I've since found that when the device won't turn off with EN/UVLO at 0V, if the load current is increased to the point where it cuts off, then the device will begin being able to be switched off with EN.  It's as if it gets itself into a confused state and remains on when EN is low. Circuit attached.

Hi Robin,

Can you help me with following.

• What is the ambient temperature?
• Are you seeing this issue from the very first power up? 
• How are you measuring this? Are you probing near the device? 
• Any bouncing at LDSTART pin or any other pins. Any clock type input to any pin? 
• Is this repeating with multiple devices?
• Do you have our EVM (to reduce PCB related variables)

Robin Crane said:

I recently saw that you have released a new equivalent" device, the 259830 which is rated to 18A rather than 15A.

They are from same device family with lot of common ip. The thing is, this is not an expected behavior for TPS25982 also so I cannot say that this will occur/not occur at your end with TPS25983. They are p2p so you can evaluate this without any additional PCB cost. 

Best Regards,
Arush

Hi Arush

It's at normal room temperature - ie around 23C.

At first power up it works OK  - I have to cycle the "INST_PWR_OFF_L" signal from the processor with a load of around 1Amp before it enters this mode.

The processor is monitoring the 2S L-ion battery volts and driving the INST_PWR_OFF_L line and we depend upon the e-Fuse Output going OFF to preserve the batteries when the Volts drop below about 6.2V  - ie above the threshold that is set with the potential divider R45/R48.

Measurement of the EN/UVLO pin with scope probe on TP8 , the output via some output filtering through on-board connector.

LDSTART pin is to 0V via 0 Ohm resistor R46.

I have seen this on multiple devices - I replaced the TPS on a board that I could get into the fault state, and it still does it with the new chip.

I don't have the EVM but I have followed the layout guidelines to the letter - I am an experience board designer so know all the potential issues.

It can take many "cycles" of EN/UVLO toggling to prompt the issue.

I noticed that the 25930 devices  have had the LDSTRT circuitry removed - and if you think that the issue may be around this (?) then please let me know.

I tried to get samples form you of the 29530 but it appears that I have to pay for them...?

Regards

Robin