Thoughts on LED strip light

[Jeff1946]

After reading project Cree (https://www.eevblog.com/forum/projects-designs-and-technical-stuff/project-cree-the-led-bench-lighting-project/ ), nice job, I began to think about doing something like it.  I have replaced all my outdoor walkway lamps with home made LEDs.  Did it simply with LEDs, series resistor and full wave rectifier.  So wondered if I could do the same for lighting strip.

My idea:  make it from strings of Duris E-5 leds.   These are 100+ lumens/watt, about 5mm long, and less than $0.40 for 100 quantities.  I have a few of these and have hand soldered to PC board to test.  You need to bake above 100C before soldering.  Anyway I was thinking of making a strip of these in series on a PC board(s) about 2cm wide.  Put about 1 per cm to allow space to solder and spread the heat.  They radiate with 120 degree light so should give pretty good uniformity.  I would drive about 28 leds in series directly from rectified 120 VAC.  Obviously need to shield from touched due to live AC.

I developed a spreadsheet (attached -- Note had to rename from xls to txt for this forum, you can reverse this) to calculate resistor size, currents etc.   Because of non-linear nature of LEDs you need to do something like this to make sure you don't exceed the pulsed power limits of the LEDs. Although power is wasted in the resistor, it is the only power wasted other than a small amount by the full wave bridge.  I probably would use several small resistors placed along the PC board.

Any comments welcome.

[IanB]

You may wish to consider a dedicated constant current LED driver.

If you put many lamps in series with a high voltage supply you have an issue that a high voltage can appear anywhere there is an open circuit or high resistance. This potentially can produce arcing and localized failures. Furthermore you can get cascade failures where a single failure leads to increased current in the remaining devices, accelerating failures in a chain reaction causing rapid self destruction.

With dedicated constant current drivers you have some built in over current and over voltage safeguards leading to a more reliable installation.

Note that the series resistor with LEDs is designed to approximate a constant current source. But this only works if the voltage drop across the resistor is significantly larger than the voltage drop over the load. If the resistor has a small voltage drop relative to the load the constant current behaviour is removed and the system becomes unstable. For instance if the series resistor normally had 20 V across it and it became exposed to 120 V the power dissipation would go up 36-fold. You would have a significant fire risk (although a suitable fuse would mitigate this risk). Although this might seem like scare mongering, I really wouldn't want to power low voltage devices directly from the mains like this.

[Kibi]

I too would advise against running the LED's directly from rectified 120V. IanB has highlighted the risks.
Rather run 10 of them in a string ~ 32V across the string and employ a current limiting circuit to limit the current to 180mA. If the supply voltage is only a a couple of volts over the forward voltage of the string then you won't drop too much heat in the current limiter. It is very important to limit the current when driving LED's because variations in temperature greatly affect the current that flows through them. The circuit that I used in my design is very simple but effective.
You could run more LED's in series per string, but you wouldn't want the string to have more that 50V drop across it, a) for safety and b) a current limiting circuit starts to get more expensive if you need higher voltage devices in it. Below 50v you can pretty much use jelly-bean parts to build a good current limiter.