Brooke Clarke 2000 -2024

Light Units
    Flashlight Design
Filament LED
High Brightness LED
    AA Battery
    9Volt Battery
Lamina Atlas 7 Watt
Task Light
E10 Base LED 
LED as a Light Sensor
Origami LED Kit
Blocking Oscillator - Joule Thief seperate web page
    Single Use Camera Flash Blocking Oscillator
     Joule Thief
Neon Blinkers 
Speed Test
White - Phosphors
White - Red, Green, Blue
Numbers & Characters
    Flipping Dot
Seperate web page for Lights & Electro-Optical Gadgets & Flashlights patents and DIY info

Light Units

There are a number of  ways of measuring the light output from a lamp, LED, etc.

Candela (was: Candlepower)

Is a measure of the luminance over some area.  For example a sheet of paper is made translucent by soaking it with oil.  On one side is a candle made to a specification and burning at the specified rate.  On the other side of the paper is a light source.  By moving the light source toward or away from the paper a location will be found where the two sources are equal in brightness as seen by your eye.  By measuring the disance from the light source to the paper it's luminance can be computed based on the inverse square law.

If the light source was another identical candle then it would be the same distance from the paper.  But if a reflector was  on the light source candle it could be moved further away.  The light output from the T series LEDs is specified in milli candelas.  A bright white LED running at 20 ma might be rated 10,000 mCd but only over a very narrow angle.


Are a measure of the total visible light in all directions.  For example a 100 Watt light bulb in the U.S. (115 VAC) puts out about 1700 Lumens.

Lumens per Watt

A measure of how efficiently a lamp converts power into visible light is Lumens/Watt.  For the 100 Watt bulb that's about 17 Lumens per watt.  Since power in equals power out the total power radiated from a lamp is the same as the power put in.  Most of the power out of electrical lamps is at non visible wavelengths, like heat.  City governments or businesses are very concerned with Lumens per Watt since they pay the electrical bill.  This is also important for battery powered items like flashlights since it puts an upper limit on the run time.
Here are some values for reference:
Edison's first lamp 1.4
Edison's production carbon fil lamps3
Edison's prod. tungsten fil lamps4 8
Infrared lamps 6-9
Incandescent lamps 10-40
1 watt warm White (3300 K) LumiLED 20
1 watt cold  White (6500 K) LumiLED 46
Fluorescent lamps 35-100
Mercury1 lamps 50-60
Metal Halide lamps 80-125
High Pressure Sodium lamps 100-140

Note 1 - All fluorescent lights, both conventional and compact use Mercury.
Note 2 - May contain Mercury
Note 3 - The most popular carbon filament lamp was the 16 candle power size that consumed about 60 watts and provided about 200 lumens.  The smallest line powered lamp was the 8 CP that used about 25 watts and provided about 100 lumens.
Note 4 - The Tungsten filament lamps were made in 20, 32, 48, 80 and 200 CP (250, 400, 600, 1,000 & 2,500 lumen) sizes for mains power.
So a LED based lamp for general illumination needs to put out 100 to 200 lumens.  If the L/W is 20 (warm white) then it takes 5 to 10 LEDs to make a useable lamp.
The L/W goes up as the current is turned down on modern LEDs, but then the cost of final lamp goes way up since many LEDs are needed.

Flashlight Design

Many of the flashlight sellers make inflated claims about the brightness of their product.  It's probably best to ignore any claim made by the seller and instead look for a review that has some type of comparison testing.  Instead of using oiled paper (see above) you can just shine two lights on a white wall side by side and compare them.

Luxeon K2
            Power 120 LumensThe main problem I've found is the high amount of wasted power.  For example the Made in China "Luxeon K2 Power 120 Lumens" flashlight that runs on two LIR123A cells.  This is a cheap flashlight that uses a 4.2 Ohm resistor to set the LED current.  Since two 3.4 Volt batteries are being used and the voltage drop on the LED is about 3.85 volts that leaves 2.95 volts or 700 ma or about 2 Watts to dissipate in the resistor and about 2.7 Watts in the LED for a total nearing 5 watts.  This means that after about 10 minutes this flashlight it too hot to hold.  Lumens per Watt doesn't mean much when you burn power in a resistor.

The K2 is rated for up to 1.5 Amps (see K2 Drive Currents below) the LIR123A cell is rated to deliver a maximum current of 800 ma where the K2  puts out 93 Lumens, not 120.  These cells are rated for 400 mAh capacity at 0.2C, i.e. at 0.2 * 400 ma = 80 ma.  There is no specification for  the 2C discharge capacity.  But the best case is 400 mAh capacity being used at 700 ma so will last at most 34 minutes. 

A step up from a resistor is a regulated current source.  For example the 7135 regulator is used in many flashlights.  This is a 300 ma linear regulator that has a constant current output.  A flashlight might have three of these so by pressing the "Clickey" switch you get 0, 300, 600 or 900 ma current into the LED. 

The most efficient way is to use a buck type Switching Mode Power Supply.  In this case the input voltage is higher than the LED voltage (typically in the 3.8 V area).  See the Experiments below.  The reason I say with an input higher than the LED voltage is because the capacity of most batteries decreases with increasing current.  For example a common 9 volt battery could be used to drive a high brightness LED at 350 ma for 1 hour (see experiment below) but when a Switching Mode Power Supply is used between the battery and LED the same battery would last for about 4.5 hours.

It's possible to use a boost type SMPS with a single cell battery to drive a high brightness LED.  This has the effect of greatly lowering the capacity of the cell in exchange for a much smaller light.

It costs more to use a Switching Mode Power Supply and so the lights are more expensive, but the battery life is much much better.  Unfortunately most of the advertisements don't tell you if they use a resistor, linear regulator or SMPS type supply.

7 Nov 2007 - In conclusion I'd say that most (90%?) of the High Brightness LED flashlights on the market have been designed more for advertising value than for benefit to the user.  It seems that the information a user would need to know to choose between the large number of makes and models like the type of current control (resistor, linear regulator, SMPS) or the efficiency of the current control, the candela brightness at some beam width, weight and size are all missing from the ads.

