z wrote:

Resistors heat == inefficient

But the resistor will always be there. You are just making a bigger
resistor, the current will drop and the light will dim in a vary
linear way.
The voltage you drop across your resistor will be the same no matter
how big it is. That is not like a rheostat on an incandescent where
you are changing the voltage applied todrop the filament.

The voltage across the resistor *does* change. Also, P=I^2R.

How does the voltage change across a fixed circuit? I think
everyone here is talking about an Xmas tree lighting situation
(now), in a series arrangement.

z wrote:

From some recent work, a blue LED at about 5mA drops about
3V. At 20mA the drop is closer to 3.3V. Now, put thirty of these in a
string and the difference is 10V. You only have 20V across the
resistor - it's changed 50%. ...and this is quite nonlinear.

The resistor is a current regulator not a voltage regulator.

AN LED DOES NOT HAVE A VOLTAGE DROP, the voltage can only be
measured with a resistor in series with the DIODE. This is not a
lamp, incandescent, or CFL. It's a ****ing DIODE.

A resistor regulates nothing. R==V/I. If you say that I is constant
because V is constant and V is constant because I is constant, you're
getting nowhere. ;-).

Bull****, you pick resistor sizes based on how much current you
need/want. The R is fixed, thus so is the voltage drop. Are you
talking about a simulated dimmer like he suggested?

"G. Morgan" wrote in message
news
z wrote:

Resistors heat == inefficient

But the resistor will always be there. You are just making a bigger
resistor, the current will drop and the light will dim in a vary
linear way.
The voltage you drop across your resistor will be the same no matter
how big it is. That is not like a rheostat on an incandescent where
you are changing the voltage applied todrop the filament.

The voltage across the resistor *does* change. Also, P=I^2R.

How does the voltage change across a fixed circuit? I think
everyone here is talking about an Xmas tree lighting situation
(now), in a series arrangement.


The circuit is not fixed. The conversation is that you change the resistor
value to change the brightness of the LEDs. There is very little change in
the voltage across the LEDs, but the big change is the voltage drop across
the series resistor as the value of it is changed to change the current that
changes the brightness of the LEDs.

wrote:
On Sat, 25 Dec 2010 22:36:45 -0600, z wrote:

I like doing experiments like this and I will be back as soon as my
high intensity LEDs arrive because this is my plan. We will see.
I have no problem admitting when I am wrong but I want to see it.
Go for it. I do this stuff all the time, though with indicators, not
for illumination.


OK Here you go

These are garden variety indicator LEDs (what I had handy)
There is a 270 ohm in series with 4 LEDs and a 1k pot wired as a
rheostat.
With the pot set to 0 ohms get a tad over 15M/A and the LEDs are as
bright as you can expect from these, may even be overdriven. I al
dropping 4.11v over the 270 ohm resistor.
Turning the pot down toward 1k ohms dims the LEDs quite smoothly
At 1K ohms plus the 270 I get 3.7 MA or so and I am dropping 4.8 volts
across the resistors.
http://gfretwell.com/electrical/LED%...experiment.jpg

I agree it is not exactly linear but I also do not understand how this
small variance makes any real world difference since most of the
voltage is still being dropped across the LED string. You certainly
would have a hard time seeing it on your electric bill.


There are a bunch of issues.
Yes, LED's are VERY dimmable.
The questions a
Can you build a reliable, affordable, LED lamp that works with the EXISTING
dimmer mounted on your wall.
Can you build an affordable dimmer that works with CFL, LED and
incandescent lamps. If you're buying a dimmer to dim a 20W LED,
will you pay for the extra cost to dim a 300W incandescent load.
Not much of a premium for old-school dimmers. For one that works
with LED, the premium can be significant. Vendors don't want
to warranty replace a dimmer you blew up by plugging in an incandescent.

And dimming is not the only problem. Lots of the X10 home automation
stuff uses SCR circuits that are not CFL or LED friendly. Ditto for motion
sensor lights...but that's for another thread.

Everybody reading this probably feels comfortable opening the box
on the wall and changing the dimmer. But Suzie homemaker may be faced
with this option.
A)Go buy a 57W incandescent for 25-cents on sale and screw it in.
B)Go buy a $40 LED lamp and a $40 dimmer and PAY somebody to install it.
The stuff doesn't exist (at reasonable prices) because the demand isn't
there. The demand isn't
there because the cost is too high because the demand isn't there
and one might argue that the technology isn't there either.

Basing your decision on the STATED lifetime numbers is ill-advised.
My budget can survive a failed 25-cent incandescent. I'm much more
sensitive to a failed $40 LED whatever the failure mode. I have NEVER
had a CFL fail because the bulb quit. I've had many fail because the
electronics exploded. Don't think I ever got 8000 hours out of one.

The dimmer on your wall doesn't work because it requires a load to charge
the cap that provides the phase shift that delays the turn-on and dims
the incandescent. Sure you can design one better, but
1) That one ain't on your wall
2) It will initially cost so much that you won't put one on your wall.

As for the LED lamps.
You've got a 160V peak or twice that depending on where you live
coming out of the wire.
Your LED is 3V or so.
It was mentioned that you can stack them up to 95% of the volts.
Not so. There are voltage variations between LEDs and temperature
variations. When you stack a bunch of them up, that variation
can figure significantly into the means you use to limit current.

Putting 40 of them in series drops the reliability by a factor of 40 or so.
Yes, I expect some statistician will take issue, but the factor ain't small.

If you use a resistor to limit the current with only small headroom,
the current becomes very voltage sensitive. You don't want your lights
to dim
every time the line voltage sags a volt.
And there's also the reliability issue.
You don't want your bulbs to burn out every time your air conditioner
compressor kicks off and spikes the line. And you don't have room or $$
for a line filter in each bulb.

Another way to limit current is to use a series capacitor. It's much more
efficient, but is critically dependent on having a sinewave input.
Line transients can pop your lamps. And the dimmer on your wall
puts a HUGE transient into the lamp every half cycle.
And the caps aren't cheap.

Then there are the MUCH MORE EXPENSIVE active ways to manage the current.

Unless you're designing a new lighting system, you don't have control
of both ends of the system. For currently installed systems,
Suzie homemaker is gonna go to home depot and buy whatever is on sale
and works with what's currently installed.

At today's prices, the only rational compromise is CFL and stick
with incandescent where you absolutely must have dimming.
YMMV

As an experiment, I did buy some of the lights-of-america 1.5W 20-led
lights for cheap at a garage sale. Make great night lights, but after
a year, they're getting dimmer. Wish I'd measured the initial light
output. And they waste over half of that 1.5W in the dropping resistor.
So much for efficiency.

On Fri, 24 Dec 2010 06:25:05 -0600, G Morgan wrote:
Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Standards light dimmers use PWM, not variable voltage. They should work
great with LEDs if not for the fact that LED's have a power
supply that converts AC power to low voltage DC and which will result
in the same low voltage DC based on the peak voltage in, the same for
1% as for 100%.

