# [Noisebridge-discuss] A simpler circuit for ... [driving multiple LEDs with minimal batteries]

Dr. Jesus j at hug.gs
Tue Jan 18 23:53:59 PST 2011

```On Mon, Jan 17, 2011 at 11:19 AM, Tymm Twillman <tymmothy at gmail.com> wrote:
>
> On Jan 17, 2011, at 10:32 AM, Dr. Jesus wrote:
>
>> On Mon, Jan 17, 2011 at 9:59 AM, T <t at of.net> wrote:
>>>>
>>>>> Ooh! Another teaching moment. How Switching Power Supplies Work.
>>>>>
>>>>> Switching supplies are efficient because the transistor "switch" is
>>>>> either fully OFF, (in which case negligible current) or in saturation
>>>>> (fully ON) in which case there's negligible voltage across it. In
>>>>> either state the power consumption is tine (recall P=V*I), compared to
>>>>> linear mode which has both significant current and voltage drop.  The
>>>>> JFET in that current limiting circuit is precisely in linear mode, and
>>>>> it will dissipate precisely the same power as a resistor (minus a
>>>>> little for the feedback resistor).  In fact a popular use of JFETs are
>>>>> as voltage-variable resistors.
>>>
>>> Hmmm... I'm not sure that's not a switched mode JFET.
>>>
>>> But without more details I will defer to the guy with at @ieee.org
>>> e-mail address and his teaching moments.
>
> makes sense in the switched capacitor filter role... but need some way to integrate the output.
>
>>>
>>>
>>>>>> It's simpler and easier to use than a resistor (you don't even have to
>>>>>> calculate a value- you just get one that's got a lower millamp rating than
>>>>>> the target LED and make sure the battery voltage exceeds the sum of the
>>>>>> voltage in the LED string)
>>>>>
>>>>> It's not that much simpler seeing as how it's not a discrete component
>>>>> that I'm aware of. It's a handy circuit if your supply voltage or load
>>>>> is variable (though I would use a LM317 for any appreciable current,
>>>>> as Igs for JFETs is rarely better than 50mA). Otherwise, simpler and
>>>>> easier to pick a resistor using Ohm's Law.
>>>>
>>>> For all switching regulators, you need some element that can store
>>>> energy (a capacitor or, in these cases, an inductor), not just dissipate
>>>> it (like a resistor or a switch).
>>>
>>> All transistors have some inherent capacitance.  That's why the
>>> original zero-capacitor zero-inductor joule thief circuit works.
>>>
>>>
>>>> So Ohm's Law becomes a little more complex.
>>>>
>>>> If you don't want to use a boost converter, ...
>>>
>>> The OP was looking for a simple way to add lights to a scarf, IIRC,
>>> and didn't have much electronics experience and some trepidation at
>>> building anything complex.  I fear we have driven her away.  Or
>>> perhaps I misjudge...
>>
>> Which solution did you end up going with, Meredith?
>>
>> Power supplies being what they are, it occurs to me that it would be a
>> good idea to have a poster near the components showing the fundamental
>> linear power supply types and the simpler switching ones.  I'll draw
>> one up.  If anyone has spare switching ICs they'd like to donate, let
>> me know before Friday and I'll add them to the poster.
>>
>> It would also be nice to put the typical conversion efficiencies next
>> to each power supply type to help the reader select an appropriate
>> circuit, but a quick look through my handbooks doesn't give me any
>> useful data.  Is there a good reference that takes into account modern
>> battery chemistries?
>
> more input / output voltage dependent than battery chemistry from what i've seen.
>
> efficiencies vary a lot; reasonable switching supply will generally be between 75% and 95%, but even board layout can impact that a bit, and the higher efficiencies will generally go to synchronous switchers (which use switched mosfets instead of a diode, and are more expensive).    in some cases, like buck/boost regulators where the output voltage can cross over the input voltage as the battery voltage drops, you will have ranges where efficiency tanks (though these are also applications where a linear regulator by itself just can't do the job).
>
> Lower output voltages will have lower efficiencies in standard switchers; much of your loss will generally be the .5v from the schottky diode drop, and if you have a very low output voltage that can be a high percentage...
>
> In both switchers and linear regulators, efficiency will change as the input voltage changes.
>
>
> probably good to note too that you get switching noise with switchers, which in some designs is a bigger problem.
>
>
> i'd be happy to supply some ICs, though the packages won't make clear what kind of regulator a part is... lots of switchers & LDOs in similar packages.  it's the surrounding parts that give them away.

I threw up a poster with descriptions of the two linear regulators
that are easy to make with the parts at Noisebridge.  There isn't a
lot of space left on an 11x17 so I think only one switching regulator
type will fit in the remaining area on the poster.

If someone writes up some text and example circuits for a simple
switching regulator I'll add it to the poster.
```