Also See the Flashlight Patents page for tube sizes and other construction info.

Filament LED

The Magnetic Levitating Lamp uses filament LEDs.  They are also used in a number of LED lamps because the have the look and feel of old fashioned filaments.

Filament LED
The interesting thing is that this one turns on with only 1 mA (Fluke 87 DMM in Diode mode) with a voltage drop of 2.522 V.
This is amazing performance.  Normally it takes more than 3 volts to get anything near a white light from an LED.
Filament LED With image brightness adjusted to look more like with eyes.

High Brightness LED

Philips Luxeon aka. LumiLED are LEDs that run at currents in the 350 to 1,500 ma range as compared to the common T13/4 type LED that runs at 20 ma.  A white LED may have a forward voltage of 3.2 v and with a current of .35 amps that's 1.1 Watts most of which is heat that needs to be dissipated.  One way of doing that is to use a special PCB material that has an aluminum core, but that's an expensive solution.  Another option is to use a copper plug and CPU thermal epoxy to attach the emitter to the copper.

Oct 2007 - Phillips now has the Rebel which is a small Surface Mount Technology LED on a small heat sink.  It's designed to be used on a normal printed circuit board with a pattern of over a dozen plated through holes nearby to act as the heat sink.  150 deg C junction temp and 350 (80 lumens) to 1,000 ma drive.

Nov 2007 - Some T10 size LED are showing up on eBay.  These appear to be high brightness LEDs packaged in the clear plastic T10 package.  These packages have a built-in lens so the candela rating is very high.  They have special mounting requirements and heat sink needs, which so far it's not clear how to meet.

The first generation emitters have a Tjunction max of 120 deg C and so need more heatsinking than the newer K2 emitters which have Tjmax of 180 deg C.
LumiLED LXHL-BW03 Warm
          White12 Feb 2007 - This is a first generation 1 watt emitter, the LXHL-BW03.  It's a warm white color that is very pleasing, not like the white that has the excess blue.  I have it rubber banded to an aluminum bracket and the bracket does not get even warm.  0.340 Amps @ 3.15 volts.  Note at cold turn on the Vf is 3.26 volts and after running a long time the Vf is 3.25 a drop of 22 mv.  The data sheet says -2 mv/deg C so the LED is only heating 5.5 deg C over ambient.

I have more LumiLEDs on order (3 watt) that run at higher powers to get a feel for the heat sinking needed.

I have this LED hung off the arm supporting a halogen desk light.  The LumiLED is considerably dimmer than the 50 watt bulb, but it's adequate for seeing what I need to see and the color is pleasing.

All the fuss about hermetic storage relates to wave soldering.  If an emitter is left out in the open it can get moisture inside, then when it's wave soldered it may explode.  But this isn't a problem for hand soldering.  It's not a problem for the "Star" parts that are already solderded to an aluminum heat sink.

Efficiency is measured in Lumens per Watt.  Here are some numbers for comparison.

Edison's first lamp 1.4 L/W
Infrared lamps 6-9 L/W
Incandescent lamps 10-40 L/W
1 watt warm White (3300 K) LumiLED about 20 L/W
1 watt cold  White (6500 K) LumiLED about  46 L/W
Fluorescent lamps 35-100 L/W - all use Mercury
Mercury lamps 50-60 L/W - uses Mercury
Metal Halide lamps 80-125 L/W
High Pressure Sodium lamps 100-140 L/W
Theoretical max for white light 225 L/W

Data sheet values for the cold white (6500K) K2:

K2 Drive Currents

350 mA
700 mA
1 A
1.5 A
1.2 W
2.5 W
3.7 W
5.8 W
Lumens decrease 10% for each 50 deg C rise in temp.
There is a tradeoff related to what current is used to drive a K-2.  At the lower currents you get higher efficiency (L/W) but less total light output.
At higher currents you get more light.  But heat sinking comes into play at the higher currents.

Experiments 1 watt

LXHL-BW03 note1
wrmWht BW
LXHL-MWGC note1 Star
LXHL-NWG8 note1 Star-O
10 deg
emitter Wht
emitter RlBlu
Lamb 3.95
note1 - these 3 are all the same warm White 1 Watt emitter in 3 different packages.
The emitter is the raw LED.  The star is an emitter mounted to a PCB with an aluminum layer. The Star-O has a small parabolic reflector.
note2 - the power supply has a 1 Amp current limit so did not test at 1.5 Amps.  The LEd lights with a current that appears to be under 1 ma (below the resolution of the Agilent E3617A power supply).  As the current is increased the light output increases (see Data Sheet DS51 Fig 11 on pg 15) but the slope is flattening slightly as the current increases.  3.76 volts @ 1 Amp at turn on, then 3.40 v after some time.  With -2 mv/deg C the temperature is going up about 18 deg c or 5 deg C/watt.
note3 - the voltage drops to 3.77 after a minute or so. meaning 9 deg C temp rise or 2.25 deg C/Watt.  I used a new bracket and applied a good size gob of silicon grease to this one. 

19 Feb 2007 - First look at the LM3402 driving a LumiLED LXK2-PW14 at 350 ma (the K2 is good up to 1.5 amps and is very bright @ 350, but can put out more light at 1.5 Amps).  So the LM3402 can drive both types.

The current in the LED is regulated at 354 ma and the Vf is 3.046 so the power is constant at 1.078 watts.
Efficiency = 1.078/Power In
Note1 - At low input voltages the regulator is not putting out the full voltage and so the LED is not getting the constant power it does for higher input voltages.  So the efficiency calculation is wrong (looks better than it would if the actual LED power was measured).

Note that driving a single LED is the lowest efficiency because most of the loss is independent of the drive voltage because of the constant current.  So adding more LEDs improves the efficiency.

Now two K2 LEDs in series LXK2-PW14 (white) & LXK2-PR14 (royal blue).
The current is the same as before 354 ma but the voltage is now 6.48 for a power of 2.294 watts.

note1 - although the LEDs will light at voltages below 12 volts, they are not as bright as at higher voltages so it's difficult to calculate the efficiency.

Low Current Operation

I don't have a good way to measure brightness so I'll describe in words what I've observed.  The high power LED seem to work down to very low currents (below the range of the current meter on the HP E3617A power supply (1 ma).  The efficiency gets better as the current is lowered.  So if you run a 1 Watt LumiLED at 20 ma, it will put out more light than a 5mm type high brightness white LED.  The data sheet indicates about 46 lumens/watt which is maybe twice as good as the high brightness 5mm LEDs.  Note that 5mm LEDs typically have a lens that focuses the light into a 15 deg beam so it's hard to compare to an emitter that's putting out a 120 deg beam.

AA Battery

20 Feb 2007 - Normally  you would not use AA batteries to power a LumiLED that draws 350 ma.  This is because at that current the batteries would last between 90 minutes and a few hours.  Note that this is the case independent of the number of series batteries when a linear voltage regulator, like the LM317, is used.  BUT it is what done in most of the cheap LED flashlights.

On the other hand with a Switching Mode Power Supply (SMPS) the power drawn from the battery pack remains constant, but the current changes as the pack voltage changes (either due to the number of cells in the pack or due to discharge).  So if 10 AA cells are used to drive a single LumiLED at 350 ma the battery current is 100 ma.  The Constant Current Performance chart on the Energizer E91 data sheet shows a run time of 15 to 30 hours for a 100 ma load.

If the Constant Power Performance chart is used and the power from the 10 AA battery pack (13.8 V * 100 ma) 1.38 watts and divide by 10 cells or 0.138 watts per cell.  The run time ranges from 12 to 20 hours.

Now looking at the Energizer L91 data sheet.  At 350 ma constant current the run time is 8 hours and at 100 ma constant current is 30 hours.
But now looking at 1.38 watts Constant Power the run time is more like 40 Hours.  The L91 was designed to supply SMPS supplies where the current increases as the battery discharges.

On the E91 Constant Current curve note that the slope gets steeper above 100 ma.  A change in slope represents a change in amp hour capacity.  So you get more amp hours when running at or below 100 ma.

9 Volt Battery

1 Watt LumiLED & 9
          volt batteryThe energizer 9 volt Alkaline battery (522) is rated for 4.5 hours at 100 ma. and maybe 1 hour at 350 ma.  So by using a SMPS the battery lasts more than 4 times longer.

It's very hard to take a photo of a 1 watt LED operating becasue it makes a lot of light.  This photo gives an idea of the relative brightness but is not in focus.

Single Cell Battery

There are applications where the 1 watt LED needs to be powered by a single cell.  The cell voltage varies with the chemistry from 1.2 v or Ni-MH to 3.4 v for rechargable lithium.  There are SMPS that are designed to raise the voltage but they do it by drawing more  current from the battery than the load consumes.  As the current drawn from a battery increases the amp hour capacity decreases.  This is why the battery data sheets have discharge curves (constant current, constant resistance and constant power are the common ones).  This method of powering a 1 watt LED works, but at the expense of less battery capacity.


When any of these, except the Star-O which had the emitter hidden, are used in a way where you can directly see the emitter it's not comfortable because of the brightness.  The output of the Royal Blue emitter is specified as about half a milliwatt and all the other emitters are specified in lumens (the Star-O in candelas) so it's hard to compare the specs.  To my eye the brightest by far is the PW14 emitter.

So far Philips has not offered the K2 in any of the "Star" packages which mount the emitter on a PCB that has an aluminum core that acts as a heat sink.  One of the reasons for this may be that the max temperature on the K2 emitters is much higher than on the other types allowing it to be mounted on top of a normal FR4 PCB.

Power Supply

The high brightness LEDs are best driven by a current source.  There are a number of different ways to do that.

Lamina Atlas 7 Watt

White LEDs really are a blue LED that illuminates a phosphor.  The light contains a lot of UV and can burn  your eye  just like looking into an arc welder.

It's been 4 days and I'm no longer suffering like I was when this was written.  It may be that it's just the point source visible light that's causing the problem.  I'm looking into what caused it.

Lamina Atlas LEDThe company has many patents going way back in the area of ceramic circuits like used for microwave components.  The key technology here is the ability to get a good heat path through an insulator.  The disk has four LED chips mounted in a square pattern.  At 1 amp forward the Vf is 7.68 Volts so the LEDs must be wired in a series parallel combination.  Part of the Atlas family is a specialized extruded aluminum finned heat sink that comes in lengths of 0.75 or 1.5 inches, this one is the 1.5" version.

This Light Engine is the Lamina NT-42D1-0425 Warm White (3000 K) rated for 8.2 Volts at 1.05 Amps (8.6 Watts input) and 162 lumens out.  They make brighter cold white (5000 K) LED, but I like the feel of the warm light.  The data sheets have quite a bit of data at 350, 700 and 1000 ma.

The photo at left was taken with 7.68 volts at 1 amp (7.68 Watts).  I was trying the three optics, wide, medium and narrow by sitting them on the LED+heatsink and with the spot beam little light was coming from the optic lens toward the camera, but the direct leakage from the LED was so bright that the flash did not fire and the image came out mostly black.  I've tweaked it in photoshop to make it viewable.  The 40 and 30 small optics are made by Fraen and I suspect so is the 10 spot optic.  Fraen makes optics that match most of the high power LEDs.  OP-4LN2-0492 10 deg beam sitting by gravity on lamp-heat sink,   OP-4FW1-0442 30 deg  &   OP-4FW1-0441  45 deg beam optics in front.  These are a solid lens in a holder.

I'm using Wakefield 126 (accessories.pdf 1.3 MB) Thermal Joint Compound, not the 120 Lamina recommends.  The way I read the specs this is twice as good when first installed and even better after the 120 has aged for 6 months.

Since the disk with the LED needs to be attached to the heatsink using thermal grease and held in place with screws you can not attach the optic until that's done.  So to make a task light some sort of can needs to be added to what you see in the photo.

Lamina NT-42D1-0425 LED

There are four LEDs inside the Lamina LED assembly.  I've had it out in my office area and you can see that it's covered with dust.  In a real world application the LED assembly needs to be in a dust proof environment

Note the use of 2-56 Round Head screws since they have a smaller head diameter than pan heads.

