Failing LEDs on temperature control board 3

[jumbleboy]

Hello

I'd be very grateful for some help on a problem we are having with a circuit board. The circuit is a PIC12F675 based temperature control board. The circuit uses a capacitor power supply to provide the power. The PIC reads a thermistor and triggers a triac when the elements need heating. It all works fine, but we are having a high failure rate on the indicator LEDs. After a while they start to flicker erratically and then soon after stop working altogether.
The 4 LEDs are in series, and when I test them with an external power supply they all still work, but one will draw around double the current of the other 3. This is enough to pull the voltage down making the LEDs flicker, though the PIC still runs happily. I have tried testing the boards on a continuous flash for days, and have tried to see if turning the circuit on and off many times will trigger the problem, without success. Does anyone have any idea of what we could do to solve this problem?

The PCB design is here -

https://www.dropbox.com/s/c7rtxvhg2e55sx8/CircuitBoard.jpg

and LED specs here -

https://www.dropbox.com/s/mxqkrm53q1wv8i3/L-C150KRCT.pdf

Thank you very much
JB

 

[VEBill]

"...a capacitor power supply to provide the power... The 4 LEDs are in series, and when I test them with an external power supply they all still work, but one will draw around double the current of the other 3. "

Not sure I understand the LED test. A better LED test would be to apply the correct current (often 10mA), and see if it lights up. Maybe measure the Vf if you're paranoid.

High voltage spikes from the power line causing one LED to short out?

 

[OperaHouse]

That type of failure is typical of over current. I have a few LED that I've done that too and they exhibit the same symptoms. You could put a resistor in the emitter of the transistor an make it pseudo constant current. That is a pretty novel power supply for a computer guy. The LED may be seeing a reverse voltage from the power line. Never been one to believe that 5V reverse spec on a LED. In all my testing the reverse voltage is about 170. However, line spikes can easily be higher than several LED in series. Peak turn on currents could also be large depending on what time of the cycle they turn on at.

 

[itsmoked]

You may be hammering them with reverse voltage. You need to shunt any reverse voltage around all the LEDs in higher voltage AC circuits. Two ways to do that. Either double up the LEDs with ones in reverse. Or put a diode 1N400x across the whole string. The LEDs would cost more but would be more informative. Try it on the test bench.

Otherwise I agree with Opera. You're over-currenting them and his suggestion to use the transistor as a regulator is a pretty good one. But before trying that I'd want to put a scope across the R3 and watch the polarity and magnitude of what's happening. As you probably know [!]you need to be very careful with your scope setup to prevent making the scope chassis live while doing this measurement![/!] If you don't understand this you would be safer to add a small resistor in Q1's emitter leg and measure across that with your scope since one side would then be neutral referenced.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

That type of failure is typical of over current.

This^^^^^^

I really don't like that circuit for LED control...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[OperaHouse]

Looks like the OP has disappeared. At least he was lucky to have such an obvious problem this early. From the two resistors in series on the input cap, it looks like this was going to use surface mount. A production product and not a hobby circuit. If he needs all that light the number of LED should be doubled to halve the current.