Please forgive me if this is total bull. Its late at night and I'm having trouble thinking clearing.
I found one of Jim's circuits, to sense current:
I'm wondering if it can be used to make an efficient LED driver... Here's what I'm thinking...
Instead of having the current load (LED) go to ground, it will go to the negative end of capacitor. The positive end of the capacitor will be connected to the Vcc (5v in my case).
The "logic out" will also be connected to the -end of the capacitor.
See diagram below.
What I'm hoping this would accomplish is to charge the capacitor until there is enough voltage across it to cause the LED to draw the desired current. Once that happens, the transistors would switch, and the CurSense output would go high, basically preventing the capacitor from charging any more. The capacitor would then continue to discharge through the LED and R1.
When the current through R1 becomes low enough, the transistors would switch again, causing CurSense to be low, giving the capacitor more room to charge. I feel like I might be missing something fundamental in why this won't work.
I have 3 different LEDs that I want to drive, each with there own version of this circuit. 1 of them has a typical VF of 3.3v, and I want to run it at around 50mA. Another has VF of 2.2v and I want to run it at around 25mA. The third is also VF=3.3v, but I want to run it at 15mA.
Am I thinking down the right path? Would this be more efficient than a simple constant-current sink?
- +V5 | o-----------------------------------------o | R1 | o-----------^v^v^v------o-----------o | | | | | | Q2 | | __|__+ C1 | | / LED \|/ ___ | ------|v // V / | \- | | |\ VV --- | | Q1 | \ | | | | | | | \ | | | | | v|-----------C--------o | | /| | | | | / | 2 < | | | | R > | | | | < | | o------^v^v^v--o-------o | | | R5 | | | o----------------------C------------o-----o 4 > R < R < > > | o----------------------o | ----- 0V ---
Thanks, Daniel.