/diy/ - Do It Yourself

>>1216066
>I'll just see about finding a fixed ~3V voltage rail on the power bank
There probably won't be one that won't drain the cells unevenly, and you don't want that.
>how to make a buck or boost converter vary its duty cycle in response to the output voltage.
For LEDs you're probably interested in current, not voltage. It depends on your particular circuit, but here's an oversimplified illustration of negative feedback. The more current through the LEDs on the secondary side, the more voltage developed across R1 and therefore Q1's B-E junction. Q1 eventually starts to conduct enough to pull down the sawtooth oscillator and reduce the amplitude on Q2's gate, thus decreasing the duty cycle of Q2 and the transformer. The nice thing about current mode is that you can add or exchange LEDs into the series string without characterizing or recalibrating much, as long as it's still within the abilities of your electromagnetics and they all demand the same current. If your LEDs are to run on different amounts of current, they will probably need separate converters.
The same general principle applies for a boost or buck converter: convert current to voltage and use that to retard the oscillator somehow.

>>1216084
>Vcc: 2.0-5.5V
As a rule, inputs should never exceed Vcc, and Vcc should always stay within the specified limits. Don't try to feed it 5V signals on a 3.3V Vcc and you'll probably be fine.