AVR Freaks

Also note the antiparallel diode is not there to balancethe current, but to protect the led. Most leds have a reverse breakdown of around 5V. Things will work without the diode, but not for long!

The capacitor in question is X2 rated.

It takes a really good spectrum analyzer to see the DC harmonics :wink:

JC

The leakage of the diode will allow reverse voltage on the led. It will eventually die. Not a problem at lower voltages, but with 110V and transients, the led is going to have a hard time.
Best to put the diode across the led and size the resistor accordingly.
Note- you are three years too late to the party!

This method has been used for many years with no problems. Rectifier leakage in picoAmps is not going to bother a pulsed LED in this circuit. 

Don't use a 1N400X diode. Use a low leakage one like a 1N4148. And consider that the Light Emitting Diode and the Rectifier are both part of the same series circuit and combine their characteristics to make this application feasible.  Pulsed optoelectronic devices.

I was going with the line voltage that would be going to the front panel for the neon indicator. The line voltage would still be going there and the neat little package with a 1N4148 rectifier and a resistor and LED would take the place of the Neon indicator. The design is in compliance with rules for pulsed optoelectronic devices.

One person suggested a resistance of 10,000 Ohms but the LED would be very dim. That person was treating it as a DC circuit instead of a pulsed circuit that is only on for a short time. The brightness and duration has to be such that it tricks the eyes. That requires a 2500 Ohm current limit resistor. LEDs can be pulsed at currents considerably higher than their average dc rating to provide the same brightness as perceived by the eyes.

A good optoelectronics application manual is a good addition to the library. That is what I used. 

Right, it seems odd that the peak reverse voltage rating never caused a problem in the design.

Not odd. It's a numbers game. Recently i had to debug a product problem. The product had sold in large quantities over a 15 year period with supposedly no bad reports. I had a hard time convincing the engineer that designed it that there was a design issue. Once we started asking questions, we found that it was common for the distributors to just replace the failed item and throw away the failed unit. Thus we had a significant percentage of failures that never got reported or investigated. Millions of units over 15 years! It took one service tech to do the wrong thing and cause a major problem that brought it to light.

I did this many times, for high power white LED's on linear bulb replacements. One good example is linear retrofits, in which the balast is used to limit the string current, just as it does with a standard flourescent tube, but since there is no inductive kick requirement to start the bulb a capacitor works just as fine (XL and XC).

A Single antiparalel diode is not the best aproach. The best way to achieve this is a bridge rectifier with the LED on the output (and perhaps a series capacitor to smooth the output) and the series limiting capacitor in series with the AC input. This ensures there's no DC Biasing and both half-cycles are simetrical. The bridge has to be rated for 400V, at least.