Hi,
This is my first attempt at a web site.  So things,  
will no doubt be, dynamic for a while. Change is
the only constant.
MORE FAB()
admin@testvector.com
Xit:
Exit Sub '--
Xit: Exit
Sub '--
Above is the picture of a 30 watt Inferred Spot Light that I made.
It consists of 216, Fairchild QED-123, 80mW/100ma, Vf= 1.7v, 880nm, 18' IR LEDs.
Organized as 6 banks, each of 6 strings of 6 LEDs.
Each bank is supplied by an LM-317 adjustable regulators set for the 100mA of Forward.
Each regulator is mounted on a 6" x 1/8" strip of brass mounted over a 12v computer fan and sealed
to pull air in from the front over the LEDs,  then the heat sinks and blown out the back.

I have several cheap, $60, B&W CCD "board cameras".  They just happen to be extremely sensitive
to the IR that these LEDs emit.  I also have a pocket LCD TV that has an aux input for video.

I can easily light up the apartment building across the street from where I live with light nobody else
can see.  At 30 watts if the spot light is swept past closed eyes,  the sensation of brightness will be
evident but there will be no visual flash or brilliance.  So humans can detect light down in the IR
range.

Eventually I want to slave the spot light to
Gizmo.
The above view was taken with IR only, plus some light from
the TV at the far left.  Not visible to the naked eye.
Xit: Exit
Sub '--
Link to WMF of the schematic above.
Fab_02
Designed by Ken Leyla cc 2006, 2007
The surrounding images are of an
evaluation version of the circuit
on the left that I designed.

Nine 3,600 mcd, 3.4vdc, 25mA, White Light LEDs.
One LM-317 (1.25vdc), adjustable linear voltage regulator.
No current limiting resistors.  
Uses 4 to 8 AA NiCad, NiMH or Alcaline/Zink cells on a standard 9v connector.

I adjusted it for 3.21vdc @ 165mA.
With 8 x 1.2v AA cells, 9.6v,  I added a binder clip to the LM-317 as a heat sink.
I placed an LM-34 between the TO-220 tab and binder clip and got 110 'F.

Four 1,000 mA cells should last around 6 hours.
Four 2,500 mA cells should last around 15 hours.
It seems to be longer than shown above.
I get so used to it,  that I forget I have it on my head and turned on.
I'm fairly quick at catching when someone is squinting now.  Nobody says, "turn it off", they
just politely squint until the bell rings.

I can easily charge a second set of batteries while using the head lights.
Brightness control and power switch.
The light assembly and battery packs are held with Velcro.
It's an odd sensation to alway have light where I'm looking.  The peripheral image of
an object maybe very dark,  but as soon as my head turns and my eyes refocus, it's
light.   
Don't need to reach for the switch when just grabbing something from a darkened
room.   It's great when I take out the garbage or check the mail box at night or go to
the garage.
The above idea works well but is not really efficient.  The linear regulator produces too much heat when the
batteries are fresh.  Blowing away battery power as heat is not good.  After working on the Split PWM for the
attenuator project above,  I wondered if I could make a regulator for battery powered devices that could put
out several hundred mA at a constant level with a wide range of inputs raw supply levels.  I didn't want to use
inductors.  I wanted about 200 mA at 3.4v from 6v to 12v.  I had some N-Channel MOSFETs but I really
needed a 'high side' switch so I got a few.

The screen captures from my scope at the right are for three input supply voltages :
With the lower threshold set for
3.18v and the upper for 3.26v,

RAW   :  Regulated :   Load   :  Data  :  Duty  :  PW  : @ Output : @ Comp. filter.
Supply :   Output   :  9x LED  :   Set  :  Cycle :  ms  :   RMS    :    RMS
6.0v b  :    3.11v   : 081.1mA  :  Top   :  97.1  : 12.6 :   5.8mV  :   5.7mV
9.0v    :    3.24v   : 141.2mA  :  Mid.  :  54.4  :  6.8 :  13.4mV  :   5.8mV
9.0v    :    3.31v   : 188.8mA  :        :  91.9  : 12.3 :  10.3mV  :   5.7mV *
13.0v   :    3.25v   : 147.9mA  :  Bot.  :  19.0  :  2.5 :  17.7mV  :   6.5mV
13.0v   :    3.32v   : 200+mA   :        :  19.6  :  4.0 :  20.0mV  :   8.0mV **

* duty cycle limit,  ** current meter limit.
Data Set
TOP
Data Set
Mid.
Data Set
Bot.
I think that I can improve the operation by using faster caps.
Raw DC :  LED 'F  : Output E & I  : Pulse Width ON / PW OFF  : 36 Pulses   : MOS Switch Temp : RMS across Load :  Duty   : Lab Supply Mode
14.0V  : 170.8' F : 3.69V & 530mA : 540.0us ON / 4.140ms OFF : CYC 343.0ms : more than cold. :    32.70.mV     :   2.1%  : limiting @ 1030mA & 13.60v
12.0V  : 170.6' F : 3.67V & 530mA : 760.0us ON / 3.920ms OFF : CYC 335.0ms :  almost cold.   :    22.38.mV     :  16.3%  :
Not limiting @ 1030mA & 11.82v
10.0V  : 170.7' F : 3.68V & 530mA : 760.0us ON / 2.020ms OFF : CYC 205.0ms :     cold.       :    21.95.mV     :  26.6%  :
Not limiting @ 1030mA & 10.01v
9.0V  : 169.6' F : 3.68V & 530mA : 1.600ms ON / 2.240ms OFF : CYC 281.0ms :     cold.       :    20.37.mV     :  41.4%  :
Not limiting @ 1030mA & 10.01v
8.0V  : 168.8' F : 3.67V & 530mA : 5.600ms ON / 1.640ms OFF : CYC 490.0ms :     cold.       :    22.17.mV     :  75.4%  :
Not limiting @ 1030mA & 10.01v
7.0V  : 153.7' F : 3.54V & 450mA : 27.80ms ON / 2.400ms OFF : CYC 1.124s  :     cold.       :    19.52.mV     :  93.5%  :
Not limiting @ 1030mA & 10.01v
New Test Data.  This is up through 530 mA.
I exchanged the 25mW LEDs for a 2.5 watt LED, the regulator sourced as much as 2 amps.