TPS549D22: gerber check

Andy Hsiao and Gary Teng 

Thank you, I was able to find U3126 in the BRD file.  Generally the layout it pretty good, but a few places it could be improved:

1) Snubber ground connection

The purpose of a snubber is to provide a controlled ground impedance for high frequency signals in the 50-200MHz frequency range.  The current ground connection has a very long path back to the source of the synchronous rectifiers, which are at the exposed pad under the TPS549D22.  After passing through the Capacitor and resistor, the nearest ground via is out by the output capacitors, adding a lot of loop inductance and limiting the effectiveness of the snubber.

Rather than connecting the snubber to the output ground, I would recommend a pour under the R6746 resistor with several vias to the ground layers to route back under R6746 and C7756 would reduce the series inductance and improve snubber effectiveness.

Widening and filling the connection from C7756 to R6746 would also help.

2) If backside high restrictions allow moving C7748 and C7749 to the backside of the PCB and directly under C7751 and C7750 will help reduce input impedance.

3) Moving C7754 (2.2nF to GND) to the backside of the PCB under the VIN pins with vias close to the VIN pins will generally make this capacitor more effective.

This will provide a parallel path with lower mutual inductance to better suppress ringing and reduce MOSFET stress.  Keep these vias far enough apart that the ground planes are able to flow between the vias to prevent slotting and blocking of thermal spreading.

4) Due to the high dV/dt transients on VIN during the turn-on and turn-off of the high-side MOSFET, I am generally reluctant to route the remote sense differential pair under the VIN bypass capacitors like this.  It can expose the remote sense to high dV/dT voltage transitions.  If possible I would recommend routing this pair on the far side of the VIN via array to reduce it's potential coupling to the high dV/dt transients on VIN.

Andy Hsiao 

That looks much better and I believe that will provide you with better performance, especially high-frequency and MOSFET stress characteristics.

One minor improvement that could be made from both a performance and manufacturing reliability point of view.  R6746, the Snubber Resistor.

1) The new layout is better than the old one was, but the resistor may suffer from skew issues during solder reflow do the different geometries.  I would recommend a small pour area around the pad connecting to C7756 so that both pads of R6746 are in pours.  This will ensure that the solder-mask swell expands both pads equally so that there is equal liquid solder surface tension on both ends of the resistor to prevent "tombstoning" where the resistor is pulled to stand vertically on one pad.  Make sure the pour does not extend to the C7756 pad, which would add the issue to the smaller capacitor which is even more susceptible to this effect.

2) I would also recommend moving the two vias at the end of the resistor to an area between the resistor pads, as this will help reduce the loop inductance presented by the resistor and it's snubber performance

I am having an issue creating the graphic images to demonstrate this.  I will add that graphic as soon as that issue is addressed.

Andy Hsiao 

As promised, here is the graphic showing the improved layout of the coper areas around the Snubber resistor.

Andy Hsiao 

Yes, that looks better.