On Sun, 26 Dec 2010 01:17:12 -0600, G Morgan wrote:
wrote:

From some recent work, a blue LED at about 5mA drops about
3V. At 20mA the drop is closer to 3.3V. Now, put thirty of these in a
string and the difference is 10V. You only have 20V across the
resistor - it's changed 50%. ...and this is quite nonlinear.

The resistor is a current regulator not a voltage regulator.

AN LED DOES NOT HAVE A VOLTAGE DROP, the voltage can only be

bull****. LEDs have a voltage drop of 1.7V. Silicon diodes by the
way have a voltage drop of 0.7V.


Put a meter across a lit LED some time if you don't believe me.

On Sun, 26 Dec 2010 01:09:29 -0600, G. Morgan wrote:

wrote:

Resistors heat == inefficient

But the resistor will always be there. You are just making a bigger
resistor, the current will drop and the light will dim in a vary
linear way.
The voltage you drop across your resistor will be the same no matter
how big it is. That is not like a rheostat on an incandescent where
you are changing the voltage applied todrop the filament.

The voltage across the resistor *does* change. Also, P=I^2R.

How does the voltage change across a fixed circuit? I think
everyone here is talking about an Xmas tree lighting situation
(now), in a series arrangement.

He's talking about dimming LEDs, as in under-cabinet lighting.

" wrote in
:

On Sat, 25 Dec 2010 13:30:47 -0500, wrote:

On Fri, 24 Dec 2010 17:35:37 -0600, "
wrote:

On Fri, 24 Dec 2010 18:24:52 -0500, wrote:

On Fri, 24 Dec 2010 15:32:11 -0600, "
wrote:

On Fri, 24 Dec 2010 15:02:02 -0500, wrote:

On Fri, 24 Dec 2010 06:25:05 -0600, G. Morgan
wrote:

Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer
switch.

I can understand why CFL's can't be put on a dimmer - but why
not LED bulbs?

This is new, last 3 years. They had to develop a ballast that
was adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes
them even less desirable as a replacement for incandescent
bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.
Try 3. something volts to light a white LED. The only way to
"dim"
LEDs is to PWM them with variable pulse width/duty cycle. The
dimming range is quite narrow.

Not true at all. Using PWM, or a variable current, you can get a
very substantial dimming range (with less change in color than an
incandescent). It's just a PITA and a phase-control (Triac) wall
dimmer ain't going to do it.

The same can be done for "overdriving" an LED. Shourt duration
pulses can significantly increase the visible light output without
overheating the junction.

No, it doesn't increase the light output at all. You may be able
to see it with less output because a flashing light catches the
eye, but as long as it's a "constant" light output (i.e. not
visibly blinking) the light output of an LED is pretty much a
linear function of the *average* current through it. Flashing of an
LEDm above the eye's critical fusion frequency does not increase
efficiency, rather the opposite. The efficiency of an LED goes
down, at high currents, as it heats.
MANY high output led applications are pulsed "overdrive"
applications, and believe me, they DO put out a LOT more light.

No, they don't, for any reasonable reading of that sentence. The
physics doesn't allow it (the opposite, in fact).

Driving
them steady at those currents would blow them in a matter of
minutes, but pulsed at 15-20% duty cycle at up to 4 or 5 times rated
current they still deliver almost rated lifespan, and, if I remember
correctly,over 5 times the rated light output.

The *average* current is all that matters. The average also does the
heating, so it's a no win to pulse them, other than it's the easier
way to dim them.


As usual you are not "completely right". I won't argue and say you
are wrong in your assertions - but my UNDERSTANDING is that PEAK
current controlls the visible light output, and average current
affects lifespan (due to junction heating). It is not totally linear.

Nope. Current controls the light output. Average current controls
the average light (which the eye detects, integrated over the
"critical fusion frequency"). Yes, peak current controls the peak
light output, if you're detecting peaks, this might be important. It
is certainly *not* if you're looking at it. Above the CFF, human eyes
average the light intensity. Having high peaks with long spaces does
*nothing* to aid perception and in fact reduces efficiency; LEDs are
LESS efficient at high currents. Pulsing LEDs is a lose-lose
proposition.

My experience is obviously different than yours.
As in many other cases, I need to say that just because you haven't
seen it, don't make it wrong or impossible. Just means your scope is
too narrow.

No, you're just wrong. It's not the first time.

See: http://www.gardasoft.com/uploads/APP...ing%20LEDs.pdf

Did you actually *READ* that app note? An overdrive factor of *6*
will produce only 3 times the light (efficiency drops by half).

"The average current must be kept below the current rating for the
LED."

IOW, you can't overdrive it for long.

The table "High Power LEDs" indicates that you can drive the LED up to
5x current for 2ms, with a 10% duty cycle. A 5X current you get 2.5X
the light or ONE HALF the average light output as you would have
gotten if you'd just driven it at 100%, DC. IOW, a loser.

also http://www.lunaraccents.com/educational-LED-driver.html
and http://www.light-speed-tech.com/ltleds.htm
and
http://www.optoiq.com/index/machine-...sing/display/v
sd-article-display/351674/articles/vision-systems-design/daily-product/
intelligent-strobe-driver-safely-controls-led-intensity.html and
http://www.gardasoft.co.uk/ and
http://www.smartvisionlights.com/pro...erdrive-series and
http://www.freepatentsonline.com/7639219.html and
http://www.parameter.se/products/Def...30&ID3=142&ID4
=155

Are just a very few references for you to look at (commercial
applications of) pulsed overdrive applications for high luminence LED
applications.

As I've shown with the first article, pulsing LEDs is a loser. You're
simply *WRONG*.


the high power Cree XR-E LEDs I used for my homemade bike light are spec'd
at 228 lumens at 1 amp,but emit half that(~114 lumens) at only .35amps.
So,they are more efficient at the lower current.

--
Jim Yanik
jyanik
at
localnet
dot com

On Sun, 26 Dec 2010 02:10:31 -0500, wrote:

On Sun, 26 Dec 2010 00:09:54 -0600, z wrote:

On Sun, 26 Dec 2010 00:52:51 -0500, wrote:

On Sat, 25 Dec 2010 22:36:45 -0600, z wrote:



Lets get to the bottom line. Are you trying to say a dimmed LED (by
increasing the size of the resistor) draws more current than one at
full brightness?

Of course not. I'm saying that it is NOT LINEAR and NOT EFFICIENT.


I guess the point is LEDs are so efficient to start with and dimming
them will cut the draw even more, why make things hard on yourself to
squeeze out an extra few pennies a year.

I could say the same thing about replacing incandescents in the first
place. Why, for a few mennies a year. In fact, that's what I do.

