Hi,

I am using SN74LV4053A as the bridge(SMBUS) between battery charger LTC1760 and Microcontroller. Recently we are facing the issues with this MUX IC, the IC suddenly stops working and there will be the permanent short between VCC and ground. This short will not resolve with resetting the device and also causing the failure of its biasing regulator(5V) IC as well.

The below image is our implementation. Please let me know if there is any issues with it

 

Given the total failure mode it seems that the device is operating beyond the maximum power dissipation. I suggest a careful reading of the specifications relative to power handling capabilities relative to the actual input power total for all of the inputs.
( This is a guess, based on what I see in the posted circuit, )

 

That was my thought as well, but inspection shows no obvious path that's loaded with any significant current flow unless there's a short elsewhere externally to this 'snippet'. But get a thermal camera on it and check. Also check pull-up values to ensure they are what they claim to be.

What is the purpose of muxing the charger to some static lines, and where do the diodes off U4 go to - just curious, can't see any issue with that.

 

My thought was that possibly even with no short circuit at all, the current was excessive. My recollection is that the analog multiplexer is not a power device at all. And the surface mount package can handle even less power.

 

My thought was that possibly even with no short circuit at all, the current was excessive. My recollection is that the analog multiplexer is not a power device at all. And the surface mount package can handle even less power.

Agreed, this is a 25mA per channel mux rather than 250mA like some of the Maxim and AnalogDevices products. But here its only switching i2C signals, not power, as far as I can see.

 

The problem sounds like latch-up. See the Wikipedia article on the subject. It explains how it happens, but it also tells how the problem is traditionally avoided. I would assume this technique is implemented in this chip.
The datasheet says "Latch-Up Performance Exceeds 250 mA Per JESD 17". Not sure what that means.
I would look at the chip inputs and outputs for any signals that exceed the supply rails.

 

Quite possible on lines from charger, esp if board layout routes them near switching power circuits. A 100R or 1k in series with the relevant pin(s) near the mux might help.