Thanks for the detailed information, it is quite helpful.  :)

The 8 LED's backside with the soldered together leads are potted in=20
epoxy, it would be easier to start over. This setup does not get used=20
frequently, but to have a LED failure while in use would not be good.=20
There could be a well drilling rig and crew @$200/hour standing by. From=20
what has been said, I think my best would be start from scratch and=20
either to connect the 8 LED's in series with a LM317 set up as a current=20
regulator. For 3.4 volts * 8 =3D 30.4 volts +3.5 volts =3D 33.9 volts, and=
=20
with the Cat-5 losses maybe 5 volts, need say around 40 volt power=20
supply, but then I'm getting out of the camera's LM7812 input range.=20
Probably better leave the current 2 parallel, 4 series configuration,=20
but provide a LM317 current regulator for each of the series strings,=20
and the remainder of the setup as I proposed without the LM317 voltage=20
regulator.
On 10/19/13 10:33 AM, RussellMc wrote:
> LEDs should (~~=3D must) be driven by constant current.
> Trying to drive them with ~=3D their notional Vf will lead to a substanti=
al
> variation in most cases.
> When operating on 12V you should have only 3 LEDs in series, plus current
> control..
> When operating with 24V you can have up to about 6 LEDs in series plus
> current control.
> The more "headroom (aka spare voltage) that you have the easier it is to
> get ~=3D constant current with simple circuits.
>
> It is exceeding bad practice (also exceedingly bad practice) to attempt t=
o
> drive two parallel strings of LEDs without any attempt to current balance=
..
> I have seen some LEDs that are so well matched for Vf that you could
> 'almost get away with" doing this, but most LEDs have enough variation,
> even in the same batch, to make this problematic.
>
> LEDs should be driven at MORE than their rated combined Vf with a constan=
t
> current source taking up the difference. You can use as little as an LM31=
7
> and one resistor as a constant current source.
> You'll need about 3.5V "headroom" for an LM317 to operate as a current
> source.
> eg for 4 LEDs. V min ~=3D 3.8V x 4 =3D 15.2V.
> Headroom =3D 3.5V say.
> Vinmin=3D 3.5 + 15.2 >=3D 18.7V.
> ie ~=3D 24V would be dandy.
>
> Vin =3D LM317 Vin
> Iout =3D LM317 ADj.
> Rcc from Vout to Vadj.
> Rcc =3D 1.25 / I A =3D 1250 / I_ma
> Dissipation in R =3D~ 6/R or 1.25 mW  per mA
> eg for 25 mA.
> R =3D 1.25 / .025 =3D 1250 / 25 =3D 50 Ohms.
> Pd_Rcc =3D 50 x 1.25 ~=3D 63 mW.
>
> If you have 4 LEDs in series and this is unchangeable, you could short on=
e
> out per group and probably get a better result.
>
> If these are on only occasionally then 'any brand will do", almost. If th=
ey
> are intended to spend thousands of hours on over their lifetime then you
> should use LEDs from the major LED manufacturers OR ones made under licen=
ce
> to them. LED lifetimes are still woefully low in many cases - despite
> claims to the contrary.
>
>
>               Russell

--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.