The difference between an LED and an incandescent is far more than
pennies and when you dim the LED you are still reducing the power you
use. Garden variety triac dimmers are not that efficient either but
we still say they save money.

Sure they're efficient. They have a fairly well fixed 1.4V across them, times
the duty cycle of the brightness setting. At half-power that's well less than
a watt. Your rheostat will easily produce that much heat.

Have you ever looked at the lumens per
applied power number on a dimmed incandescent? They are horrible.
Bear in mind they are still drawing power when they are virtually too
dim to see.


The application I am making right now will be low voltage so my little
experiment is really a prototype. I will just need to make the
necessary adjustments for the high intensity LEDs I bought.

On Sun, 26 Dec 2010 01:54:30 -0500, wrote:

On Sat, 25 Dec 2010 22:36:45 -0600, z wrote:

I like doing experiments like this and I will be back as soon as my
high intensity LEDs arrive because this is my plan. We will see.
I have no problem admitting when I am wrong but I want to see it.

Go for it. I do this stuff all the time, though with indicators, not
for illumination.


OK Here you go

These are garden variety indicator LEDs (what I had handy)
There is a 270 ohm in series with 4 LEDs and a 1k pot wired as a
rheostat.
With the pot set to 0 ohms get a tad over 15M/A and the LEDs are as
bright as you can expect from these, may even be overdriven. I al
dropping 4.11v over the 270 ohm resistor.
Turning the pot down toward 1k ohms dims the LEDs quite smoothly
At 1K ohms plus the 270 I get 3.7 MA or so and I am dropping 4.8 volts
across the resistors.
http://gfretwell.com/electrical/LED%...experiment.jpg

Now, scale that to your 120V application.

I agree it is not exactly linear but I also do not understand how this
small variance makes any real world difference since most of the
voltage is still being dropped across the LED string. You certainly
would have a hard time seeing it on your electric bill.

It will *not* be a small variance when you add another 30 or so in series and
a smaller ballast resistor. Add in the sine wave excitation and you're going
to find your results don't match your experiment very well.

z wrote in :

On Sat, 25 Dec 2010 22:29:03 -0500, wrote:

On Sat, 25 Dec 2010 20:04:00 -0600, "
wrote:

Try it. You'll find that you are *very* wrong. LEDs are *not* in
any way linear. You'll also find that the efficiency goes down as
you lower the brightness (the resistor takes more of the line
voltage).

Why do you think the voltage across the resistor changes?

Because it does?

E = IxR
Ohm's Law.
it's simple math.
if you provide less current to the LED's because the resistor is a higher
value,the voltage dropped across the resistor INCREASES.(and the power it
dissipates) the resistor converts the additional power to heat.

You have a
certain voltage dropped across each junction and the resistor takes
the rest.


The resistor is a current regulator not a voltage regulator.





--
Jim Yanik
jyanik
at
localnet
dot com

On Sun, 26 Dec 2010 01:17:12 -0600, G. Morgan wrote:

wrote:

From some recent work, a blue LED at about 5mA drops about
3V. At 20mA the drop is closer to 3.3V. Now, put thirty of these in a
string and the difference is 10V. You only have 20V across the
resistor - it's changed 50%. ...and this is quite nonlinear.

The resistor is a current regulator not a voltage regulator.

AN LED DOES NOT HAVE A VOLTAGE DROP, the voltage can only be
measured with a resistor in series with the DIODE. This is not a
lamp, incandescent, or CFL. It's a ****ing DIODE.

A resistor regulates nothing. R==V/I. If you say that I is constant
because V is constant and V is constant because I is constant, you're
getting nowhere. ;-).

Bull****, you pick resistor sizes based on how much current you
need/want. The R is fixed, thus so is the voltage drop. Are you
talking about a simulated dimmer like he suggested?

Perhaps you should read the thread. This time for comprehension.

On Sun, 26 Dec 2010 08:48:30 -0500, Art Todesco wrote:

The string itself is a rectifier. The current through them will be DC,
no reason to further rectify it.. In the case of a string of xmas
lights they have nearly the full line voltage across them. Works fine.

Jimmie
Yes the string (I assume your are talking about a Christmas string)
is a rectifier, but it is only a 1/2 wave rectifier. Adding a full
wave rectifier, as others (and me too) have done, makes the string
brighter and eliminates much of the annoying blinking. Yes it still
blinks, but full wave rectification makes it a whole lot better.

The manufacturers could put an anti-parallel LED in the same package with the
same effect. The power dissipation is the same (application dependent) so the
package wouldn't need to be changed, just another junction on the same die.

z wrote in news
On Sat, 25 Dec 2010 21:59:15 -0600, z wrote:

On Sat, 25 Dec 2010 22:29:03 -0500, wrote:

On Sat, 25 Dec 2010 20:04:00 -0600, "
wrote:

Try it. You'll find that you are *very* wrong. LEDs are *not* in
any way linear. You'll also find that the efficiency goes down as
you lower the brightness (the resistor takes more of the line
voltage).

Why do you think the voltage across the resistor changes?

Because it does?

You have a
certain voltage dropped across each junction and the resistor takes
the rest.

sorry, new netbook and keyboard/trackpad is driving me nuts

No, that would assume an ideal diode. They are not ideal, the I-V
curve is not a vertical line. The more diodes you put in series and
the lower the voltage across the resistor, the more pronounced this
becomes. From some recent work, a blue LED at about 5mA drops about
3V. At 20mA the drop is closer to 3.3V. Now, put thirty of these in a
string and the difference is 10V. You only have 20V across the
resistor - it's changed 50%. ...and this is quite nonlinear.

The resistor is a current regulator not a voltage regulator.

A resistor regulates nothing. R==V/I. If you say that I is constant
because V is constant and V is constant because I is constant, you're
getting nowhere. ;-).


a resistor acts as a current LIMITER,not a "regulator",and it stipulates a
constant source voltage.
Since line voltage typically varies,you need some sort of regulator
circuit(usually an IC)to keep LED current from exceeding it's limit and to
keep a constant brightness.

--
Jim Yanik
jyanik
at
localnet
dot com

z wrote in :

On Sat, 25 Dec 2010 23:25:40 -0500, wrote:

On Sat, 25 Dec 2010 22:15:14 -0600, z wrote:

On Sat, 25 Dec 2010 21:59:15 -0600, z wrote:

On Sat, 25 Dec 2010 22:29:03 -0500, wrote:

On Sat, 25 Dec 2010 20:04:00 -0600, "
wrote:

Try it. You'll find that you are *very* wrong. LEDs are *not* in
any way linear. You'll also find that the efficiency goes down as
you lower the brightness (the resistor takes more of the line
voltage).

Why do you think the voltage across the resistor changes?

Because it does?

