Inside an LED Christmas light bulb. (Or, How LEDs work using 120V.)

[rebelrider.mike]

I bought these bulbs years ago to replace the incandescent ones on my outdoor light strings.


They go bad way less often than the traditional lights, but I have had a couple kick the bucket. I decided to open up a dead one just to see what's inside. This one was full of water (cause of death) so I drilled holes to drain it first. Then cut the decorative cover off.


I wasn't expecting another covering over the LEDs, and this one was full of water too. So I chopped the base in half to see what's in there. Looks like next time I can just crack the plastic part off the base and save my self some time.



Inside was a tiny circuit board, plastic insulator, and muddy water. I thought the circuit board was actually pretty cool.



On one side is an MB6S bridge rectifier, and a resistor. On the other is a capacitor and another resistor. So it looks like a very basic AC to DC converter with a smoothing capacitor. I guess the two resistors are to help limit current?

Anyway, I put about 8V directly to the LED board's positive and negative, and got 3 of the 5 LEDs to light up.


The little LED board was in pretty bad shape, so I de-soldered the LEDs and 4 of 5 actually still work. They're now in a box for future projects. I looked up the rectifier chip online and discovered it's good for 600V, 0.5A, so I tried running 120VAC to it. I got 106VDC out. So these would need to be put in a string of maybe 50 in series in order not to get too much current through each bulb. Makes sense I suppose because they're designed to be on a 50 bulb series circuit.

So I have a neat little rectifier board to play with. I can't put it on my little 50V oscilloscope, but maybe I can find my old transformer and get some lower voltage for it. Would be fun to see how flat the DC voltage is.

 

[daromer]

The resistors and cap are to lower the voltage. A common circuit to run diodes on AC.

 

[rebelrider.mike]

I've been considering how to make some 120V LED indicator lights and remembered this old thread.

First off, I took apart all my dead C9 LED Christmas light bulbs to try to reverse engineer them. I did ok with the schematic:

I'm having trouble with the math though. I've been using Ohm's law to try to work out voltage drops, as well as the modified version of Ohm's law to figure out the resistor value needed for the LEDs: R=(V total - Vf LED) / A LED. I was able to measure 2 of the capacitors, but none of the resistors survived. Fortunately, their values are written as 102 and 105. I don't know the purpose of the capacitor. Does it somehow help lower voltage or is it just a noise filter? As cheap and small as these things are, it seems odd to add a capacitor there when there is none to smooth the rectified output.

The only LEDs to survive in all my dead bulbs were the three I found 2 years ago. Turns out they're still in a jar on my desk. (I don't clean often.) I was able to measure them at 2.4V, 200mA. I feel like I should be able to do all the calculations that I need to understand what's going on.

The whole reason I'm trying to reverse engineer this is to see if it would be a good way to make my own indicator lights. So far, I have a big resistor that drops the voltage from 120V to 2.2V, but the resistor loses over 3W to heat. This bulb runs 5 LEDs on only 0.5W, so there's got to be something going on here that I don't understand.

 

[Korishan]

Noise suppression, yeah. And make note that's a ceramic cap, not an electrolytic, which is polarized.

 

[rebelrider.mike]

I took another crack at it today. Double checking the resistor values I discovered that what I thought was a 5k resistor was in fact a 1M resistor! I thought that was not possible, so I found a multimeter with a 2000k setting, and measured 1MΩ. I was then able to find on the board a tiny trace connecting the resistor across the L and N connections. So I guess the 1M and the capacitor are both noise suppression.

I found a C9 string and with 22 bulbs going at once, the string draws 13.75V, 0.21A. Now I've finally got a real idea of how much power and current each bulb draws. 1.165W and 0.01A.

Now I really have no idea why this circuit works. A 1kΩ resistor should not be enough to drop the voltage sufficiently to run 5 LEDs on 120V. And I should be seeing over 6W of power used, and I'm only seeing about 1W. I definitely don't know what I'm doing.

 

[rebelrider.mike]

Big Clive to the rescue:

View: https://www.youtube.com/watch?v=Q23uh7AjjXw


At 18:38 he puts together a capacitive dropper to run an LED on 240V without a high (ish) wattage resistor. I don't understand it yet, but this has got to be how the LED bulb is working. He also mentions that LEDs aren't great diodes for AC because they tend to have a high reverse current leakage. Or they start to leak reverse current at low reverse voltages. Something like that. Anyway, a second diode or rectifier is recommended. Good to know.

So there may yet be a lit LED at the end of this tunnel.

 

[rebelrider.mike]

I believe I finally have the answer. Searching around, I found the equation C=1/2π*f*Xc.
C is the capacitance in Farads (I think), f is the AC frequency, and Xc is the reactance of the capacitor (measured in Ohms).
The reactance in this case is the resistance that would normally be calculated for the LEDs resistor: R=(Vin-Vf)/A where Vin is the starting voltage, Vf is the LEDs forward voltage drop, and A is the LED current.

I've already seen some disagreement as to whether Vin should be the AC voltage (as in 120VAC) or the DC voltage after the rectifier (as in about 169VDC).

I've been using a program called LTSpice to try to simulate the various circuits I've been attempting to understand. Unfortunately, I suck at using LTSpice, so my results aren't reliable. But... In this case it seems to be working. Here is how the C9 LED bulb circuit turned out:

With no smoothing capacitor, the DC Volts and Amps Jump from 0 to their desired specs and back, but the values are looking like I figure they should.

Since that seemed to work ok, I did two more simulations based on the LEDs I want to light. One using a resistor, and one using a capacitor:


This is rather encouraging. I think I will see if I can find some components to put together and see if it works in real life. I have a pocket oscilloscope that I barely know how to use. But I should be able to read the voltage oscillation on it.

 

[rebelrider.mike]

I found an interview with the creator of LTSpice that gave me an insight into why I hate the interface so much. Basically, it's written by a programmer for other programmers. Since I hate programming, it makes perfect sense. Fortunately, I also discovered that I don't have to use parts from the lists provided in the program. I can instead add generic "ideal" components and then add what they call a SPICE directive to indicate specific characteristics of the component I want. For example, I can finally tell the program that I want an LED with a voltage drop of 3.3V. It took me a while to figure out the proper syntax, but I finally got it working.

While I was fiddling around with that, I wanted to measure the actual voltage drop in the simulation, so I added a 10Ω resistor between the LED and the 0V reference. I was surprised to see the difference it made in the behavior of the current. I guess I accidentally rediscovered what people called an RC filter. Also, the voltage drops in the simulator were exactly the same as the ones I calculated. So that makes me feel better about using the simulator.


Turns out I don't have any of the parts I need, so I have to wait for an order before I can see this in real life.