2
\$\begingroup\$

I am designing the single phase inverter. The input of the inverter is around 35 Vdc, and output is 24 Vac. The output currents will be below 1A. I control the inverter with F280039C from TI on a daughter card. I have multiple ADC signals coming from inverter to microcontroller via op-amps. (currents and voltages sensing from inputs and outputs).

I was thinking to design it with separated power ground and control ground and joining together two grounds by ferrite beads. I highly rely on single-phase inverter design from TI. Then I start to reading about separation of grounds, and now I am not sure is it a good approach. I read some articles, for example on altium.com, but they mostly divide approaches to isolated (with transformer) and non-isolated designs. The inverter case is a bit different. Should I stay with approach from TI and use just ferrite beads or create a single ground plane and just separate components on the board?

My diagram and concept of PCB: Diagram - sorry for yellow color, I realized that after exporting from my eINK PCB concept

\$\endgroup\$
9
  • \$\begingroup\$ Welcome! Please post schematic or block diagram of what you are proposing. \$\endgroup\$
    – winny
    Nov 18, 2022 at 11:15
  • \$\begingroup\$ Hello and welcome. Just curious, is 24Vac correct? That reference design is 220Vac \$\endgroup\$
    – jonathanjo
    Nov 18, 2022 at 11:32
  • \$\begingroup\$ Unless I am wrong, GND and BUS_GND should not be wired together. Used in schematics DCH010505 are isolated regulators so, no wiring. ti.com/lit/ds/symlink/… \$\endgroup\$
    – Antonio51
    Nov 18, 2022 at 11:32
  • \$\begingroup\$ @winny I just added images. \$\endgroup\$ Nov 18, 2022 at 13:21
  • \$\begingroup\$ @jonathanjo yes I want to use 24V ac, and it's true that ref design is 230 Vac. I cannot find any 24V design. \$\endgroup\$ Nov 18, 2022 at 13:22

1 Answer 1

0
\$\begingroup\$

Separating grounds isn't a good approach. The problem is the return currents create a common mode voltage through the resistance of ground planes.

For 0.5oz copper, resistance is 1mΩ/sq(inch)

Lets say you have a 1A current running through that 1mΩ, you'll get 1mV of common mode voltage or 'ground bounce'. (this also only matter if the current is changing, because if it's steady then you'll see a constant offset which isn't a problem for most applications)

What about 1oz copper? We'll that's double the thickness and half the resistance so only 0.5mV

What about 1A through 5 square linear inches of 0.5oz copper? that would be about 5mV, for 1oz 2.5mV.

So if the design can tolerate mV, then your fine... uV not so much.

So calculate it! Find the biggest current, it will run through the shortest path and estimate the copper resistance and you can find how much voltage will be on the ground plane.

Another thing... you can reroute currents by rearranging components on the board so high currents do not flow back across sensitive areas. I have been able to do this with uV designs on boards sourcing a few amps of current with no isolation or ground separation.

\$\endgroup\$

Your Answer

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

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