Photo taken with about 4.8 Volts and less than 1 ma current (Agilent E3617A display .000 Amp).

LumiLED -
Cree - no warm white
Seoul Semiconductor - P4 - broken web pages 16 Feb 2007
Lamina - LEDs aimed at use in buildings

Lamina Atlas,
            Warm White, Heatsink, 10 degree OpticIf you look at the back of the optic you can see the clear plastic part that has a hemispherical hole that fits over the LED silicon.  The optic is actually seated on the silicon and can be rocked a little.  Drilling holes in the heat sink would serve no purpose.  The intended mounting scheme is to place the application PCB over the top of the LED star board with large via holes in the PCB that line up with the pads on the star board then solder them toghther.   Then the optic pins, which go completly through the application PCB can be mushroomed using a hot iron.  But that makes for an akward assembly operation thereafter.

A better solution is to put the star LED board and optic into a aluminum can from the front.

Task Light

Task LEDThe parts shown are available as an assembled board that has a combined buck Switching Mode Power Supply and a micro controller to allow programming the drive current and a number of handy user functions and the assembled LED head on a 6" length of Loc-Line.  Called the FlexiLED Task Light.  There are slightly different versions with the micro controller  commands suited for fixed, portable or bicycle use.

The Loc-Line is stiff enough to hold the head where you point it.  Much superior to the old flex metal goosenecks that had a lot of built-in spring.

The relative scale of the Task Light and the Lamina Atlas in the above photo is close to accurate.  The Task Light head is about 1" dia and the Lamina 10 deg optic is about 2" dia.

There are a number of settings that relate to batteries such as warning when the battery voltage goes below some value or an auto off mode that will turn the light off after some time period. 

To program it you push the white button visible on the PCB (or connect your own button to the supplied thru holes) and the LED will blink to acknowledge the commands.   Not the most user friendly interface, but just the thing for a bicycle or other application where once setup very simple button presses suffice.

E10 Base LED 1 Watt

E10 base LED lamp
T10-WHP from Super Bright LEDs.
I got this as a universal lamp for testing various flashights

The base is Edison 10 mm (E10) the standard for all the early flash lights.  Specified to be 1 watt with a 2.8 to 12 Volt input range, 35 Lumen 120 degree beam.
E10 based lamps do not have pre focused filaments and many flashlights have a way to change the axial position of the lamp to find focus.  That's the next test scheduled for this lamp.

But didn't expect it to be so high tech.  Here is some I-V-P data:

I ma
P watt
Note 1 for voltages between 2.something and 3 the brightness is increasing and constant for higher voltages.
This type of load curve incicates a Switching Mode Power Supply, not a simple resistor or linear voltage regulator.
And it all fits in a very small space.  How did they do it?  let me know.  Although the Joule Thief circuit will fit this space I doubt it would supply 1 watt.

LED as a Light Sensor

RS 276-143 IR LED extreme Close upWhen light falls on the silicon diode in a LED it acts like a small solar cell generating electricity.   So it can be used as a light sensor. 

Pointing the Radio Shack 276-143 IR (940 nm) LED at the Sun shows 924 mv (Fluke 87 DMM 10 M Ohm input).  But when the meter is configured to measure current there is no output (i.e. this solar panel will not deliver any power).  The output is about 930 mv when the LED is placed 1/2" below the glass of a halogen desk lamp.  When held under my desk the output is more like 90 mv, so there's some weak IR bouncing around inside coming from sunlight in the window.

When the IR Filter from a SDU-5/E Strobe Light is placed between the halogen desk light and the LED the output drops only a few percent.  Note that this filter blocks all visable light from the strobe so well that when the strobe is running and you place the filter right at your eye you can not see anything.

Connecting the LED to a Fluke 87 DMM in DCV mode and pointing the LED at my eye while it's touching my reading glasses to help hold it steady, then turning on Min Max mode.  When looking left or right there are beeps indicating the the LED is seeing my open eye movements.

Repeating the test with both eyes closed also causes a beep or two indicating that it's working but not as sensitive.

See the Electro-Optical Gadgets page for a very sensitive way to measure light using an LED and a PIC microcontroller.

LED Drive Circuits

ELM - White LED Head Lamp - a 2 transistor 2 AA circuit Pocket LED Light  a 1 AA 1 transistor blocking oscillator
KnurdLight PIC Example Project - This is the 3rd generation of a 4 white LED hat brim light packaged in a waterproof tupperware box a little larger than the 2 AA cells that will power it for about 20 hours.  Full documentation on line.

LED Flashers

Interesting observation about a home made bicycle safety flasher:
" . . . An interesting problem occurred when I was designing it.  I am somewhat prone to migraines.  The day I was first working on it, I had it set up with 50mS continuous pulses, which would light each LED at about 5 Hz with no rest between pulses.  The LEDs are extremely bright up close, they hurt the eyes with a fresh battery.  I found the effect of the continuous hammering of the 5 Hz light pulses mesmerizing, and I found myself staring at it blankly, even hypnotically.  Later that day I got a humongous migraine. The next day, shaken from a night of pain, I turned on the headlamp again and realized it was stimulating another head-pounding headache.   I changed to the 900mS between pulses after that, and it never had the same effect on my aching head again.  Maybe I have discovered the elusive nonlethal weapon?"  -Piclist-


LM3909 D-cell, Jumbo LED, 500 uF 50V Cap

LM3909 D-cell, Jumbo LED, 500 uF 50V Cap

This now obsolete chip when combined with an external capacitor, 1.5V battery and an LED will flash the LED for a very long time (years).  The data sheet suggested that the battery might even last longer on the flasher than it would just sittingon the shelf.  I think they were used in emergency flashlights, like on commercial aircraft to let you know the batteries were OK.

The LM3909 will not flash the newer color LEds, like a white or blue since thier voltage is higher than a red LED.

