This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LP3982-Adj latches in dual power supply config with UCC384-Adj if neg. power applies first

Other Parts Discussed in Thread: LP3982, UCC384

I designed a dual power supply for +/-4V6 which is powered from +/-5V0. Here I use LP3982-Adj (pos) in combination with an UCC384-Adj (neg). +/-4V6 supplies a few OP-Amps, total current is about 15mA. This works fine if +5V and -5V are applied at the same time or if +5V is applied before -5V. In that case, if -5V is applied some ms before +5V, the LP3982 latches up what means, it doesn't generate any output voltage. I did some measurements to see what happens: the negative rail reverse powers the LP3982 output via the load. This makes the LP3982 latch up. This effect is reversible after +/-5V supply is turned off, it doesn't damage the device. The effect is reproducible.

Here some details:

+5V first off second
-5V second on first
Operation normal only neg latched
IN(2) +5V0 -0V482 +5V0
OUT(1+4) +4V65 -0V596 -0V591
SET(5) +1V323 +0V115 +0V115
CC(6) +1V223 -0V483 -0V447
FAULT(8) +4V6 -0V108 -0V170

Regarding the absolute maximum ratings of the LP3982, the out-pin is driven out of spec. Even though this is clear, I'm wondering why this standard application and situation doesn't work with the LP3982. As this problem just popped up after the circuit design is finished and we're in series production, I'd like to keep the LP3982 and solve the problem with - hopefully - some minor changes on the PCB.

For protection, I put a schottky diode at the output. If I use a small one (1N5711) there is no big change in the reverse voltage at output pin: -0V520. This doesn't prevent the circuit from the latch-up effect.  If I put a bigger diode (1N5817), the reverse voltage is clamped to -0V250. This works well, no more latch-up. This is just an information to understand the latch-up mechanism, I'd like not to add an additional device to the circuit.

Does somebody have a smarter idea?

-> Does toggeling the SHDN(7) pin help?

-> Is there a way to delay the attack time of the UCC384-Adj.?

-> Is there a simple way to do power - sequencing, maybe to use the FAULT pin?

-> why is LP3982 so sensitive? UCC384 doesn't mind the same kind of "mistreatment"

-> maybe it's risky to ask in a Ti forum: would MAX8860, which is 99% compatible, work better? Any experiences???

Every comment welcome...

Christian