Why not fuse the 12V line - that's what I wanted to do, but due to the
way I had done the board layout, it was simpler to fuse the ground
line. Besides, this is a fiberglass airplane, so it is unlikely to come
in contact with dangling ground wires. But your point is valid, and it
might be worth changing in the future.

I had not thought of the decoupling capacitor for the 7805. Again, a
good point.

I knew about the low power dissipation in the 0.69Ohm resistor. The
only reason I used a 3W resistor was because I wanted to use the same
board for both circuits. A smaller resistor in a big slot won't look as
pretty (I know, lame reason..)

Someone else brought up the point about switching regulators for
greater efficiency. Actually, the efficiency here is not all that bad.
The red-side transfers 50% power to the LED, and the green-side
transfers 80% power to the LED. Most switching regulators are around
80-90% efficient, so I concluded that switching regulators were not
worth the extra complexity and the possible RF interference to gain a
little bit of extra efficiency. In addition, most off-the-shelf
switching regulator chips put out constant voltages, so they would
still need a circuitry to convert to a constant current application (so
we are back to the same problem). Also, most off-the-shelf switching
regulators put out standard PC motherboard voltages (3.3V, 5.0V..), and
non-standard voltages would be costly components. I once tried to
design my own switching regulator, but I could not entirely isolate the
noise. I could hear a hiss on my FM radio.




Rob Turk wrote:
"Andrew Sarangan" wrote in message
oups.com...
Actually, those are current sensing resistors, not current limiting.
The voltage across the resistor is compared with a fixed reference
voltage, and it is the FET which limits the current. So, a higher
resistor does not necessarily mean lower current.


Ahh yes. I saw the current sensing resistors but I overlooked the resistor
values for the divider that creates the fixed reference. I'm still wondering
about a few things though, hope you can elaborate a bit.

- You have the fuse placed in the ground wire. However, the largest
'touchable' area on your PCB and heatsinks are all attached to the area
behind the fuse. A simple protruding screw or rivet would be enough to turn
the circuit into an unfused system. Why not fuse the 12V line?

- The LM7805 really likes having a small decoupling capacitor at it's output
to prevent oscillation (and even self-destrucion on cheap non-brand 7805's).
How about adding 100nF between the 5V output and ground?

- The current sense resistor in the 1400mA version will dissipate about (1.4
x 1.4 x 1.3) 2.6W, so a 3W resistor is OK. The 700mA version would only
dissipate (0.7 x 0.7 x 0.69) about 0.36 W. You could do with a smaller
version (0.5W) there??

Mind you, I'm not trying to be negative about your design, but just curious
about some of your design decisions. The Luxeon's are really nice gadgets
8-)

Rob