On Mon, 18 Jan 2010 17:47:44 -0500, Charlie Smith
wrote:

John Fields wrote:
On Mon, 18 Jan 2010 15:30:55 -0500, "Charlie Smith"
wrote:


Thanks for the offer John and I am interested. Here is what I would like
your circuit to do:

At 100% brightness, make the current in the LED's be 18 mA.
At 75% brightness, make the current in the LED's 18 mA.
At 50% brightness, make the current in the LED's 18 mA.
And at 25% brightness, make the current in the LED's 18 mA.

---
If you want the peak current to remain the same but the brightness to
vary, then you're describing PWM, which isn't what I offered.

My circuit would be more like:


BRIGHTNESS CURRENT
% mA
------------+---------
100 20
75 15
50 10
25 5
0 0

Not _exactly_ like that though, because the brightness VS current curve
isn't totally linear, but it's a dimmer, for goodness' sake, so you
crank it until it's bright or dim enough for you and who cares about
lumens per degree???
---

If it can do this then I am absolutely interested. I agree, there is the
risk of noise in the circuit since nobody that I know has done this that I
am aware of. No data exists. I plan to build it and place my hand-held
within 6" of the board and see if it talks to me. If it does, then its time
for Plan B. If not, then I am not so interested if it turns 3 or 4 Watts
into heat. That's roughly the amount of energy one incandescent instrument
bulb would consume. Insignificant. The heat rejection is built into the
board to handle that much and more. But if you have a better way, I really
would seriously review it. I am not married to any one approach. I just
want the best compromise. Right now from my perspective, varying voltage
does not seem to be it.

---
What you seem to be missing is that the varying control
voltage/resistance is used to vary the current through the lamps and,
therefore, their brightness.

Here's a simple example to illustrate the concept: (View in Courier)


+14V-+-------+
| |
| [R1]
| |
| [LED]
| |
| C
[POT]--B NPN
| E
| |
GND--+-------+

With the pot's wiper at the GND end of the element, there'll be 0V on
the base of the transistor, so the transistor will be cut off and only
leakage current will flow through R1, the LED, and the c-e junction of
the transistor, keeping the LED off.

As the wiper is rotated more and more toward +14V, however, a point will
be reached where the voltage on the base will rise enough for current to
be injected into the base, causing the current through R1, the LED, and
the c-e junction to increase as well, eventually illuminating the LED.

Then, with a fixed supply, since the transistor is causing the current
through the LED to vary as the voltage on the base is varied, it's
acting like a voltage-variable resistor with the entire system, BTW,
dissipating no more and no less power than a PWM controlled dimmer.

The ball's in your court; what do you want to do?

JF

You are correct, I was missing your point regarding base current on the
transistor. I can wire this up here on the breadboard and get a feel
for how it works.

---
It won't work very well at all because it's common emitter, so its going
to have a very narrow dimming range, probably about 5 or 10 degrees.

But then, it's not supposed to, [work well] since it's just an example
designed to demonstrate voltage controlled resistance.
---

My main concern is that somewhere between 50 and 25%,
a part of the panel will switch itself off and part will stay on. And
what goes off will depend on manufacturing variation on the LEDs,
exactly how much voltage the regulator is putting out and the
temperature in the cockpit. The practical consideration for me has to
be how often will I be operating in that envelope. Maybe never. Maybe
all the time. I won't know until I build the thing and see how well it
works. You are also correct in that what I described is PWM. Be fair.
You did ask what did it bring to the table and what it brings is the
ability to dim an LED at constant current.

I really have no care at all as to how efficient the circuit is. I do
care about how much total power it pulls.

---
At full brightness with a 14V supply and 20 LEDs drawing 20mA each,
either system will draw 400 mA from the supply and dissipate 5.6 watts
plus whatever the quiescent circuit requirements happen to be.

With the lights dimmed to off, the power dissipated by either system
will be zero watts plus whatever the miniscule quiescent circuit
requirements happen to be, so the point is that since either system
dissipates the _same_ amount of power for equal dimming, that's a wash,
and the circuit topology will depend on other factors.
---

Either system will be
efficient enough for me; by a far cry. Let me look at this alternate
set-up here over the next day or so. Either system will have to work on
5-8 volts regulated power so thats just a matter of resistors. If the
alternator dies, I want the panel to work as long as possible and the
3080 chip will function until well below voltage the LEDs will work at
and the load the lighting is taking off the battery will be minimal.

Thanks for the suggestion.

---
You're welcome, but don't use that circuit to seriously evaluate
_anything_ since it won't be part of the final circuit.

I'll put something together either tomorrow or Wednesday, and I'll post
it here when it's done. In the meantime take a look at Jim's
circuit...

Not bad for an old guy! :-)


JF