out that the efficiency didnt really
change by as much as you thought it did.

The kinds of numbers i'd need to see here are
maybe three sets of:
input voltage
input current
output voltage
output current

These three sets would be with full batt in,
1/2 batt in, and low batt in, so that if you
want to know the results using 6v down to 4v
you would have the three input voltages set as:
6v, 5v, and 4v, while taking readings for
input current, output current, and output voltage.
Also, you would need to specify how you are
making the measurements (what kind of meters).


Take care,
Al

[ian]

MrAL

I wasn't so much concerned with the loss in efficiency as I was with the loss of control. I also doubt I'd get high marks for my measurement techniques. I just used an analog meter in series with the LED and in series with the power supply for total current supplied vs LED current.
Here are some measurements I made.........

E%_____uH______I(t)____I(led)____Duty
45%___8,000____100_____115________40/36
72%___8,000____3.5______5.7_______20/10
66%___8,000____37_______55________20/15
65%_____340____55_______82________20/15
70%_____340____4.7______7.5_______10/5
68%___8,000____5.2______8_________10/5
67%_____120____3.7______5.6_______10/5
68%___8,000____8.9_____13.6_______10/6
72%___8,000___28_______45_________10/7
74%___8,000____5.4______6.7_______10/7
71%___8,000____8.8_____10.5_______10/8
75%___8,000____7.4_____9.3________20/16
72%___8,000___20_______24.2_______20/18
82%___8,000___12.8_____20_________20/15

I think I used 3.0V as a forward drop for efficiency calculations, but like I said I was more concerned with trying to get a steady 20mA through the LED than I was with efficiency.

With these values you can see I needed a high frequency to maintain efficiency. Since the period was so small there was no large variables for the duty cycle % to accurately compensate or control the curent. I was using a PIC @ 8MGHZ using the PWM feature.

Again, even with 80% efficiency the gains weren't enough for the amount of control circuitry and programming I'd have to add to get even ideal power savings.

<small>[ February 03, 2006, 12:42 PM: Message edited by: ian ]</small>

[Chris Smith]

Lets see?

"I think I used 3.0V as a forward drop"

Two D cells [3 volts] no electronics involved.

Calculated resistance value = 0
Next standard 5% resistor = 1 ohm
Power dissipated in resistor = less than 1 watt
Efficiency = 99.99%

Still trying to dance?

Two D cells, 2500 ma divided by " was more concerned with trying to get a steady 20mA" comes to 125 hours running time.

I Still don’t see your problem?

Beat that, or continue to pretend to manipulate electricity MORE efficiently?

But then you never described what "more efficently" means in your world?

More light, brighter light, for what usage, longer lasting, less strain on the led, etc., etc., etc.

[ian]

Guys, please don't respond anymore to Chris' idiocy. We all see the problem with his last post. I'm sure Chris himself knows this wouldn't work and is just trying to provoke. Clue into that.

[MrAl]

Hello again Ian,

Oh ok, i think i see a few possible problems...

First, probably the biggest error in measuring
eff is assuming the LED always drops 3.0 volts.
This can easily lead to errors of 10 percent or
more in some efficiency calculations.
Normally, the voltage across the LED would be
measured also and the actual reading would be
used in any calculations.

Second, i think you'll see much better accuracy
using a digital type meter set to a scale close
to what you are measuring. Analog meters lead
to errors for a bunch of various reasons.
If you can get ahold of two inexpensive meters
you might be happier

Third, when the current is measured in a switching
regulator like this it's best to insert a small
non-inductive type resistor in series with the
load, measure the voltage across the resistor,
and use Ohm's Law to calculate the current.
The value could be 0.1 ohms to maybe 1 ohm,
which would make calculating the current from
the 'across' voltage very easy to do in your head.
For your circuit probably 1 ohm would be best,
then whatever voltage you read is also the current:
If you read 20mv then the current is 20ma, etc.
Be aware that meters (even digital ones) sometimes
switch rather high value resistors in series with
the load to measure current, especially on the
low current ranges (20ma, 2ma, 200ua, etc.)
This could lead to VERY high errors in the
calculated efficiency.