You have a
certain voltage dropped across each junction and the resistor takes
the rest.

sorry, new netbook and keyboard/trackpad is driving me nuts

No, that would assume an ideal diode. They are not ideal, the I-V
curve is not a vertical line. The more diodes you put in series and
the lower the voltage across the resistor, the more pronounced this
becomes. From some recent work, a blue LED at about 5mA drops about
3V. At 20mA the drop is closer to 3.3V. Now, put thirty of these in a
string and the difference is 10V. You only have 20V across the
resistor - it's changed 50%. ...and this is quite nonlinear.

The resistor is a current regulator not a voltage regulator.

A resistor regulates nothing. R==V/I. If you say that I is constant
because V is constant and V is constant because I is constant, you're
getting nowhere. ;-).


Lets get to the bottom line. Are you trying to say a dimmed LED (by
increasing the size of the resistor) draws more current than one at
full brightness?

Of course not. I'm saying that it is NOT LINEAR and NOT EFFICIENT.

You may have a more efficient way of dimming a LED lamp but how many
decades would it take to get your money back over just using a
rheostat? Sometimes KISS is the best rule.

A reostat is a BAD idea. The worst, in fact. It's easy enough to
perform this electronically, but the results aren't good. Doing the
job right is somewhat more difficult.

I like doing experiments like this and I will be back as soon as my
high intensity LEDs arrive because this is my plan. We will see.
I have no problem admitting when I am wrong but I want to see it.

Go for it. I do this stuff all the time, though with indicators, not
for illumination.

I just did one where I used current sources instead of resistors to
eek out the last couple of tenths of a volt on the supply. It was an
application where we were using red and green LEDs but the owner like
blue but the product is LiIon powered so we have to operate from 3.6V
to 5V (when charging we run off the input). The current source
allowed a constant intensity over this range (3.6V is pushing it)
while having suffiient drive at the low end.


when I worked at Tektronix,I had a semiconductor CURVE TRACER that plotted
out the V-I curve on the CRT screen.
for a LED,it's not a sharp bend,it's more like a hockey stick,a "knee" and
then a sloping current increase as voltage increased. heating of the LED
chip causes the angle to change.
you could measure the V drop at the desired current,and not risk destroying
the LED by overcurrent or overheating.
LEDs can pop very fast when overdriven.(as the rheostat guy will discover)
for high power LED's,it will be an expensive lesson.
My 3Watt Cree XR-E's cost $5.50 apiece. (from DealExtreme)


--
Jim Yanik
jyanik
at
localnet
dot com

" wrote in
:

On Sun, 26 Dec 2010 01:09:29 -0600, G. Morgan
wrote:

wrote:

Resistors heat == inefficient

But the resistor will always be there. You are just making a bigger
resistor, the current will drop and the light will dim in a vary
linear way.
The voltage you drop across your resistor will be the same no matter
how big it is. That is not like a rheostat on an incandescent where
you are changing the voltage applied todrop the filament.

The voltage across the resistor *does* change. Also, P=I^2R.

How does the voltage change across a fixed circuit? I think
everyone here is talking about an Xmas tree lighting situation
(now), in a series arrangement.

He's talking about dimming LEDs, as in under-cabinet lighting.


those are not single LED's,they are LED arrays.
Probably parallel strings of LEDs.

--
Jim Yanik
jyanik
at
localnet
dot com

On Sun, 26 Dec 2010 11:29:15 -0600, Jim Yanik wrote:

" wrote in
:

On Sat, 25 Dec 2010 13:30:47 -0500, wrote:

On Fri, 24 Dec 2010 17:35:37 -0600, "
wrote:

On Fri, 24 Dec 2010 18:24:52 -0500, wrote:

On Fri, 24 Dec 2010 15:32:11 -0600, "
wrote:

On Fri, 24 Dec 2010 15:02:02 -0500, wrote:

On Fri, 24 Dec 2010 06:25:05 -0600, G. Morgan
wrote:

Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer
switch.

I can understand why CFL's can't be put on a dimmer - but why
not LED bulbs?

This is new, last 3 years. They had to develop a ballast that
was adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes
them even less desirable as a replacement for incandescent
bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.
Try 3. something volts to light a white LED. The only way to
"dim"
LEDs is to PWM them with variable pulse width/duty cycle. The
dimming range is quite narrow.

Not true at all. Using PWM, or a variable current, you can get a
very substantial dimming range (with less change in color than an
incandescent). It's just a PITA and a phase-control (Triac) wall
dimmer ain't going to do it.

The same can be done for "overdriving" an LED. Shourt duration
pulses can significantly increase the visible light output without
overheating the junction.

No, it doesn't increase the light output at all. You may be able
to see it with less output because a flashing light catches the
eye, but as long as it's a "constant" light output (i.e. not
visibly blinking) the light output of an LED is pretty much a
linear function of the *average* current through it. Flashing of an
LEDm above the eye's critical fusion frequency does not increase
efficiency, rather the opposite. The efficiency of an LED goes
down, at high currents, as it heats.
MANY high output led applications are pulsed "overdrive"
applications, and believe me, they DO put out a LOT more light.

No, they don't, for any reasonable reading of that sentence. The
physics doesn't allow it (the opposite, in fact).

Driving
them steady at those currents would blow them in a matter of
minutes, but pulsed at 15-20% duty cycle at up to 4 or 5 times rated
current they still deliver almost rated lifespan, and, if I remember
correctly,over 5 times the rated light output.

The *average* current is all that matters. The average also does the
heating, so it's a no win to pulse them, other than it's the easier
way to dim them.


As usual you are not "completely right". I won't argue and say you
are wrong in your assertions - but my UNDERSTANDING is that PEAK
current controlls the visible light output, and average current
affects lifespan (due to junction heating). It is not totally linear.

Nope. Current controls the light output. Average current controls
the average light (which the eye detects, integrated over the
"critical fusion frequency"). Yes, peak current controls the peak
light output, if you're detecting peaks, this might be important. It
is certainly *not* if you're looking at it. Above the CFF, human eyes
average the light intensity. Having high peaks with long spaces does
*nothing* to aid perception and in fact reduces efficiency; LEDs are
LESS efficient at high currents. Pulsing LEDs is a lose-lose
proposition.

My experience is obviously different than yours.
As in many other cases, I need to say that just because you haven't
seen it, don't make it wrong or impossible. Just means your scope is
too narrow.

No, you're just wrong. It's not the first time.

See: http://www.gardasoft.com/uploads/APP...ing%20LEDs.pdf

Did you actually *READ* that app note? An overdrive factor of *6*
will produce only 3 times the light (efficiency drops by half).

"The average current must be kept below the current rating for the
LED."

IOW, you can't overdrive it for long.