Replacement For The LM3909 LED Flasher / Oscillator -
1.5 Volt LED Flashers -
Futurlec - New Old Stock LM3909

Origami Cat LED Kit

Requires cutting into square and punching three round holes (can use LED to enlarge a slit).
Folding so that ..... is visible after fold, or so that -------- is hidden after folding.
Bend LED leads about in half so they are all within the battery diameter.
Install LED and battery inside and make (9) fold.

Origami LED Kit
Origami LED Kit

Blocking Oscillator

Joule Thief moved to it's own web page.

Blocking Oscillator w/o LED

If a Zener diode is connected between the Base and Emitter (cathode to emitter) it will clamp the base voltage to the Zener voltage protecting the E-B junction from breakdown.
Now if the LED is disconnected and the supply is at 1.0 Volts the collector waveform is a voltage spike about 82 volts high and 200 ns wide at 40 volts.  The duty cycle is small.


Blocking Oscillator - uses a dual winding transformer to form a very low cost oscillator circuit.  Operation may depend on the saturation current of the drive transistor and/or core saturation of the transformer.  Typically no air gap in transformer.  A secondary winding can be added (three windings total) to provide a high voltage output.  Vout = (Vbat - Vces) * Ns/Np  The voltage step up is due to the transformer winding ratios.  The magnetic circuit has minimal energy storage.  Low power circuit.  Used in throwaway cameras for the strobe high voltage generation circuit.

Flyback Oscillator - the primary of a transformer is charged up then drive is turned off allowing the collapsing filed to generate a high voltage in the secondary.  The transformer has a small controlled air gap.  The magnetic circuit must store all the power needed for the output. Medium power circuit.

Inverter - A pair of transistors drive the primary of a transformer.  The magnetic circuit has minimal energy storage  (no air gap).  The duty cycle is very close to 50%.  Can be high power circuit.
By adding a very narrow air gap the performance is enhanced by greatly lowering the residual magnetism.  This eliminates large current spikes at turn on.  These gaps are made by lapping the two faces to be as flat and smooth as possible.

Single Use Camera Flash Blocking Oscillator

This circuit can easily be removed from a single use Kodak flash camera.  It runs from a single AA battery.
Kodak single use
              camera Flash PCB Front
Kodak single use
              camera Flash PCB back
The button in the center of the PCB starts the flash charge cycle.
The two sheet metal parts on  the right go to the shutter switch.
Notice the Neon bulb at the upper right is on.
The 120 uF 330 V cap has a little more than 300 Volts after a the charge cycle. 
The energy stored in the capacitor is 1/2 * C * V*V (Watt Seconds) = 0.5 * 120e-6 * 300 * 300 = 5.4 Joule (1 Joule = 1 Watt Second)
The circuit draws just over 1 amp and the current tapers down to 1 mA in about 17 seconds.  The Neon lamp glows but does not blink when the charge cycle is done.

For driving the synchronizer in a Western Union Self Winding Clock the current is controlled by a series resistor.  The time constant is L/R so for a 100 ma drive and 300 Volts the resistor would be about 3 k Ohms.  If the synchronizer coil was 1 Henry (a very high estimate) then the time constant would be 1/3000 or about 330 micro seconds.  That's very fast compared to the operating times of mechanical systems and fast compared to the thermal time constant of the coil.  As the resistor is made smaller two things happen.  The time constant gets longer, allowing more time for the mechanical movement, and the current increases, which is probably OK until it gets near the wire exploding value.  Note that wire heating will not be a problem since the synchronizer is pulsed only once per hour.

Don's Xenon Flash and Strobe Page
Strobe Lights and Design Guidelines, Useful Circuits, and Schematics - has info on similar circuits
Sam's Strobe FAQ - scroll the page, the links go to other pages

Blocking Oscillator Patents (also see Joule Thief & Flux Gate patents)

Blocking oscillators are used in many applications because they are simple (low cost, small size, light weight, etc.).
They can be free running or triggered.  When triggered the output is a pulse rather than a spike and the pulse width or delay may be the key parameter and/or the power in the pulse.  There are various modes of operation:
2211852 Blocking Oscillator Apparatus, Max Geiger (Telefunken), Aug 20, 1940, 331/146 ; 336/182; 336/185 -
has good description of why windings are not done layer by layer when low capacity is important
2816230 Blocking Oscillator Circuit, Lindsay, Dec 10 1957, 331/112 ; 331/146- transistor
2857518 Transistor Blocking Oscillator, Robert C. Reed (North American Aviation), Oct 21, 1958, 331/112 ; 331/146; 331/183 -
triggered power pulse generator
2894212  Blocking Oscillator, Dalton L. Knauss (Hoffman Elec), Jul 7, 1959, 331/148 ; 327/191; 327/596 -
 saturable core mentioned sharp rise and fall times
2976489 Blocking Oscillator, Maurice R. Bums (North American Aviation), Mar 21, 1961, 327/267 ; 331/149 -
tube circuit used to generate a precise time delay
2988709 Transistor Blocking Oscillator for Telemetring, Herbert K. Janssen, Jun 13 1961, 331/112 ; 324/711; 331/65; 331/66-
used to generate audio frequency sawtooth whose frequency is proportional to input resistance
3005158 Core Saturation Blocking Oscillator, Robert J. Spinrad (AEC), Oct 17, 1961,
327/177 ; 327/191; 331/105; 331/112; 331/113R; 331/148; 331/149; 336/110; 336/155; 340/870.24; 340/870.33; 377/96-
Equations for active device saturation modes and core saturation modes
uses a permanent magent in the transformer to control the magnetic bias and hence the pulse width
2605423 Oscillator, Leon Bess (Navy), Jul 29, 1952, 331/146 ; 331/148 - linear sawtooth type, 2 triodes
2717961 Frequency Division, Charies W. Johnstone (Navy), Sep 13, 1955, 331/149 ; 331/146; 331/148; 377/96-
2740047 Electric Pulse Generators, Alan J Bayliss (GE), Mar 27, 1956, 331/51 ; 327/231; 331/148; 377/96- divide by 10 tubes
2838669 Counting Network, JR Horsch (GE), Jan 10, 1958, 377/97 ; 327/100; 331/148; 331/149; 331/177R; 365/206-
core saturation used to regulate input pulses
2846581 Transistor Pulse Generator Circuit, Hendrik Volkers (Philips), Aug 5 1958, 331/112 ; 331/148; 363/18- more reliable transistor osc starting
2886706 Blocking Oscillator Pulse Width Control, SC Rogers (Bell Labs), May 12, 1959, 331/112 ; 331/148-
concerned with overcoming variations due to transistor parameter variations when using transistor saturation
2903677 Timing Track Recording Device (for mag drum clock), D. L. CURTIS (Hughes), Sep 8 1959, -
uses center tapped mag head as transformer of blocking osc a key idea is recording a clock signal on the drum rather than using a toothed wheel or trying to use an external crystal
2908870 Generation of Very Short Microwave Pulses, Clyde D. Hardin (Army), Oct 13 1959, 331/87 ; 331/148; 342/202 -
tube type blocking oscillator directly drives magnatron for high resolution radar\
bifilar wound ferrite toriod with low inductance (low capactance)
output X-band 10 ns pulses 150 Watts  (800 Volt drive pulse impedance matched to the magnatron)
2964716 Displacement to Frequency Converter, N Berman (United Aircraft), Dec 13 1960, 331/113A ; 331/181-
Angular position of an input shaft changes magnetic properties of core changing frequency of oscillator
similar oscillator topology to the Royer inverter oscillator, i.e. balanced design multiple feedback windings.