Fourth, i wonder what the series resistance
of your inductors are? i would guess 8,000uH
would have a pretty high dc R value...perhaps
you could measure this value using your ohm meter.

Fifth, a good operating frequency (if you can
set it) would probably be around 50 to 100kHz.
Small value inductors (uH) dont work well
at lower frequencies because the peak current
goes too high (leading again to eff issues) and
at higher frequencies other factors become a
pain in the neck (skin effect, lead lengths,
stray capacitance, inductor core losses, as
well as rise and fall times). Heck, at some
upper frequencies the transistor may no longer
be supplying a true 'pulse' to the circuit, but
rather a 'triangular' shaped wave which eats
up efficiency very fast (due to long rise and
fall times).

Hope this all helps some...


Take care,
Al

[Dimbulb]

The project should have been described as an
efficient Pic Stamp based PWM 20mA led controller.

I still wonder why you did not make clear what you really wanted to do and why it was kept secret.

For me it is frustrating when the scope of a project is not disclosed and good ideas deemed irrellavant. The thread was long has some good ideas but the reading is tedious to find them. It was wasteful to discount so many of the good ideas that could have lead toward improvement.

The high amperage comment was good humor something you can learn from.

[MrAl]

Hi there,

Sometimes you ask a question and dont include
all the details because you dont realize what
you are really asking until some people start
asking you questions about what it is exactly
that you are trying to do. I see this all the
time all over the web on various message boards.
Myself, i like as much info as possible too
because that way i can mold my reply much better.

Take care,
Al

[Gadgets]

Originally posted by dimbulb:
I still wonder why you did not make clear what you really wanted to do and why it was kept secret.

I have to agree, I am new to this forum, and have now read all 16 pages of this thread. Enjoying the popcorn served up along the way. It was interesting to see how Ian was rude in his way of turning down the multiple offerings from several people here, but at the same time not be able to resolve the problem himself. Mostly I find it interesting on how he has basically refused to tell what his project is. Makes me wonder what all you guys are actually helping Ian to achieve. It also amazes me that people continued to help, argue, call names, etc. over a topic asking for help to (basically) get long battery life, and not blow up an LED if the actual project design is flawed and could "run wild", would it not be smarter to spend more time on the over all operation and programming to ensure it wouldn'd go crazy an you in the first place? Had you (Ian) offered up your entire project idea, not with all the identifying details maybe, but the over all project, maybe someone would have come up with a better, potentially easier, way to get your end result with out having to focus on this one area of your project.
Either way I think this is the longest post I have read in a very long time, that never actually resolved the issue concretely one way or another.

Ian, I do hope you manage to get what you are looking for, for whatever it is you are building.
My thoughts not knowing what your project is, and the fact that somewhere in the mess of things you mentioned the use of Solar Panels tells me this may go outside, why not build your circuit to the 80% efficiency I think you have achieved, use 4 D cell rechargeable batteries, and attach a small solar panel to it, to re-charge your batteries. That's my only offering at this point because I truely think this topic has all but run out. It has been a interesting read though.

Good luck to you Ian, hope it wooks out for you.

[Chris Smith]

We have ALL missed Ian's Revision numbers, revision pages, and the guide to the cosmos,.... let alone his actual goals here, because they are still a secrete.

He still has to date, failed to name his actual goals, level his parameters, or name anything at all useful about what he wants out of this mystery,...... to the beginner electronics hobbyist,....... let alone the serious,...... but still thinks he is in a room full of open ideas and return concepts?

The lights are out Ian, and every one serious has gone home, long ago.

But even the blind think they are within a organized structure, if they just bounce off a few walls?

One must state their goals, before receiving the magic bullet. Fact of life.

That or click your heels twice, and say I wish.

Im still waiting, as most of you are with your chipped tooth or root canal exposed into the cold wind,..... for what it is that your actual goal is supposed to be.

Is it the mystical water into hydrogen mole, saving us all for the oil monster?

There has to be some fantastic goal for mankind in the end, for so much wasted hyper space time and ink to be wasted?

Even Mother Teresa awaits your esoteric goals?

Is it the Cure for cancer, end to war,.. What?

We wait your reply with baited breath,...really?

And we still love canucks, they are soo funny!