The table "High Power LEDs" indicates that you can drive the LED up to
5x current for 2ms, with a 10% duty cycle. A 5X current you get 2.5X
the light or ONE HALF the average light output as you would have
gotten if you'd just driven it at 100%, DC. IOW, a loser.

also http://www.lunaraccents.com/educational-LED-driver.html
and http://www.light-speed-tech.com/ltleds.htm
and
http://www.optoiq.com/index/machine-...sing/display/v
sd-article-display/351674/articles/vision-systems-design/daily-product/
intelligent-strobe-driver-safely-controls-led-intensity.html and
http://www.gardasoft.co.uk/ and
http://www.smartvisionlights.com/pro...erdrive-series and
http://www.freepatentsonline.com/7639219.html and
http://www.parameter.se/products/Def...30&ID3=142&ID4
=155

Are just a very few references for you to look at (commercial
applications of) pulsed overdrive applications for high luminence LED
applications.

As I've shown with the first article, pulsing LEDs is a loser. You're
simply *WRONG*.


the high power Cree XR-E LEDs I used for my homemade bike light are spec'd
at 228 lumens at 1 amp,but emit half that(~114 lumens) at only .35amps.
So,they are more efficient at the lower current.

Tell that to "free-lunch" Clare. The datasheet he linked had a 2:1 efficiency
reduction (4x current for only 2x lumens) rather than 30%, but that was by
overdriving them above their average current spec.

On Sun, 26 Dec 2010 11:15:58 -0600, AZ Nomad
wrote:

On Fri, 24 Dec 2010 06:25:05 -0600, G Morgan wrote:
Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Standards light dimmers use PWM, not variable voltage.

Actually, phase control, not PWM. Similar to, but not to be confused with.
With a nonlinear load, like an LED, it's quite different.

They should work
great with LEDs if not for the fact that LED's have a power
supply that converts AC power to low voltage DC and which will result
in the same low voltage DC based on the peak voltage in, the same for
1% as for 100%.

Completely clueless.

On Sun, 26 Dec 2010 12:02:15 -0600, Jim Yanik wrote:

" wrote in
:

On Sun, 26 Dec 2010 01:09:29 -0600, G. Morgan
wrote:

wrote:

Resistors heat == inefficient

But the resistor will always be there. You are just making a bigger
resistor, the current will drop and the light will dim in a vary
linear way.
The voltage you drop across your resistor will be the same no matter
how big it is. That is not like a rheostat on an incandescent where
you are changing the voltage applied todrop the filament.

The voltage across the resistor *does* change. Also, P=I^2R.

How does the voltage change across a fixed circuit? I think
everyone here is talking about an Xmas tree lighting situation
(now), in a series arrangement.

He's talking about dimming LEDs, as in under-cabinet lighting.


those are not single LED's,they are LED arrays.

He's building his own.

Probably parallel strings of LEDs.

*Series* strings. Parallel does nothing but cause more trouble. ;-)

On Sun, 26 Dec 2010 12:19:56 -0600, zzzzzzzzzz wrote:
On Sun, 26 Dec 2010 11:15:58 -0600, AZ Nomad
wrote:

On Fri, 24 Dec 2010 06:25:05 -0600, G Morgan wrote:
Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Standards light dimmers use PWM, not variable voltage.

Actually, phase control, not PWM. Similar to, but not to be confused with.
With a nonlinear load, like an LED, it's quite different.

Who said anything about phase control?

PWM is pulse width modulation, control of power by duty cycle.

PWM works just fine on non linear loads as its output isn't operated linearly.
The output of PWM is either on or off. What controls average power is the
percentage of the time that the device is on. LEDs do fabulously well with PWM
actually.

In article , Jeff Thies wrote:

Hey Don,

What do you know about what circuitry is in the commercial LED lamps
and dimming compatibility.

For those who don't know, Don Klipstein knows more about lighting
than anyone I know of:

http://members.misty.com/don/ledx.html

We'll see if he is around and interested.

Sadly, I have yet to look inside any retail-available LED lights over 2
watts, nor most 2 watts and under. I don't remember too well what the
packages that I looked at so far say.

Best I can say is, look at the fine print on the packages for
compatibility with dimmers.

The Philips ones at Home Depot *may* be a good bet. However, I did
little more than glance at their photometric and color specifications
after getting "sticker shock".

So far, I am seeing only here-and-there applications where LED "bulbs"
appear to me more appropriate than CFLs or incandescents, mostly for light
output near or less than that of a 25 watt incandescent and with a lot of
"on time" per day.

One more thing: Most white LED units claiming 100,000 hour life
expectancy, especially cheaper and non-major brand ones, significantly
fade in 4,000 to 50,000 hours. I generally recommend green or blue LEDs
for nightlights. Any LED "security lighting" that must be at least
basically white should be by or have LEDs made by *major* brands of
"lighting grade" LEDs, such as Philips/Philips-Lumileds, Cree, Nichia,
Osram/Sylvania, Citizen/Cecol, and the like.
Also, the usual "bullet shape" low power LEDs are unlikely to get past
10,000 hours before significantly fading if they are white unless they are
greatly underpowered. I know of one LED nightlight "bulb" that does make
use of underpowering to achieve true long life from cheap white low power
LEDs. That is the Feit 3-LED candelabra base one.

One more thing: For outdoor or basement nightlighting or security
lighting, I strongly recommend that any white lights (LED or otherwise)
be of cooler color (more bluish, higher color temperature). This is
because night vision is significant in such dimmer lighting environments,
even if colors and sharp outlines are visible. A spectrum richer in
mid-green to mid-blue favors greater illumination in this case.
--
- Don Klipstein )

On Sun, 26 Dec 2010 13:11:39 -0600, AZ Nomad
wrote:

On Sun, 26 Dec 2010 12:19:56 -0600, zzzzzzzzzz wrote:
On Sun, 26 Dec 2010 11:15:58 -0600, AZ Nomad
wrote:

On Fri, 24 Dec 2010 06:25:05 -0600, G Morgan wrote:
Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.


L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Standards light dimmers use PWM, not variable voltage.

Actually, phase control, not PWM. Similar to, but not to be confused with.
With a nonlinear load, like an LED, it's quite different.

Who said anything about phase control?

Dummy, common incandescent (triac/SCR) dimmers use PHASE CONTROL, not PWM.
There *is* a difference.

PWM is pulse width modulation, control of power by duty cycle.

No, it is PHASE CONTROL. Dimming is done by controlling the firing angle of a
triac or SCR. For an incandescent (resistive load) this works out to be
similar to PWM, but it is a *lot* different when you're talking about a load
as nonlinear as an LED. PWM is often used to dim LEDs, but from a DC supply.
PWM works just fine on non linear loads as its output isn't operated linearly.

PWM works, phase control doesn't work well.

The output of PWM is either on or off. What controls average power is the
percentage of the time that the device is on. LEDs do fabulously well with PWM
actually.

It would help if you had slight clue.