3185939 Generator with Blocking Oscillator Controls, John I. Moss, May 25, 1965, 331/52 ; 331/112; 331/172; 331/47; 361/203; 607/71- muscle stimulator

Neon Blinkers

Before semiconductors Neon lamps were used to make blinking lights Neon blinkers were the thing to have.  I had one in the early 1960s on the shelf above my work bench.  It consisted of a 90 Volt dry battery and series resistor and a cap in parallel with a neon bulb.  It would blink about once every couple of seconds.  It was on all the time (today you might say it was on 24/7) for some number of years.  I'd bet a dollar to a donut that the designers of the LM3909 had one of these.

Next to the Neon Blinker was a glass jar with the label "IITYWYPAPITJ".  When visitors came to my room they usually would ask "What is that blinking light?" and my response was: if it tell you will you put a penny in the jar.

Starting late September on the PIC List there has been an on going discussion about how to make these based on problems encountered with the lamps not going out, but instead staying on.  Once this happens the cap can not charge up and it quits blinking.

One solution to this is presented in  U.S. patent 2714692 (USPTO, Pat2PDF, Google) Portable Electronic Identification Light, W.D. Nupp, J.M. Rosen, Aug 2 1955, 315/232 ; 315/241R; 315/245; 340/908.
I have redrawn the circuit (without changing any connections) to make it clearer and reduce the file size.
The main circuit is B charging C1 through R1 causing the 3 Watt Neon bulb PL1 to flash about once per second.  But PL1 does not self extinguish.
So the 1/4 watt neon bulb PL2 (which does self extinguish) is driven  at about 10 CPS set by R3 and C2.
When PL2 fires it reduces the voltage across PL1 enough to cause it to extinguish allowing C1 to then charge etc.
patent 2714692 Fig 1
There are some possible extensions of this circuit.

Speed Test

Bob Blick's speed tests showing how LEDs and laser diodes turn one (very fast).

White - Phosphors

This is typically used for white light flashlights.  These combine a very bright blue LED with a phosphor that's yellow.

There is also a white phosphors "cap" that can be installed over a blue LED to provide diffused white light. JKL Whitecap -

White - Red, Green, Blue

Nichia America Corporation - Nichia, model NSTM515S, true RGB, $20-$24 apiece
These are a specalized single package that can generate many different colors.  But they only make sense if you do not have room to use seperate Red, Green and Blue LEDs.  By using seperate LED you get much much more light output for much much less money.

Numbers & Characters

2024 April 11: YouTube: A STRANGE old Electromechanical "Flip Dot" Display. How Does It Work?, 7:10 - single digit, 4x6 Dots, motor driven with cams. Totalizator.

The 7-segment LED can produce the numbers 0, 1, 2, 3, 4, 5, 6, 7, 8, 9.  In addition the letters for hexadecimal numbers A, b, C, d, E, F.
The rest of the alphabet is really not doable in a way that someone not trained can recognize, although some English words can be displayed that anyone would recognize (HELLO bOb), so for some applications 7-segment characters may work, but not where a comprehensive vocabulary is needed.

LED displays take some power to operate.  As the character height goes up the number of LEDs per segment also goes up and with it the power needed.  A 6 digit clock hh:mm:ss can easily draw over a watt with one inch high characters.  Managing the power is an issue.  One approach is to multiplex the display.  If each segment is only driven for 1/10 of a display period and it's current is 10 times higher the apparent brightness will be about the same.  But the on time needs to be small compared to the thermal time constant so needs to be in the micro seconds and during testing or any foreseeable malfunction the on current must not be on all the time or something will get smoked.  Another approach is to use a commercial multiplexing display IC (Maxium).  These are popular with Basic Stamp applications that are not fast enough to multiplex directly like can be done with a PIC.  Yet another option is to use a DC driver IC (Allegro).  These are programmed using a serial data and clock stream and are cascadable.  Each chip can drive 16 segments so it takes a number of chips.  A big advantage is with DC drive there are no worries about burning out something and the LEDs are more efficient with DC than with pulsed drive.

The characters are slopped and there is a decimal point in the lower right corner.  If two 7-segment displays are put side by side with one upside down and with the decimal points facing each other the decimal points form a colon ":" which is the common separator in clocks, so you get hh:mm:ss.  Most 7-segment displays have a pin out that allows installing them either right side up or up side down without rewireing the ground connections.  So only the drive logic needs to be changed to accommodate the upside down digits.

The 14 or 16 Segment LED can generate all the numbers and letters and so can be used in general purpose applications where you want the self illumination and freedom of what can be displayed.  It takes more drive circuitry but you get a self illuminating display.

