LED's rarely if ever fail open, unless handling at manufacture breaks a bond wire; in this case it would be found at test. A more usual failure (still tiny) would be S/C and would result in less than a percent drop in light,insignificant in comparison to the 10% or more caused by failure of a HB LED. This failure is more likely due to the much higher current, die temperature and thermal cycling of these in comparison to the smaller LED's running at just 10's of mW's.
With enough paranoia at a pinch you could add a cheap 0.5c diode which would act as an antifuse across each LED and this will fail s/c under the HV supply to short out the open LED!
I could only partly agree with the statement of Robert; Boosting at rectified 230VAC mains means that the output voltage needs to be much higher than 325VDC. A number of LED's with a forward voltage of 350V or 400Vdc is not common. Next to this you also have the risk of the Christmas light effect. The chance of having one LED out of 100 LED's failing as an open circuit is higher that with 4-8 LED's. Single stage converter with high PFC and without large storage capacitor will result in light flicker. Based on the insights from Robert, I would prefer to use a non-isolated fly-back with medium volage LED's.
Please share your thoughts.
This idea of using a boost converter and low power LED's is the subject of two Patents/applications both UK and US. Including a semi(not quasi)-resonant technique, the design achieves as high as 98% driver efficiency, 0.993 PF and due to the low cooling requirements a life of 100K hours at 45C or 400K hours at 25C including long life (12k at 105C)electrolytic caps. There is 'no'(2%PP) flicker, even when dimming. I hope to license this to interested parties. email@example.com
I had a friend in the 80's that tried to make money selling those incandescent stick-on diodes. I recall his sales pitch was that the diode increases the life of the light bulb - maybe he pitched power reduction as well. Both of these claims might be true - I have no idea - but in the end you have a dimmer bulb output. That fact was probably not included in my friend's sales pitch!
kinnar, I don't know where you got your 25 Hz number but normally the number is around 70-80 Hz. It is somewhat dependent on the decay time of the phosphor though. I am assuming we are using blue leds with yellow phosphors. Easy check to know if a Christmas string is LEDs or incandescent, If you move you head back and forth rapidly here in the USA you can see single color leds strobe, I think this is caused by LEDs being driven for half of the 60 Hz cycle, still 60 Hz though. This should be even easier to see in countries that use 50 Hz although if one were to add the expense of full wave rectification this strobing would probably not be visible.
The diode in series was a bad idea then for incandescents, and it's a bad idea now for LEDs, although for very different reasons. Now, instead of starting with a 50/60 Hz sine wave with (hopefully) low distortion, the power supply is forced to begin with a half-cycle AC wave with lots of distortion--and for what possible effciency or other benefit?
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...