Created by
Created May 25, 2012
Last modified May 27, 2012
Tags bridge-rectifier   led   led-array  


The lamp is made using an dead CFL body. The lamp is already in place & working very fine with brightness equivalent to 40W incandescent lamp.
The energy consumption is almost NIL, as the energy meter shows no movement. The operation is super cool, there is no heat generated even after over-night operation.


The circuit is simple. All the components ratings are mentioned in diagram itself.

Please feel free to give your suggestions.



There are two problems with your circuit:

First, you cannot run LEDS in parallel: they are current driven devices.

There has to be some sort of current limiting or defining component in series with a LED or a string of LEDs.

You are driving them from a voltage source: the 470nF cap acts as a high impedance so the source voltage across the LEDs is set by the zener which looks like a low resistance voltage source.

Therefore, the string with the lowest overall forward drop will draw a - possibly destructively - higher current than the other strings.

There will also therefore be a wide variation in brightness across the strings.

You at least need to put resistors in series with each string and drive them from a voltage significantly higher then the range of string drops you might expect allowing for diode variations and temperature related forward drop changes.

The second problem is that simply using a capacitor as a voltage dropper may not be considered safe in this sort of application. It works but it would probably be considered unsafe as a capacitor failure - or the result of it - may pose a serious fire risk.

There's a lot of commercial interest in LED lighting these days: have a look at ( and other semiconductor vendor's sites for hints and info about driving LEDs.

by signality
May 25, 2012

@signality I would like to tell you that this Lamp has been already prepared & is functional since one week. I keep it ON from evening until morning. The output seems equivalent to 40W incandescent lamp. The operation is super cool, I mean there is no heat generated by the lamp.

Thanks for mentioning the problems associated with this circuit. There will be serious problem in case of Capacitor failure. How about adding a Fuse before capacitor.

In the series-parallel arrangement, will it be good to add fuse in every branch? so that if an LED fails, then consequently all LED in that series connection will fail, hence to protect shorting, a Fuse can be inserted.

Please give your suggestions.

by k.rajnikant
May 27, 2012

There are three big problems with the design.

The diodes need to all be in series, in order to not have one string of the current. You may have lucked out, if all your LEDs came from the same batch, from nearby spots on the original wafer. In a real production run, the LEDs could be from different batches, with slightly different forward voltage drops, and then there could be significant differences in the brightnesses of the strings. Myou can fix this by putting all the LEDs in one series string.

The input cap is a big and unsolvable problem. In the real world, you have to design things soothe bulb can be in a house located on a mountaintop, between a vandegraff testing lab and a battleship welding company. Or in a house where the fridge compressor kicks out a 800 volt jolt back into the power line when it shuts off. That means you need a capacitor rated for 1200 to 2500 volts. Try finding one that is smaller than an adolescent kumquat and costs under $6.

The capacitor also draws almost pure reactive current. You can get away with that in a one-off device but power companies frown on millions of those devices hanging on their lines. Some countries have laws against that kind of thing.

It's cool that it works, but in the real world you'd probably have to scrap the capacitor input design and that just leaves you with the option of a small SMPS. Much more complex but no big unsolvable gotchas.

by arduinohacker
May 27, 2012

Leave a Comment

Please sign in or create an account to comment.

Revision History

Only the circuit's creator can access stored revision history.