Another general purpose LED display is the Dot Matrix.  Common is the 5x8 in either common cathode or common anode.  These have an advantage when you want to scroll.  The problem with segmented displays is that you can only scroll the whole character and that gets too jumpy to read, but with a dot matrix display you can scroll one pixel at a time allowing the message to be read while it's moving.  There are commercial single line message boards available for under $200.  Matrix displays are also used for outdoor advertising or at ball games where a color image can be displayed not just alphanumeric characters.  These take a huge amount of power.

Above some small number of characters it's much more cost effective to switch to a LCD.  The Liquid Crystal Display operates on almost no power, if fact there are versions that do not require any power to maintain a fixed display, only drawing power to make changes.  General purpose LCDs come is different flavors just like the LED displays.  They include numeric only (using the same 7 segment method as the LED), alpha numeric, using the dot matrix method and a table of character fonts that's usually less than 256 characters.  Also graphic LCDs like for a laptop screen or modern cell phone with color camera.  Custom LCDs can have icons like a battery symbol or clock hands.

Flipping Dot

Another type of display is the flipping disk dot (Wiki).
Flip Dot 7 disk barThe stick is about 106 mm long and each disk is about 14 mm in diameter.
Note that a common dot matrix display (Wiki) is 5x7 dots and so if you stack 5 of these 7-dot sticks you can form any character.
This flip dot stick came from Alpha Zeta in Poland that has taken over the FP Electric Flip Dot products including these flip dot stripes.

These have a magnetic memory and so consume no power once they are set.  In the photo above you can see one of the 7 dots has been flipped, but stays flipped when power is removed.  These were used on bus destination signs but have been replaced by LEDs.  They also were used at airports for the arrival and departure display boards which made a distinctive sound when they were updated.  Still are used for highway road condition and status signs.

I think the factory flip time of 50 ms can be greatly reduced by using a high voltage drive and a series resistor, similar to the way teletype and stock ticker printers work.  A related but different drive method is instead of driving from a fixed voltage source a capacitor can be used where it is sized to deliver just enough energy (1/2CV2) to make a reliable flip.  This would be handy when a low voltage battery was used to drive the display at high speed.


 512395 Producing Illuminated Letters, James H. Rogers, Jan 9, 1894, 345/30 ; 200/46 - covers an 11 segment display using lamps as well as a matrix display. (note the patent date)
 683133 Interchangeable electric display apparatus, George Lafayette Mason, Mason Monogram, 1901-09-24, - lights in segmented display
1311118 Three-position annunciator, Wesley H. Geiger, 1919-07-22, - Railroad
2245493 Electric indicating means, Erich Nothe, Siemens, 1941-06-10, - aircraft cockpit On/Off indicator
2367299 Magnetic signal, Ernest S Mclarn, Frank A Leibe, International Standard Electric, 1945-01-16, - aircraft cockpit On/Off indicator
2388448 Indicator, Edward V Sundt, Purcell Joseph, Littelfuse, 1945-11-06, - aircraft cockpit On/Off indicator
2415452 Electrical indicating device, John A Taylor, Milton A Knight, 1947-02-11, - aircraft cockpit On/Off indicator
2585974 Electrical indicating device, John A Taylor, Milton A Knight, 1952-02-19, - aircraft cockpit On/Off indicator
3163949 Digital illuminated display device, Paul R Gley, Dependable Printed Circuit Corp, 1965-01-05, - 21 segment (almost a dot matrix).
3303494 Magnetically operated signs, Taylor Maurice Kenyon, Winrow Donald, Ferranti Ltd, 1967-02-07, 340/815.62 ; 335/266; 335/281 - dot matrix for sign
2959219 Control Apparatus, Hajny (Baso Inc), Nov 8 1960, 431/50 ; 137/66; 335/266; 335/281; 361/162; 361/210; 431/48; 431/54; 431/80 - bi-stable magnetic fuel shutoff
3042823 High Speed Electronic  Memory, Willard (IBM), Jly 3 1962 - matrix of gas discharge elements
3140553 Magnetically Operated Sign, Taylor (Ferranti Ltd), Jly 14 1964 - an earlier version of the fliping dot - to replace matrix of light bulb type sign
3540038 Multi-Color Single Axis Magnetically Actuated Display or Indicating Element, Taylor (FP), Nov 10 1970
3942274 Strip Module for Sign Element, Winrow (FP), Mar 9 1976,
3975728 Electromagnetic displays with resiliently mounted components, Winrow (FP), Aug 17 1976
4069480 Blanking circuit for electromagnetic display, Christopher Gordon Helwig, Ferranti Ltd, 1978-01-17, - dot matrix for sign
4156872 Display element write sensor, Helwig (FP), May 29, 1979, - H-bridge drive of display element and read back


It turns out the the IR remote that controls my Blu-ray player turns on the lights in the ceiling fan when I press "Return" so I wanted to investigate that and see what would be needed to use a single "Universal" IR remote instead of the stack of remotes I no use.
Also the order of turning on the various electronics boxes is important, for example:
TV/Monitor first, then the audio/video receiver, then the blu-ray player to make the digital copy stuff happy.

This is a way to read the codes generated by IR transmitters as typically found in remote controls for electronics, toys, and ceiling fans.
It has the ability to both receive and transmit.
Available from: Dangerous Prototypes, shipped from Seeed Studio

Getting it Going

1. Download the firmware package and expand the ZIP file.
2. Plug the unit onto a USB cable and when the New Hardware window opens and asks for a driver (there is no driver needed) you need to browse to the location of the firmware package and inside that the inf-driver folder.  Then the OK button will go from gray to usable, press OK.
3. Open Device Manager (under My Computer / Properties / Hardware) and the COM port folder.  Now unplug the Toy and notice which COM port turns off, in my case it was COM 18.
4. to check if the Toy is working open Hyperterminal and point it to your COM port.  Take the defaults, and click on the phone icon to take it off-hook.  You should see:  DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DETECT!DET
5. Download and extract the WinLIRC program and run the .exe.  It will fail because there's no .cg (configuration file), but press continue and then select the USB IR Toy and your com port, then click CREATE CONFIG.  You should get a DOS window.  Follow the instructions using a remote.

