# September 16, 2018

Did some experiments today with some stepper motors.  The only library I found that can control them was designed for use with a CNC machine.  It takes G-code.  That is more complected than I had hoped but I have worked with G-code before—when I made a giant fricking robot with vision and lasers for work.  I was able to do some basic control but never got the speed control as turned as I would have liked.  The stepper motors, for my first project, are just going to move a laser mirror around.  So I think I will look for a more basic control library.  Otherwise, I'm just going to have to strip this one down.

# September 15, 2018

Last week I wrote about an experiment I did using a 100 watt white LED COB for a strobe light. This weekend my 100 watt blue LED COB arrived and I was eager to see if it improved the freeze effect of my strobe experiment. The short answer: no.

I started to have my doubts about the strobe duration I read about. 1/1000th of a second, or 1 ms, was what I was using. While I don’t remember exactly where I saw the number I had been searching for number used for photograph strobe lighting. So I went in search of other numbers. One product I saw was an LED strobe used for high speed photography. They listed light pulse duration between 20-1000 µs. My setup can do timings this fast so I decided to give it a try. The results: 100 µs light pulses stop a spinning fan blade.

The problem is, even with a 100 watts of LED light, a 100 µs light pulses isn’t much overall light. The product I was reading about uses 2,000 watts of LED light—20 times the light output I have available.

Time for a little math. I was using 1 ms light pulses at 100 watts which is 0.1 joules for energy. By switching to 100 µs pulses, I have 0.01 joules. I would need to dump in 10 times the power—1000 watts—to get the same amount of light energy. I could do a bank of 10x100 watt LED COBs, but there is one other thing to consider.

LEDs output is limited by their temperature. The temperature increases because there is a voltage drop across the diode in where the energy is turned into heat. Thus the current that goes through an LED must be low enough to allow the heat to escape. A 100 watt LED with the correct heat sink is rated for continuous usage. We are only pulsing the device for a 100 µs at a time, and at most once every 100 ms. This gives a duty cycle of 0.1%. I’ve read “Most high-brightness LEDs have an effective upper limit of 6X to 8X over current capabilities, based on a typical 150 microsecond (µs) pulse width and 0.15% duty cycle.” That means my single 100 watt LED should be able to handle between 600 and 800 watts for 100 µs. I would only need 2x100 watt LED COBs in order to produce 0.1 J light pulses.

# 1 comment has been made.

## From Noah

#### September 19, 2018 at 8:14 PM

It's very interesting that there is only a certain limit to the tradeoff of duty cycle % to driving current. I'll bet that there are some wickedly complicated heat-transfer equations out there somewhere!

# September 14, 2018

The is a picture taken on east side of Park Street right at the Pheasant Branch Trail.  A huge landslide has taken place as a result of the flooding.  It will be quite sometime before the trails are reopened.

# September 13, 2018

Pumpkins at a stand along County Q.  I haven't yet seen any giant white pumpkins like I had last year, but it isn't October yet so I will standby.

# September 12, 2018

Fellow cyclist riding by on my commute to work.  The standing water is flood water and has been slowly receding.

# September 11, 2018

There is a yard along Farwell Drive just north of Warner Park.  For the past several years I've noticed I ride by this garden with the most lovely colors highlighted by the setting sun.

# September 10, 2018

This is likely one of the last sunrises I will see from this location for the awhile.  The sun will not be above the horizon on substituent rides into work.  Perhaps when daylight savings time takes place in November I will see the sunrise from this location again, but losing more than 2 minutes of daylight each day that may not happen.  It is nice to have something that only takes place periodically like this.  It is a reminder that fall is on the way.

# September 09, 2018

This is a project I have been thinking about for several days.  It started to come together today when the remainder of my parts arrived.  I picked up a basic strobe light a couple weeks ago but wasn't impressed by its performance.  Using an oscilloscope I could see the pulse given to the LEDs was about 20 ms in duration.  This just isn't enough to really get the stop motion effect like one obtains with a xenon strobe.  After doing some reading I believe the target time for a good strobe effect is a 1 ms or shorter light pulse.
LEDs are extremely fast and should be able to switch in a matter of nanoseconds.  So all I needed to do was switch them that fast.  Problem is, most modern power supplies that drive LEDs use an inductor based switched-mode power supply.  They don't always like being switched on and off with exacting precision.  To get around this I figured I could make a linear current regulator.  That involves an op-amp, a FET and a couple resistors.  Honestly, I'm too lazy to put this together.  The easier way is to use a power supply, a big capacitor, a FET and a high current resistor.  That is actually quite easy, and what I have done.
I could have used the LEDs from the strobe light I bought, but I wanted more power.  With only 1 ms to produce light I would need something very bright.  So I picked up a 100 watt LED COB (Chip On Board).  This is actually an array of LEDs on a single large chip.  Normally you need a heatsink and a fan because of how much power these devices dissipate so I ordered a heatsink and fan.  Turns out the heatsink and fan I picked up are worthless as it doesn't fit the chip.  However, I am only giving the device a very short duty cycle so I figured for testing I could get away without one.
The circuit is very simple.  I use an Arduino Nano to turn an output on for 1 ms, and then pause for some duration determined by a potentiometer.  The output drives a FET.  This drives power through the LED and into a current limiting resistor.  I use a bench power supply to set some desired voltage.  It is crude, but completely functional.  The LED COB wants 2.6 amps of current at 38 volts.  I put in a 2 ohm current limiting resistor so set the power supply to 42.2 volts.  That gives the LED all 100 watts and wastes 13.8 watts through the 2 ohm current limit resistor.
The results: close but not quite.  I found I could lower the voltage to around 20 VDC and still light the array, but very dimly.  This was nice because I could look at it.  White LEDs work by using a blue LED and phosphorus.  The blue light is absorbed by the phosphorus and re-emitted at various wavelengths producing white light.  Unfortunately, the phosphorus isn't nearly as fast as the diode.  It has a long afterglow of tens of milliseconds.  While I have a fast attack on my strobe light I have a long decay.
I have ordered a 100 watt blue LED COB.  There should be no phosphorus and thus it should be faster.  I can try the experiment again and see what I end up with.
Despite not getting the times I wanted this strobe light is significantly better than my other LED strobe.  It is crazy bright and will likely be functional.  Still, I want to see if I can do better.

## From Noah

#### September 10, 2018 at 9:03 PM

Looks good! Is the decay time of the phosphorus in your white LEDs enough to ruin the strobe effect?

It's very interesting to see the shortcuts used to replicate 'white' light as we see it.

## From Andrew Que (http://www.DrQue.net)

### Middleton, WI

#### September 12, 2018 at 9:37 PM

Sadly, I think the phosphorus decay is taking away from the strobe effect.  A good strobe light should freeze frame the scene.  I was flashing the strobe at a fan, but the blades were still blurry.  A 1 ms flash should is 1/1000 seconds and that should have been enough to freeze the blades during a flash.  I even ran 500 microsecond pulses which is 1/2000 seconds.  Although there was less overall light, the blur didn't seem to change.  The real test will come when I get the blue LED.

# September 08, 2018

State Street in the early morning hours.

# September 07, 2018

Early morning twilight on my ride to breakfast.