Trouble shoot LED failure

[bhavlo]

I am brand new to multisim, in fact downloaded the trial version yesterday.  I want to see if it will help me trouble shoot an LED failure issue that I have encountered multiple times.

The circuit is 120/240 Vac LED power indicator in a plug-in product.  The LED circuit is across the 120/240 Vac line which consists of a 1.2k ohm resistor in series with 0.22uf capacitor in series with a LED that has a 1n4002 diode in parallel with it for reverse bias protection.  The problem has been that on repeated plug/un-plug action the LED has failed numberous times.  There is about 20uf of additional capacitance from the rest of the product in parallel to the LED circuit, described above.  I believe that the transient response due to all the in-rush of current to the capacitance is killing the LED.  I would like to learn if I can model the transient response with Multisim so I can simulate a fix as well.

If anyone could help me learn how to do that with Multisim I would appreicate it.

[Sharanya_R]

Hello Bhavlo,

The following tutorial will help you configure a transient analysis in Multisim;

Configuring a Transient Analysis in Multisim

If you have any additional questions please let me know.

[johnsold]

bhavlo,

Simple circuit analysis will give some idea of what is happening.  The reactance of the capacitor will limit the current to about 10 mA at 120 V and 20 mA at 240 V.  Because the LED current is half wave rectified, the average LED current will be less.

The transient response at the time the power is applied can be calculated by assuming that the capacitor is initially a short circuit. If power is applied at the instant of peak voltage the initial current will be 283 mA = (240 * sqrt(2) / 1200). The time constant is 264 us. After one time constant the current will be down to ~37 % of the peak or 104 mA.

The manufacturer of the LED probably does not specify the transient response the way power rectifiers are specified.  I looked up one LED at random and it has a surge current rating of 1 A for a pulse width <= 10 us.  My guess is that your circuit is marginal under worst case start up transient conditions.

Lynn

[bhavlo]

Thanks for your reply.

I do understand most of what you descibed, what would you suggest to make my design more robust?

[johnsold]

bhavlo,

Fundamentally, you need to reduce the transient current.  The capacitive current limiter is nice because it has low losses or wasted power.  If you increase the series resistance, you reduce the instantaneous current at the expense of higher steady state losses.

Do the failures occur at both 120 V or mostly at 240 V?  How much current do you need to get adequate illumination of the LED? Does your device have a switch or jumper to change between 120 and 240 V or is it a "universal" input circuit?  Do you need to monitor the line voltage or could you monitor some regulated output voltage where the variations are not an issue? Can you afford to add some kind of current regulator circuit?

Lynn

[bhavlo]

The failures mainly occur with the 240Vac circuit.

10ma is plenty for good illumination

There is not a switch between 120v & 240v, the board just gets built different, a 330ohm & 0.47uf for 120V -OR- 1.2kohm & 0.22uf for 240V

Since I already have a PWB I am really kind of stuck with the existing design across the line voltage and can't really add regulation at this point.  The best I can do is change R & C values or tack additional parts on the back side of the board.  I am hoping I can find a better solution with those restrictions for now.

Thanks for all your help.

[johnsold]

Interesting.

I looked up the LED listed on your circuit diagram.  The peak current rating is 100 mA with a pulse duration of 100 us.

With your values I calculate a peak current of 514 mA at 120 V and 283 mA at 240 V.  The time constants are 155 us at 120 V and 283 us at 240 V.  So now I am not sure what is causing the failures. Lower peak current with a longer time constant as opposed to higher peak current with shorter time constant.

My first thought was to increase the 1200 ohm resistor but it already is dissipating almost half a watt so a higher value will require a resistor with a higher power rating which is probably not acceptable.  If the time constant is more important than the peak current, reducing the 1200 ohms to 700 ohms would match the tme constant used on the 120 V circuit and would give a peak current of 485 mA.

My "gut feel" says that is the worng way, but your data suggset it might be worth a try.

Lynn