On Sun, 26 Dec 2010 16:02:42 -0500, wrote:

On Sun, 26 Dec 2010 12:41:26 -0600, "
wrote:

Probably parallel strings of LEDs.

*Series* strings. Parallel does nothing but cause more trouble. ;-)

OK the science comes fast around here
I took a flashlight apart.
It appears they use a 4.5v LED, 9 in parallel with NO resistor at all.
The internal resistance of 3 AAA cells seems to be the limiter

I've seen that too; a very poor design. There is nothing to current-share
across the LEDs. Counting on the internal resistance of a battery is really
****-poor.

With just the batteries in there the lights are pulling about 244ma
(they are fairly new batteries)
This is hurt your eyes bright.
I put my 1k pot in there and even all the way off I am dropping .05v,
current around 211ma.

I assume you mean, "all the way *ON*", as in "zero" ohms (the wiper resistance
is about 1/4ohm).

The slightest movement of the pot, only putting
a couple ohms in there rapidly starts dropping the current. Somewhere
around 900 ohms we are at 3ma, dropping 1.3v
and the light is "indicator bright".

1.3V/.003A = 433 ohms.

OK so back to the junk drawer for some smaller resistors
With 10 ohms in there it is dropping 0.8v 77ma

Ok.

the light is noticeably
less but still pretty bright. When I double that with 20 ohms the
light dims quite a bit, current drops to 48ma and voltage 0.94v

You can already see that it's nonlinear and you're using a DC source.

I guess somebody has to crack open a 120v LED bulb or just do an
experiment like this to see how they work but I know what I need to
know about a low voltage setup like I want to make. I am thinking a 25
or 50 ohm pot will do the deed for me.

I thought you wanted an AC powered light? If all you want is to dim a
low-voltage DC string, PWM is the way to go. It's pretty easy. There are
cheap chips to do this automagically.

In article , G. Morgan wrote:
Home Guy wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.

L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.
--
- Don Klipstein )

On Sun, 26 Dec 2010 15:07:26 -0500, wrote:

On Sun, 26 Dec 2010 11:32:25 -0600, "
wrote:

The difference between an LED and an incandescent is far more than
pennies and when you dim the LED you are still reducing the power you
use. Garden variety triac dimmers are not that efficient either but
we still say they save money.

Sure they're efficient. They have a fairly well fixed 1.4V across them, times
the duty cycle of the brightness setting. At half-power that's well less than
a watt. Your rheostat will easily produce that much heat.

I would really like to see the math on that.

I normalized current to an amp (100W). 1A * 1.4V * .5(duty cycle - dimmed by
half power) is .7W. Scale that percent, give or take, for your choice of
lights.

I'd have to know what circuit you're intending to use, but a ballast resistor
for LED lighting can easily get there. Dimming makes it worse.

If I reduce current by
50%, I am reducing power by 75% (the 2 in I2R)

Overall, you will waste more, percentage wise, at lower settings. The
rheostat will dissipate more (V^2/R).

In my little string I was producing a total of 0.18 watts at full
brightness (15ma) and that dropped to .036w at 3ma. Even if all of
that was all in the rheostat, so what? The reality is you are still
dropping most of the voltage on the LEDs and that is where the lions
share of the heat will be.,

You aren't lighting anything with your little string. Scale that up to
something useful.

Since we are talking LEDs, what do you know about a 10 mm -130,000 mcd
Intensly Bright Blue LED? That is one of the types I ordered but I am
not sure about the specs.

I can tell you that they'll be about 3.3-3.6V each at 30mA, or so. More at
higher current. It's a blue thing. ;-) If you give me a P/N I can probably
look them up.

On Sun, 26 Dec 2010 15:17:19 -0500, wrote:

On Sun, 26 Dec 2010 11:49:47 -0600, Jim Yanik
wrote:

LEDs can pop very fast when overdriven.(as the rheostat guy will discover)

Since the rheostat will be in series with the proper "full load"
ballast resistor, overdriving is not an issue. Current is going to be
dropping as a square of the change in resistance and so will the heat.
As long as the rheostat is rated as high as the ballast resistor,
where is the possibility of a problem?

High power rheostats aren't all that common, or cheap, these days.

I suppose the open question is, how big is the resistor in a
commercial lamp?

I haven't seen a lamp with a rheostat is decades.

In my case it doesn't matter at all since I will be driving mine from
a wall wart. I am tempted to just take a cheap 4.5v LED flashlight
apart (a couple bucks at H.D) . That is probably cheaper than I can
buy 9 LEDS and for my purposes, it provides plenty of light. I just
want to spread them out a bit. I even have a 4.5v wall wart.

If you're powering this thing with a DC wall wart, simply PWM the thing. For
less than a buck or two in parts, you're all done, and you can forget about
the rheostat and waste heat.

On 12/26/2010 05:15 PM, Don Klipstein wrote:
In , G. Morgan wrote:
Home wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.

L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.

but it hasn't been done with CFLs yet, so don't hold your breath waiting
for it.

nate

--
replace "roosters" with "cox" to reply.
http://members.cox.net/njnagel

In , wrote:
Significantly edited for space
On 12/24/10 17:35:37 -0600, " wrote:

SNIP back-and-forth leading to this

The *average* current is all that matters. The average also does the
heating, so it's a no win to pulse them, other than it's the easier way
to dim them.

As usual you are not "completely right". I won't argue and say you
are wrong in your assertions - but my UNDERSTANDING is that PEAK
current controlls the visible light output, and average current
affects lifespan (due to junction heating). It is not totally linear.

There is a major myth about peak rather than average light output of
LEDs determines how bright they appear to humans, even when pulsed rapidly
enough to appear continuously on.

I mention its origins and the truth in:

http://members.misty.com/don/ledp.html

My experience is obviously different than yours.
As in many other cases, I need to say that just because you haven't
seen it, don't make it wrong or impossible. Just means your scope is
too narrow.

See: http://www.gardasoft.com/uploads/APP...ing%20LEDs.pdf

Shows efficiency decreasing with overdrive.

Effectiveness of combining overdriving and pulsing is for machine vision
applications where strobing is suitable.

also http://www.lunaraccents.com/educational-LED-driver.html

Mentions usefulness of pulsing, but not for increasing visual luminous
efficacy.

and http://www.light-speed-tech.com/ltleds.htm

Mentions short pulse overdrive - apparently for strobing, single-pulse
or machine vision applications.

and http://www.optoiq.com/index/machine-...ng-processing/
display/vsd-article-display/351674/articles/vision-systems-design/
daily-product/intelligent-strobe-driver-safely-controls-led-intensity.html

Machine vision is in the name of the link.

and http://www.gardasoft.co.uk/

Mentions for machine vision.

and http://www.smartvisionlights.com/pro...erdrive-series

Stated to be for machine vision lighting.

and http://www.freepatentsonline.com/7639219.html

That is for a visibly strobing application - mentions 10 flashes per
second.

and http://www.parameter.se/products/Def...11&ID2=30&ID3=
142&ID4=155

For cameras, frame grabbers and machine vision.

minor snip from here
--
- Don Klipstein )

In article , Nate Nagel wrote:
On 12/26/2010 05:15 PM, Don Klipstein wrote:
In , G. Morgan wrote:
Home wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.

