🚀 go-pugleaf

On Fri, 04 Oct 2002 17:19:23 -0700, Lizard Blizzard <NOSPAM@rsccd.org>
wrote:

>budgie wrote:
>
><Snip>
>
>> In the auto situation, the simplest mitigation is to feed the
>> resistor/LED chain thru an LM317 configured as a constant current
>> source.  Just the IC and one resistor added, and no voltage variation
>> issue.
>
>The LM317 has a minimum current which may be greater than a single LED 
>can handle.   It will work good for a hundred mA or more.

Say wha?  If you use an LM317 as a current source, and open circuit it
(which is what you do with the brake circuit idle) then pray tell HOW
there is a a minimum current.  I think you are confused with the
standard configs for constant voltage, where there is an adjustment
pin bias current of typically 50uA.  Different config, different
issue.

>But probably 
>the biggest objection to using the LM317 is that they cost a buck or two 
>each, so if you use one for each series string of LEDs, you can end up 
>with more cost for the regulator than the LEDs.

Absolutely.

>If you're going to spend the money on an LM317, change it to an LM334, 
>which is a true current regulator.  Even cheaper might be a JFET with 
>the gate connected to the source.  Problem with that is that the JFETs 
>have such a wide range of Idss that you might pick a dozen of them and 
>they might all have a wide variation in current.

Problem with JFETs is firstly finding a family/type which has an Idss
figure matching the LED load requirement.  If you can do that, the
next issue is (as you mentioned) the parameter spread in N devices.
That's why I would lean to the 317T - a whopping $A1.32 in singles.

>Here's a URL for a LED current regulator using the grain of wheat 1.5V 
>25 mA bulbs from Rat Shack.  I don't particularly like using lamps for 
>obvious reasons.

Nor would I contemplate incandescents in this application.