On Sat, 25 Dec 2010 20:14:45 -0800 (PST), JIMMIE
wrote:

On Dec 25, 10:04*pm, "
wrote:
On Sat, 25 Dec 2010 11:53:35 -0500, wrote:
On Fri, 24 Dec 2010 19:12:39 -0600, "
wrote:

On Fri, 24 Dec 2010 17:06:06 -0800 (PST), Red wrote:

On Dec 24, 11:50*am, "
wrote:

Yes, you are correct, except that the LEDs have to have a "threshold"[*]
voltage to light at all. *If the dimmer is triggered when the AC (sine wave)
is too low the LEDs don't light. *The range of adjustment will be very small.
[*]as others have pointed out, this isn't an on/off thing but is highly
nonlinear

* I think the key is how the current source is constructed. Limiting
the power lost there is important.

Yes, this is a very expensive thing to do, compared to a *cheap* Triac dimmer.

Note:http://joby.com/store/gorillatorch/switchback

Joby makes several models of quality dimmable LED lights, but as you
said they are not cheap. *The one in the link will dim thru a range of
5 - 130 lumens, mostly linear and without flicker. *I have one and it
works great. *I have no idea of the circuitry they use to dim the LED
and I'm not going to take it apart to find out.
This response is just to let the doubters know that the technology to
dim LED's is real and available.

Sure dimming LEDs is possible, and actually not all that hard if you start
with DC (the unit you linked is battery powered). *Doing it from the AC line,
cheaply and efficiently, is more challenging. *It's easy to do sloppily, but
then there is no gain over an incandescent bulb.

I still do not get how you arrive at that conclusion. A string of LEDS
will draw 15 - 20MA at full brightness and if you increase the size of
the current limiting resistor the current will drop from there in a
very linear manner.

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).

If you use a current source it works, linearly, but is no more efficient
(think of the current source as a non-linear variable resistor).- Hide quoted text -

- Show quoted text -

Nothing linear about a diode, he should actually try ploting a curve
and see how much the resistance of a diode changes as you change the
current, even much more so on a series string of the things like a
lamp cluster.. Also no big deal to convert to DC to run an LED they
are after all diodes they will do it for you. Thats the really cheap
way out . Slightly more expesive and only arguably of better design is
using a match head size bridge rectifier, cost about as much as a
match too.

No, they are diodes, but it's not a good idea to use them as
rectifiers. The reverse breakdown of LEDs is pretty poor. As you
note, a bridge rectifier is cheap enough. The probelm with running
them on DC is storing the energy over the cycle (filtering the DC).
That isn't cheap or particularly efficient.