TPS3813K33-EP: RESET! Window exceeds spec value at cold temperatures.

Hey Michael,

Would you be able to show your schematic?
Is there a large capacitive load?

Thanks,
Daniel

Hello Daniel --

Due to the nature of our business, I cannot share the schematics.  I can tell you that the output of the WDT is lightly loaded, going only to the Reset Input of our processor. 

This occurs in approximately 5% of our systems (overall).  When we have a failing WDT, we replace it with one from stock, and it passes test after the rework has been completed.

It does come in waves, so, I would suspect that it is Lot Code based.

Mike

Hey Mike,

I can't do much without seeing a schematic.
The watch dog timer window was tested with CL = 50 pf and RL = 1 MOhm so I was wondering what was used in the final design. I would expect the window to increase with a larger capacitance and lower resistance.

Thanks,
Daniel

Good Morning Daniel -

I can tell you this:

We're using the TPS3813K33MDBVREP module with:

WDT = 0V

WDR = Vdd

RESET! is driving one line to a uProc with a typical load capacitance of 5pF.

Fallout for this is approx 5%

On failing modules, the RESET! pulse is borderline (300ms) at -40C and failing (~314ms) at -55C.

When we replace the TPS3813K33MDBVREP module the system passes the low temp test.

The modules do have PLGH stamped on them.

Is this enough to go on?

When you test the modules, are they tested at -55C to spec, or are they tested at a higher temperature with an "off-spec"?

Thanks;

Mike

Hey Mike,

You said the top said PLGH?

The top side marking for this device is PLGM.

Can you double check that?

Thanks,

Daniel

Sorry about that!

It is marked: PLGM.

Mike

Hey Mike,

Well at least its the right part!

The part is tested for -55 C, so the last thing I can think of given the information you have given me is the input to the device may be slow.

Do you know the rise/fall times of the inputs?

Thanks,
Daniel

Hey Mike,

Did this issue resolve itself?

Thanks,
Daniel

Hello Daniel -

No it didn't.

We're sending failing modules to an independent lab to get confirmation of the failing mechanism. 

Additional failing modules are being sent back to TI thru our board-house.

Mike