L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.

but it hasn't been done with CFLs yet, so don't hold your breath waiting
for it.

It has been done for CFLs. I have seen dimmable CFLs at Target for a
year or something like that already, maybe almost 2 years. They have
existed for much longer than that.
--
- Don Klipstein )

On Sun, 26 Dec 2010 21:31:57 +0000 (UTC), (Don Klipstein)
wrote:

One more thing: For outdoor or basement nightlighting or security
lighting, I strongly recommend that any white lights (LED or otherwise)
be of cooler color (more bluish, higher color temperature). This is
because night vision is significant in such dimmer lighting environments,
even if colors and sharp outlines are visible. A spectrum richer in
mid-green to mid-blue favors greater illumination in this case.

For indoor, basement, indoor security, and night-lights, I'd think you'd want
more of a red so you don't spoil your night vision. ...or maybe a white
photo-flash and cover *your* eyes, while the no-good gets blinded. ;-)

On 12/26/2010 05:38 PM, Don Klipstein wrote:
In , Nate Nagel wrote:
On 12/26/2010 05:15 PM, Don Klipstein wrote:
In , G. Morgan wrote:
Home wrote:


I'm seeing more LED lightbulbs turning up on store shelves.

I don't think I've seen one yet that is ok to use with a dimmer switch.

I can understand why CFL's can't be put on a dimmer - but why not LED
bulbs?

This is new, last 3 years. They had to develop a ballast that was
adjustable. CFL's and LEDS are not a good comparison.

They're crazy-priced as is. Not being able to dim them makes them even
less desirable as a replacement for incandescent bulbs.

L.E.D.S. Are going to difficult (impossible) to dim. Remember
they are DIODES that only need .7V to illuminate. AFAIK ---
LED's are not dim-able.

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.

but it hasn't been done with CFLs yet, so don't hold your breath waiting
for it.

It has been done for CFLs. I have seen dimmable CFLs at Target for a
year or something like that already, maybe almost 2 years. They have
existed for much longer than that.

I've tried all of the ones available - at least four different brands -
at my local stores. (at significant cost, I might add.) They ranged
from unacceptable (really weird colors when dimmed, made dimmer hum,
only dimmed to maybe 50% brightness) to didn't even come close to
working as advertised (e.g. instead of dimming they would flicker and
hum and eventually shut off - no perceptible difference between them and
regular non-dimmable CFLs) This with several year old standard, off the
shelf Lutron dimmers.

I've heard tell of one particular dimmable CFL that is supposed to be
used with a special dimmer, I forget who makes it, sylvania maybe?
haven't cared enough to order them, and they're not readily available in
stores. In any case, the combination would be significantly more
expensive, if it did work (ever wonder why the demo displays in the home
centers don't simply show a CFL on a dimmer so you can try them? If
someone had something like that I would consider taking the plunge) than
simply shoving an incandescent in the fixture and replacing as
necessary. Wouldn't save enough in electricity or lifespan to come
close to paying for itself.

nate

--
replace "roosters" with "cox" to reply.
http://members.cox.net/njnagel

On Sun, 26 Dec 2010 18:13:03 -0500, wrote:

On Sun, 26 Dec 2010 16:24:20 -0600, "
wrote:

On Sun, 26 Dec 2010 15:17:19 -0500, wrote:

On Sun, 26 Dec 2010 11:49:47 -0600, Jim Yanik
wrote:

LEDs can pop very fast when overdriven.(as the rheostat guy will discover)

Since the rheostat will be in series with the proper "full load"
ballast resistor, overdriving is not an issue. Current is going to be
dropping as a square of the change in resistance and so will the heat.
As long as the rheostat is rated as high as the ballast resistor,
where is the possibility of a problem?

High power rheostats aren't all that common, or cheap, these days.

I suppose the open question is, how big is the resistor in a
commercial lamp?

I haven't seen a lamp with a rheostat is decades.

... but you are talking about incandescents.

Other than toys, I haven't seen an LED lamp, either.

In my case it doesn't matter at all since I will be driving mine from
a wall wart. I am tempted to just take a cheap 4.5v LED flashlight
apart (a couple bucks at H.D) . That is probably cheaper than I can
buy 9 LEDS and for my purposes, it provides plenty of light. I just
want to spread them out a bit. I even have a 4.5v wall wart.

If you're powering this thing with a DC wall wart, simply PWM the thing. For
less than a buck or two in parts, you're all done, and you can forget about
the rheostat and waste heat.

You keep talking about this waste heat like we were dissipating
hundreds of watts. I only say rheostat to define a single ended pot.
If the ballast resistor is 1/2w or even 1/4w why would the pot need to
be any bigger?

You're not talking about illumination at these power levels.

In , zzzzzzz wrote:
On Sun, 26 Dec 2010 21:31:57 +0000 (UTC), (Don Klipstein)
wrote:

One more thing: For outdoor or basement nightlighting or security
lighting, I strongly recommend that any white lights (LED or otherwise)
be of cooler color (more bluish, higher color temperature). This is
because night vision is significant in such dimmer lighting environments,
even if colors and sharp outlines are visible. A spectrum richer in
mid-green to mid-blue favors greater illumination in this case.

For indoor, basement, indoor security, and night-lights, I'd think you'd
want more of a red so you don't spoil your night vision. ...or maybe a
white photo-flash and cover *your* eyes, while the no-good gets blinded.
;-)

I've tried all sorts of things for seeing around with little light.
My experience says to make use of night vision. A cool white LED can
illuminate a room to extent that I can walk around and see everything,
using a couple to a few percent as much light as if I used red light.
--
- Don Klipstein )

On Sun, 26 Dec 2010 23:56:01 +0000 (UTC), (Don Klipstein)
wrote:

In , zzzzzzz wrote:
On Sun, 26 Dec 2010 21:31:57 +0000 (UTC), (Don Klipstein)
wrote:

One more thing: For outdoor or basement nightlighting or security
lighting, I strongly recommend that any white lights (LED or otherwise)
be of cooler color (more bluish, higher color temperature). This is
because night vision is significant in such dimmer lighting environments,
even if colors and sharp outlines are visible. A spectrum richer in
mid-green to mid-blue favors greater illumination in this case.

For indoor, basement, indoor security, and night-lights, I'd think you'd
want more of a red so you don't spoil your night vision. ...or maybe a
white photo-flash and cover *your* eyes, while the no-good gets blinded.
;-)

I've tried all sorts of things for seeing around with little light.
My experience says to make use of night vision. A cool white LED can
illuminate a room to extent that I can walk around and see everything,
using a couple to a few percent as much light as if I used red light.

