I recently started making some LED drivers with the LM3414HV eSO-8, powering 3x 10W LEDs at 33v/664mA from a 55v supply. (Yes, I know that's not full power into the LEDs.) Currently, I'm using mostly spec parts (SDR1806-470ML 47µH inductor / 520KHz from a 39K resistor), 2.2µF/100v input filter capacitor, SS2PH10 diode, 4.7K current limiting resistor, etc. The only change I made to the circuit is a 10µF, 16v capacitor from VCC to ground, as I am hoping to power a PIC12F1822 for automated dimming control. (That has not been installed on my boards; there is no load connected to VCC, and nothing connected to the DIM pin.)
Basically, the driver works just fine if I gradually turn up the voltage to 55v, or if I put a 1-ohm resistor in series with input power. If I directly connect the LED driver to the 55v supply (i.e. like a power switch), the LM3414HV instantly burns out, shorting LX to GND, and 0.14v on VCC. That has happened with both chips I've tried--and I can easily verify with a multimeter that LX is indeed shorted internally to ground.
My question: The datasheet specifies a 1µF capacitor from VCC to ground. I'm suspecting that the use of a 10µF capacitor is overloading and burning out the internal 5.4v regulator, causing a meltdown of the driver MOSFET upon abrupt powerup. Can someone confirm/deny this?
If that is indeed the case, I'm assuming that I can use the 1µF capacitor per spec, and use a series resistor to a 10µF capacitor for the PIC. All I need is a power supply that can power the PIC (i.e. >2v, drawing less than 1mA) for 2 seconds after power is shut off.