2
\$\begingroup\$

I currently have this layout on my PCB:

1

As it appears, the grounds from the different components are connected to each other and then connected into the supply's ground.

In the context of amplifiers, an answer from one of my questions explains that small signal grounds should be separated from large signal grounds to prevent distortion because of the signals mixing within the PCB traces. I did that on one of my amplifier boards and it worked.

Questions:

  • What happens if I separated all of the grounds from the components and directly connected them instead into the supply's ground?
  • Is this technique better than just separating the grounds into small and large signals before connecting them into the supply's ground?
\$\endgroup\$
2
  • 3
    \$\begingroup\$ What an ironic name and question. \$\endgroup\$ Commented Sep 28, 2018 at 11:47
  • 1
    \$\begingroup\$ @HarrySvensson Yeah. I don't fully understand myself. \$\endgroup\$
    – Ground
    Commented Sep 28, 2018 at 11:51

2 Answers 2

4
\$\begingroup\$

The main reason is that the PCB tracks and wires don't have zero impedance.

If you think about it, higher currents will produce more voltage across a given impedance ( your ground wires) and this causes the grounds at these points to wander around in terms of voltage. So the point is no longer at ground which can cause noise to appear in some configurations.

What happens if I separated all of the grounds from the components and directly connected them instead into the supply's ground?

Is this technique better than just separating the grounds into small and large signals before connecting them into the supply's ground?

  • This is called star grounding (although there will be some local ground connections) and was common in the days of point to point wiring. In practice you will do both of the above, as the local circuit sections will need a local ground which is then star connected to the main ground, IE you dont do it on a component to component level if that makes sense...

Best practice is to usually have higher and lower currents connected to a single point, and have a separate ground connection for small signal sections.

\$\endgroup\$
4
\$\begingroup\$

Here is how experimentalists at diyaudio.com achieved low-nanovolt gnd-trash levels needed for 100 dB dynamic range and SNR, for vinyl record playback using moving-coil cartridges with 100 microvolt output levels.

schematic

simulate this circuit – Schematic created using CircuitLab

Notice the THREE regions of this GROUND PLANE:

LEFT region for the first JFET, a lownoise device, and all the input terminations and coax shield and bypass caps for the VDD (heavy R+C filtering);

CENTRAL region with two purposes

.... (1) grounds for the RIAA frequency compensation [50Hz pole, 500Hz zero, at voltage levels 30dB (31X) higher than from the movingcoil cartridge],

.... (2) the attenuation, IN THE NARROW REGION of the GROUND PLANE, of currents that might circulate between the PowerAmplifier coax shield-return and bottom of C6. This output signal current --- 1 mA --- times 1 mohm, produces 1 uV of ground voltage GRADIENT. This gradient may occur only on the right side of the PCB near the pulldown R7 of the output buffer JFET;

RIGHT region, where high ground currents flow (1 mA signal current to and from the power amplifier) and from the two 470 uF bypass caps on the 35 volt rail. These currents must have a DC path to/from the first gain stage, but AC currents must be strongly attenuated.

The long thin central region provides that attenuation.

BUT MORE WAS DONE. Here is how those two PCBs (audio Left and RIGHT) were powered.

schematic

simulate this circuit

\$\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.