Tinkering with home LED illumination.

[spldart]

I have one of those 'Lights of America' accent lights with 30 LED's and runs
like 3 to 3.5 watts. I use it for both spot lighting and when I'm working on
a pc's hardware and when I'm web-cam'ing on Skype. I had a problem where
10 of the 30 LED's were intermittant, turns out they individual who was assembling
this thing did a poor job of soldering, so I needed to take it to work to fix it.
I hardly ever actually use a warranty.

Anywho.. i saw that little circuit board down in there and thought to myself 'hrm'

So I mapped it out (see attached image of post it note schematic)

I then remember how I wish I could have this thing at only 50 to 60% standard
intensity at times so i look at the schematic... see those to caps in the top
left of the post it note... .82 and 1.0 uf at 250 volt? Those decide more than
anything the amount of bias on those led's
So I drop them by replacing with two .56 uf at 400 volt. (what I had on hand) See
notes on post it note on changed voltages... This dropped bias of each LED from
approx 14ma to approx 8ma.

But then I think... what if I would like to have it back to it's original intensity
once in a while? hehe
So I install that switch (see attachment) and it parallels in and additional .68uf
at 400 volts... Bring overall parallel capacitance to near the original 1.82....
1.8 iiec

Just me being stupid in the light of that very lamp on 'Low'

Mystery component is a small glass fuse covered with shrink wrap.

Yeah... That electrical tape will get replaced by epoxy

 

[audioguru]

Did you notice that when you increased the current by almost double, then the brightness of the LEDs increased only a small amount?

Because our vision's sensitivity to brightness is logarithmic, not linear.

Two LEDs look only slightly brighter than one LED.
Ten LEDs look twice as bright as one LED.
Ten times the current in an LED makes it look only twice as bright.
We can see in dim moonlight or in bright sunlight. A logarithmic difference.

 

[spldart]

I never attempted to increase... Just decrease... And I did noticed that the drop in intensity didn't seem to drop linearly.

Tell you what though.. flip that switch on the fly and you can see the significant drop in intensity real time.

i wonder if you get the same kind of gains in hours life outta the LED junctions that
you get in incandesent lighting when you reduce Vdrop across the filament.

 

[spldart]

Just noticed that it would look like i increased value in capacitance when i replaced that 105 and 824 cap. The new larger caps are outta old Sony XBR tv chassis's, deflection
board. They are 564 at 400v/ .56 uf at 400 volts That takes the original parrallel 1.82
uf down to 1.12uf and then my switch cuts in an additional .68 mf at 400 volts bringing
that portion of the circuit to 1.8uf... Nearly where it was stock. (hi position)
400 volt rated caps are larger per uf than 250 volt rated caps... i just used what i had
laying around

edit.. my spelling is atrocious

 

[BrownOut]

Cool idea about how to dim your lights. What if you switched in/out one or more of the LED strings?

 

[spldart]

In stock form.. if you open a string the rest get more bias per junction... Possibly exceeding their ratings... And I found that changing those 3 100 ohm resistors for 220's only reduced the designed bias per junction from 14 ma to 13 ma per. So that wasn't a practical way
to reduce/dim the lamp overall. It's an interestingly balanced circuit

edit. (all readings taken on a fluke 83 dmm and or extrapolated from ohms law and
existing readings) (one reading taken from a kill-a-watt meter, not very accurate but
kewl nonetheless)

 

[BrownOut]

Leave it to a Sugarland guy I grew up in Northwest Houston; Aldine Mail Rt./I-59 area.

 

[spldart]

Found the other two... I should have taken more of the wiring with the switch but I didn't.