6. After following the instructions in the DOS window that involve pressing buttons on the remote and exiting the program, open the config.cf file.
The file for the Hiaku Fan follows:

# Please make this file available to others
# by sending it to <lirc@bartelmus.de>
# this config file was automatically generated
# using lirc-0.9.0(IRToy) on Sat Aug 03 16:27:02 2013
# contributed by
# brand:                       ..\config.cf
# model no. of remote control:
# devices being controlled by this remote:

begin remote

  name  ..\config.cf
  bits           16
  eps            30
  aeps          100

  header       8970  4574
  one           491  1733
  zero          491   626
  ptrail        506
  repeat       8972  2311
  pre_data_bits   16
  pre_data       0xFF
  gap          108514
  toggle_bit_mask 0x0

      begin codes
          power                    0xC03F
          FanUp                    0x0AF5
          FanDn                    0x827D
          Whoosh                   0x20DF
          LightUp                  0x42BD
          LightDn                  0x8A75
          Sleep                    0xCA35
          Timer                    0x02FD
          Clear                    0xF807
          LightOnOff               0xC23D
      end codes

end remote

XyloBand (Wiki)

The audience for the Biden announcement the night of 7 Nov 2020 had light sticks that changed color all at the same time.
One way that might work is based on the Xyloband technology.
YouTube: BigCliveDotCom: Inside an illuminated (RGB) Xylo Band as used in concerts -

Xyloband Hacking - hand held controllers (HTX) -
9318043B2 Systems and methods for coordinating portable display devices, Dennis Chang, Mobbers Inc, 2016-04-19, - mentions prior art in the description, including Xyloband.  This patent is about coordinating mobil phones in a venue like a stadium so that an image can be generated.

Xyloband - What's inside one -

The light show that was part of the fireworks was done by quad-copters flying in formation.  It takes a swarm of hundreds of them to do this with precision station keeping.


There are a couple of things involving LEDs with the name Firefly.

Bio-mimicry device

The idea is to mimic the light flashes of an actual firefly.
6851208 Simulated firefly, Timothy L. Carter, Spartus Technology, 2005-02-08, - set of 16 Solar powered LEDs. - D580074 Firefly light emitting diode (LED),
8928227 Light emitting bio-mimicry device, Thomas John Padula, Autumn Collett Cardone,

Firefly Wind Direction indicator

 sensitive to 1"/second.
8291759 Fluid flow direction detection, Thomas Center Galley, Arnell Jean Galley, Harbinder S. Pordal, James F. Yoder, Intrinsic Minds, 2012-10-23, - a resistor at the center is heated and any wind will blow the plume of war air outward from the center.  There are two rings of eight thermistors that sense the direction the plume of warm air is moving.

7971478 Wind sensor, Harvey Harrison, Richard Paskowsky, Harrison Eng, 2011-07-05, - Dual (to get polarity of wind) crossed hot wires gives speed and direction.


Agilent (HP) - LEDs, Newark Electronics for small qty buys -
All Electronics - LEDs
American Bright Optoelectronics -
American Opto Plus -
B. G. Micro - LEDs
Best Hong Kong - eBay - Nov 2007 135 Cd White, 300 Cd Green 10 mm LEDs
Bivar, Inc. -
Brock's LED Flashlight Links -
Carclo-Optics - High Brightness LED optics - (CTP-COIL  is US warehouse the products are magnifying glasses)
Color Kinetics - C series can lights with 16 million colors and based on many RGB type white LEDs. US patent 6016038: Multicolored LED lighting method and apparatus
CMG Equipment - LED flashlights and one with a PIR switch to come on when movement is detected
Cree Research (includes Nitres, Inc.)- silicon carbide (SiC) based products including low current and high brightness LEDs
Data Display Products -
Dialight Corp - assemblies for trucks, signals, etc.
Don Klipstein's LED & Light Bulb Pages - LED Main Page -
Electronic Goldmine -
Everlight Electronics - LEDs and other electro optical products
Fraen - Optics for High Brightness LEDs - Future stocks some
Gerrys PIC Based Electronics Projects - RGB and other LED projects
HB Electronic Components, Cina - accepts PayPal
HDS Systems - rugged LED light
Hosfelt Electronics - LEDs
Infineon - lamps
Joule Theif - origional (as far as I know) version of single cell LED circuit
Kingbright - lamps
L2Optics -
LCK-LED - high brightness and others, on line store
LED Corp - White LED in conventional flashlight bulb base & flashlights
LEDdynamics - LuxDrive -
LED Museum - small qty sales -
LED PWM Control using a PIC
Ledtech Electronics, Inc. -
Ledtronics - lamps and indicators
Leotech Electronics - lighting & signs
LighThings - dealer in many brands of LED products
LSDiodes - good prices on very high brightness LEDs, individual LEDs, not digits or characters
Lumex - LED, displays, etc.
LumiLeds - traffic lights, automotive, signs by Philips
LuminArt - a sphere whose color can be changed by turning a knob and brightness by another knob. How big, power source?? about $190
Marktech Optoelectronics - N. America technical sales for Toshiba -
Mouser -  King Bright Multi-Color Lamp T-1 3/4  LF59EMBGMBW -
Mule Lighting -
Nicha -
One Stop Displays - OLED
Opto Technology - LED cluster lamps in heat-sinkable TO-66 packages
Para Light Electronics - lamps & displays
Philips - LumiLeds - Luxeon
Photon Micro-Light - keychain LED lights, many models trade color, brightness, battery life
Polymer Optics - plastic LED lens
Quantsuff's Circuit Page - single transistor blocking oscillator & two transistor single inductor boost circuits, combined FET & bipolar blocking osc circuit
Radio Shack - long list of LEDs
Roithner-laser - large selection of LEDs covering a wide range of wavelengths
Sloan Company -
The LED Light - lamps, flashlights, 120 VAC lamps, more
Toshiba. Hit the "Products List" selector and select optoelectronics. LEC chips
toyoda-gosei - LEC chips

Back to Brooke's PRC68, Products for Sale, Alphabetical List of Web pages, Flashlights & Patents, Electro-Optical GadgetsElectronics, Home pages

page created 16 Sep. 2000.