7
\$\begingroup\$

When designing the PCB layout for a DC-DC boost converter, one does not place a ground plane under the inductor, to prevent induced ground currents. Also, for SMD MOSFETs, the drain (switch node) usually serves as a heat sink, so, since it should be small but sink a lot of heat, it makes sense to pour it on both sides, connecting them with vias. Which means no ground plane underneath the switch node (I'm using an IRFH5025, if you'd like to visualize what the package looks like).

However, I am not clear on whether it is beneficial to place the ground plane under the rest of the MOSFET, specifically the gate trace, sense resistor, and current sensing filter.

On one hand, this might limit magnetic interference from the inductor. On the other hand, it would increase MOSFET gate capacitance and provide another path for return current, right underneath the quite sensitive current sense trace.

Here's a picture with the area I'm considering marked in orange:

enter image description here

So, should I leave this area clear of the ground plane?

\$\endgroup\$
3
  • 1
    \$\begingroup\$ Power MOSFET gate capacitance is so high that the ground plane won't add anything significant - you can cross that off the list of concerns. There may be other valid concerns though. My own feeling is that more copper to move heat away from the area is a Good Thing, but I'm interested to hear other views. \$\endgroup\$
    – user16324
    Jan 11, 2013 at 13:29
  • \$\begingroup\$ More copper won't really move any heat away, as I'm talking about the ground plane. Most heat gets dissipated by the MOSFET through the drain (and its "powerpad" SMD connection), which is the switch node, not ground. \$\endgroup\$
    – Jan Rychter
    Jan 11, 2013 at 13:39
  • 1
    \$\begingroup\$ Consider that gate inductance is proportional to the area enclosed by the gate trace and the return current path, and this can place more of a limit on switching speed than extra capacitance (which is probably small compared to the MOSFET, and easily overcome by a strong gate driver) \$\endgroup\$
    – Phil Frost
    Jan 11, 2013 at 14:28

1 Answer 1

Reset to default
3
\$\begingroup\$

In many high-power AC/DC designs that I've worked on, we generally divide the ground planes up so that the power return is kept clear of the control return. The two planes are connected together at a single point. Where possible we try to shield as much control as possible with the control return plane, even gate drive signals.

Bear in mind that a good MOSFET driver sources and sinks a lot of current, so as long as your gate trace isn't too lengthy it's quite difficult to induce enough signal on the connection to interfere with the MOSFET drive.

\$\endgroup\$
3
  • \$\begingroup\$ I was more worried about interference with the current sensing network. I can't really have a star ground in this design, but power return is shorter than control return in my case. I wish I could post a picture, but SE doesn't allow me, as I'm new. \$\endgroup\$
    – Jan Rychter
    Jan 11, 2013 at 14:01
  • \$\begingroup\$ @JanRychter post a link to the image in your question, and someone else with enough rep can embed it properly. \$\endgroup\$
    – Toby Lawrence
    Jan 11, 2013 at 14:44
  • \$\begingroup\$ After some thinking and re-reading this answer I decided to take it as a "YES" answer to my question — so I extended the ground plane to cover the area marked in orange. \$\endgroup\$
    – Jan Rychter
    Jan 16, 2013 at 9:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.