VCC net routing topology.

Go To Last Post
7 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Consider the attached picture. The highlighted net is VCC, at 3.3v. There are 5 transistors on the left side, that will be switching max 100mA @ 14.4KHz (one at a time), and under them is a quad opamp with a 4KHz low-pass. What would be best practice for routing this VCC net? Should I keep all junctions connected, or section the links at the marks?

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Would suggest avoiding loops on Vcc first of all. It's better to have the digital, power and analog Vcc lines in star configuration).
Only thing to concern imho is low pass op-amp, but usually these have good immunity for synchronous and eventually this is a matter of signal level.
Shall Gnd have good routing would just suggest cutting one top marked point plus connection from switch transistors to op-amp.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GND connections are solid, there is a full pour on top and bottom. You suggestion is a good one, to split up the opamp supply, but what exactly is the consequence of VCC loops? I understand current loops, but that applies to a return path. Wouldn't the connections lower the net impedance, increase the capacitance, and help filter the noise more?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I like to "star" out the VCC on the basis of where it's going. Sensitive parts are on a different branch of the star to high-current or noisy parts. In your case, I'd suggest 2 branches, one for all those transistors and the other for the op-amp.

Within a branch, I personally have no problem with loops. Sometimes it's the only way to connect up all the decoupling caps and such, depending upon where parts are placed.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Seems like you could widen those traces as well.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

VCC net is 20mils, on an outer layer I would only need 2.5mils width in 1oz copper for a 10'C temperature increase per linear inch...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Wider tracks have less inductance, it's nothing to do with current.

Leon Heller G1HSM