experience with TPS54618 anyone?

Does anybody here have experience with this chip: http://www.ti.com/lit/ds/symlink/tps54618.pdf , page 1?

This circuit is inside a tablet and is extremely sensitive to cold. I figured out that , if I touch the VSENSE input via a 10MΩ (!) resistor it shows the same behaviour as when the circuit is cooled down. I didn't measure the output voltage yet but the effect is that the display freezes, so something happens to the output voltage at this moment. A reset solves the problem.

Is this circuit supposed to be that sensitive?

I didn't do any measurements yet, component density is very high and I wanted to ask first if someone experienced similar issues.

 

The SENSE pin is basically the same thing as an input terminal of an op amp, and touching these points with anything can lead to weird behaviors, as they are high impedance points that are exceptionally good at amplifying noise as well as signals.

My guess is when touching that point you raise the voltage at sense... which will decrease the Vout voltage that may cause a brownout failure requiring that reboot. That's just a guess, a working theory of why touching there does that. A voltmeter on Vout would help confirm this.

Now are far as the cold temperature problem goes... marginal solder joints can open with temperature changes, as can marginal capacitors, in fact any marginal component is liable to fail with a large temperature change. That's one reason why high reliability equipment is put in an over to cycle the temperature.

One method used forever is to get a can of cold spray (sold at electronics stores such as Radio Shack), spray various areas of the board till you find a susceptible spot, then try to find the one part that has the trouble.

 

That's why I used a 10MegOhm resistor. The voltage at Vsense should actually not be very much influenced because of the voltage divider at the output ( I don't have the values :-/). They are for sure much less than 10M. .... Or maybe not. Although the voltage divider resistors themselves don't react to cooling spray.

That's exactly what I did. Unfortunately the component density is too high to be able to spray one by one.

What is needed is a schematic