That's interesting and a bit counterintuitive. I can see fine around the
bedroom, at night, with only the backlight of my XM radio but it's quite
bright. I guess it's not enough to trigger the iris but still make use of the
higher sensitivity wavelengths. OTOH, for astronomical viewing (and
submarines one uses red lights.

(Don Klipstein) wrote:

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.

I was taught (in college) that once break-through voltage is
equaled, that's it, on or off. Now I have to learn WTF went wrong
in my imaginary circuit if that is true.

wrote:

On Sun, 26 Dec 2010 21:17:20 -0600, G. Morgan
wrote:

(Don Klipstein) wrote:

Brightness of LEDs is controlled by how much current flows through them
- same as with dimmable CFLs. All that is necessary is to make the
current control circuitry in the LED "bulbs" compatible with and making
use from the usual dimmers, similar story as with CFLs.

I was taught (in college) that once break-through voltage is
equaled, that's it, on or off. Now I have to learn WTF went wrong
in my imaginary circuit if that is true.

KRW is right that there is some difference in the voltage drop across
the junction as the current rises., It is not a big difference but it
is a difference.(1.79 - 1.96 in the ones I tested over a range of 3ma
to 15ma) It could just be heat doing it at the high end. I did notice
in my flashlight tests that have no internal esistor that a very small
series resistance starts dropping the current very fast so they may be
using this "wall" of a heated junction to regulate the current,
similar to a nichrome wire.

The only way to get an LED to work is put a resistor in series.
Are you sure the (bulb, lamp, LED) was not a 'package'? They sell
them both ways, people who want only the LED (specialized), and
people who just want it to work. The later is what I think is
causing this confusion.

On Dec 26, 2:54*am, wrote:
On Sat, 25 Dec 2010 22:36:45 -0600, wrote:
I like doing experiments like this and I will be back as soon as my
high intensity LEDs arrive because this is my plan. We will see.
I have no problem admitting when I am wrong but I want to see it.

Go for it. *I do this stuff all the time, though with indicators, not
for illumination. *

OK Here you go

These are garden variety indicator LEDs (what I had handy)
There is a 270 ohm in series with 4 LEDs and a 1k pot wired as a
rheostat.
With the pot set to 0 ohms get a tad over 15M/A and the LEDs are as
bright as you can expect from these, may even be overdriven. I al
dropping 4.11v over the 270 ohm resistor.
Turning the pot down toward 1k ohms dims the LEDs quite smoothly
At 1K ohms plus the 270 I get 3.7 MA or so and I am dropping 4.8 volts
across the resistors.http://gfretwell.com/electrical/LED%...experiment.jpg

I agree it is not exactly linear but I also do not understand how this
small variance makes any real world difference since most of the
voltage is still being dropped across the LED string. You certainly
would have a hard time seeing it on your electric bill.

You should see the PWM method. brightness is variable from next to
nothing to full bright. They also retain their color while white
lights turn yellow using your method.

Jimmie

On 12/26/2010 4:31 PM, Don Klipstein wrote:
In , Jeff Thies wrote:

Hey Don,

What do you know about what circuitry is in the commercial LED lamps
and dimming compatibility.

For those who don't know, Don Klipstein knows more about lighting
than anyone I know of:

http://members.misty.com/don/ledx.html

We'll see if he is around and interested.

Sadly, I have yet to look inside any retail-available LED lights over 2
watts, nor most 2 watts and under. I don't remember too well what the
packages that I looked at so far say.

Best I can say is, look at the fine print on the packages for
compatibility with dimmers.

The Philips ones at Home Depot *may* be a good bet. However, I did
little more than glance at their photometric and color specifications
after getting "sticker shock".

So far, I am seeing only here-and-there applications where LED "bulbs"
appear to me more appropriate than CFLs or incandescents, mostly for light
output near or less than that of a 25 watt incandescent and with a lot of
"on time" per day.

One more thing: Most white LED units claiming 100,000 hour life
expectancy, especially cheaper and non-major brand ones, significantly
fade in 4,000 to 50,000 hours. I generally recommend green or blue LEDs
for nightlights. Any LED "security lighting" that must be at least
basically white should be by or have LEDs made by *major* brands of
"lighting grade" LEDs, such as Philips/Philips-Lumileds, Cree, Nichia,
Osram/Sylvania, Citizen/Cecol, and the like.
Also, the usual "bullet shape" low power LEDs are unlikely to get past
10,000 hours before significantly fading if they are white unless they are
greatly underpowered.

Thanks for joining in. I hope it is more joy than aggravation. I take it
that the failure is with the phosphors and not with the junction
generating the UV? Interesting as most of us had been thinking of LEDs
as being forever. I have noticed that many of the new LED traffic lights
have sections out. I don't know whether that is a circuitry break or
whether an LED actually failed. In the environment they are in I would
think connection failure from thermal cycling. Either internal or external.

Jeff



I know of one LED nightlight "bulb" that does make
use of underpowering to achieve true long life from cheap white low power
LEDs. That is the Feit 3-LED candelabra base one.

One more thing: For outdoor or basement nightlighting or security
lighting, I strongly recommend that any white lights (LED or otherwise)
be of cooler color (more bluish, higher color temperature). This is
because night vision is significant in such dimmer lighting environments,
even if colors and sharp outlines are visible. A spectrum richer in
mid-green to mid-blue favors greater illumination in this case.

" wrote in
:

On Sun, 26 Dec 2010 12:02:15 -0600, Jim Yanik wrote:


Probably parallel strings of LEDs.

*Series* strings. Parallel does nothing but cause more trouble. ;-)


I meant several series strings -in parallel-.

curiously,some free Harbor Freight 9 LED flashlights I have have all 9
LED's in parallel,and no current limiting resistor,depending instead on the
internal resistance of the 3 AAA cells.

--
Jim Yanik
jyanik
at
localnet
dot com

" wrote in
:

On Sun, 26 Dec 2010 16:02:42 -0500, wrote:

On Sun, 26 Dec 2010 12:41:26 -0600, "
wrote:

Probably parallel strings of LEDs.

*Series* strings. Parallel does nothing but cause more trouble. ;-)

OK the science comes fast around here
I took a flashlight apart.
It appears they use a 4.5v LED, 9 in parallel with NO resistor at all.
The internal resistance of 3 AAA cells seems to be the limiter

I've seen that too; a very poor design. There is nothing to
current-share across the LEDs. Counting on the internal resistance of
a battery is really ****-poor.

hey,they were FREE,including the batteries. ;-)
HF gives them away free with a coupon from their many ads.

of course,they don't use the batteries efficiently.

--
Jim Yanik
jyanik
at
localnet
dot com