Author Topic: Power ground and control ground  (Read 1148 times)

0 Members and 1 Guest are viewing this topic.

Offline alpachino10Topic starter

  • Newbie
  • Posts: 1
  • Country: us
Power ground and control ground
« on: September 08, 2016, 03:07:34 pm »
Hi,

I have a question on grounds in a circuit:

Should the power and control grounds be separated (isolated?) from one another in a power circuit (for example a buck converter)? What are the challenges and considerations of this matter?

I'd appreciate if someone could explain it. Thanks.
 

Offline ajb

  • Super Contributor
  • ***
  • Posts: 2721
  • Country: us
Re: Power ground and control ground
« Reply #1 on: September 08, 2016, 05:41:12 pm »
It depends.

Generally, with DC-DC converters it's vastly more important that all of the connections to the ground node be as low impedance as possible, which means a single ground plane to which the input, output, and any control signals are referenced, so it's not usually useful or practical to separate grounds from the converter's perspective.  However, when it comes to connecting that DC-DC converter to other circuitry you may want to do a star ground at the output of the converter (essentially running dedicated ground paths from the converter to different loads), but that's a design choice to be made based on the requirements of the other circuitry and the converter itself doesn't really enter into it. 

Generally you'd segregate the grounding paths within a system to prevent large ground currents from traveling through sensitive circuitry.  For instance, if you've got a precision analog section and a power section with large DC currents, those large currents will result in a voltage drop along the grounding path due to the unavoidable resistance of the PCB tracks (or wires), meaning that different points along the path will be at different potentials.  If your analog section makes contact with different points along that ground path, the differences in voltage will show up as an offset error in analog signals.  If you have large AC component in the power stage, then you have to worry about that coupling into sensitive analog nodes via stray capacitance between the power and analog nets.  In either case, making sure that ground currents from the power stage don't flow through the analog section will probably be a good idea.  Digital circuitry is generally going to be less susceptible to disturbance, but could still see problems in certain situations.

Principle challenges in star ground systems can include dealing with offsets when making measurements or sending signals from one section to another, and making sure that you don't inadvertently cause a ground loop between sections, which would defeat the whole purpose of star grounding in the first place, or could result in large currents flowing in places where they're not meant to.

Really, it comes down to understanding how the non-ideal aspects of real-world circuits could conspire to disrupt the proper function of a system.  In particular this means understanding that any wire or PCB trace will ALWAYS have some amount of stray inductance and resistance, and adjacent traces and wires will ALWAYS have some amount of mutual capacitance.  Separating ground paths/star grounding is one way of overcoming those factors that is suitable in certain situations.  Other options could include galvanic isolation (eg, optocouplers) between sections of a system, differential signalling/measurement, or just making the ground net as low impedance as possible (that last is partly why you almost always use full ground planes on PCBs).  All of these may be used in conjunction with star grounding depending on the application.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf