[Noisebridge-discuss] driving multiple LEDs with minimal batteries

Christoph Maier cm.hardware.software.elsewhere at gmail.com
Thu Jan 6 02:58:03 UTC 2011


Is the lumbrella using this component:
http://www.diodes.com/datasheets/ZXSC380.pdf
or a different one?

The ZXSC380 needs only one extra component, an inductor,
and you (Jonathan) explained how it works in one of the recent 5MoF talks.
(actually, it's a boost, not a SEPIC, but who cares?)

I just checked digikey; the part is in stock, ZXSC380FHCT-ND.
and there are plenty of SMD inductors to choose from
(the inductor needs to be picked with _some_ care, so that it can
store enough LI^2/2 energy per cycle,
but that shouldn't be a principal problem)
The remaining question is if the flexible PCB material has enough
resolution for a SOT23 footprint.
For the inductor, one of the larger SMD parts with a ferromagnetic
core looks like a good choice.
Apparently, Meredith's flex PCB material from Dupont is commercial grade,
so with the right printing or phototransfer process, this should be doable.

Just my 88 cents' (a piece) worth ...

Christoph

On Wed, Jan 5, 2011 at 5:04 PM, Jonathan Foote <jtfoote at ieee.org> wrote:
> Yes exactly. It's a current source: it delivers as much voltage as
> necessary to result in the desired current, (up to some reasonable
> limit which depends on the circuit but sounded like 30V in this case).
>
> At 20 mA, the power requirement for that is VI = 0.6 watts,  which is
> taxing but do-able for an alkaline AA (I'd use a few in a pack).
> Typically these switching circuits are ~90%  efficient so you can more
> or less ignore their power draw (another reason they rock).
>
> So the forward diode drop is the practical limit on this. For blue or
> white LEDs that's more than 3V typically, so you won't get more than
> 10 or so in series before you hit 30V. For reds Vf is closer to 1.7,
> so you could do a few more.



More information about the Noisebridge-discuss mailing list