LTC4418 power path switch-over failure

Question

I am experiencing a very strange issue with the attached LTC4418 power path controller design. I have one instance, where the controller does not switch off path 2 when path 1 when it is enabled. The circuit is looking for 12 (or 24V) from a car battery as the primary source, but has an internal 7.2V battery for backup. On this particular board, path 1 is energized, but FET for path 2 stays on - resulting in power being sourced directly to the battery. The battery is not rechargeable, so this is very bad. 
 To top things off, if I connect a power supply instead of the battery, the issue does not occur, and the circuit works properly. The LTC4418 has been replaced and the issue stayed the same, the battery control FET (Q2) has also been replaced - also with no change in the behavior. Attached are two scope shots presenting both cases. Please advise if you have any ideas to develop this further. 
 Thanks! PDF

JRH · Accepted Answer

Figured out my issue. FET Q304 in the schematic was bad (shorted source to drain). When a power supply was connected to the battery (power path 2) input, it would not sink current from the power path 1 supply, so the PWR_SYS (VOUT) rail would rise and the circuit would appear to work correctly. When a real battery was connected to power path 2, the power path 1 supply would instead try to drive the battery through the faulty FET Q304 and the body diode of FET Q303. After taking more scope shots of VG1, I realized that there should not have been enough threshold voltage to enable these two PFETS.

Electronic Design Services - PartnerZone

LTC4418 power path switch-over failure

Top Replies