LM3414HV: Blows out if 1N4003 put in series with the positive power input line

Hello Sid,

If you place a diode in series with the input what happens after the switch on the LM3414 opens? When the LM3414 switch closes the current in the inductor ramps up. When the switch opens the current wants to continue to flow which uses the freewheel diode (which is returned to Vin). If the source is open it will cause Vin to rise at the rate of whatever capacitor is charging. A schematic would help but you can also watch Vin at the IC and see what it is doing.

Best Regards,

Hmm...I hadn't thought of that, but it makes a lot of sense.  Also goes to show how such a small thing can have major implications.

My schematic is below, with the diode marked (Dx), but not in the circuit.  Outside of a TVS (not tight enough) or a zener (don't know that it could handle much wattage @ 62v), I'm really not sure what to do.   A large part of the problem COULD be the capacitor I have on the LED output to reduce flicker from the PWM, as it's the same size as the input filter capacitor. 

  - Maybe a much smaller output capacitor (or bigger input capacitor) would help?
  - Perhaps there's a better place to put the diode?  C4 (100uF) has to be across VIN/GND to prevent chip blowup from inductive spikes on power up, but maybe there's some way to isolate the main power input from C4/C1...?

(edit to properly upload schematic picture)

Hello Sid,

Are these voltage transients part of the source?  Is the voltage source the problem in your design?  Have you tried this on the bench with a solid voltage source?  If there is a lot of inductance in the source, such as long leads, this is a different issue you have to deal with.  I calculated the energy in the inductor and if the voltage would rise far enough to damage the IC and it should not happen.  100 uF will not change much voltage when the inductor energy is dumped into the capacitor.  Can you look at the input voltage with an oscilloscope when these transients are happening?

Best Regards,

I most definitely have "long leads"--we're talking over 50 feet of 12/2 wire (with a regular light switch and two magnetic [read: inductive!] circuit breakers) to the 51.2v battery bank.  (51.2v nominal; it ranges from 50-56vDC.)  There are a number of big loads connected to the battery bank (through breakers), but none of those were running when I was testing my LED driver module.  Oh, and my 44v LED strip is actually about 8 feet long (two 4-foot "fluorescent" LED tubes that I rewired to 44v)--would that wreck the inductance calculations?

When I was first trying these chips with a 50v bench power supply (plus a foot or so of alligator clips), they were instantly blowing out upon powerup until I put the 100uF filter capacitor (C4) across the input.  After over a year of installed use, I haven't blown any of them out--that is, until I tried adding the 1N4003 in series with the input lead.  I thought that that should decouple the LED driver from anything before it, but apparently not.  If I connected my little module to a solid bench power supply, and added the 1N4003 diode to the positive line, I'd be willing to bet a tube of LM3414HVs that the chip will blow in 0.5 seconds.

I really can't "look at the input voltage"--yes, I have a digital storage oscilloscope, but with the 1N4003 diode added, the LM3414HV blows up half a second after powerup.  And without C4 (without the diode, too), it blows out instantly upon powerup even with a bench supply.  I'll have to purchase more of these chips for any further tests, as I don't want to blow out any of my functioning lights.

Hello Sid,

If you can watch the input voltage (at the IC) during turn-on it would be helpful otherwise you are trying to troubleshoot a problem without knowing what it is. Even if you damage the LM3414 you need to see what voltage is damaging it. It may be Vin but it may be something else.

Best Regards,

I just noticed in the LM3414 datasheet where the absolute maximum ratings are 67v on VIN and LX for 500mS.  And coincidentally, it has taken between 500-750mS to blow the chip out. 

Looking at the application schematic, though, I see no way that the "kick" from the inductor could get around to C4 and VIN.  After the internal switch turns off, bypass diode D1 will be forward biased, dumping all the inductor's energy into C5 and the 44v LED string.  There is no way C5 could get over 67v, as the LED string is directly connected to it.  For that matter, the only way LX could get past 67v is if D1 is blown or not working (as it will clamp the potential at LX to VIN +0.6v).  I've checked D1 with my meter, and it seems to be working fine.

Found a couple more LM3414HVs in my parts box, but I'll have to get the oscilloscope another time.  I might just leave well enough alone and give up trying to figure out what's amiss...