0
\$\begingroup\$

How to make sure you have "adequate" ground in a self-enclosed portable circuit? I read that a good ground ensures steady signals. But, if you run a circuit from a battery it is not connected to anything else.

What do you use for "adequate" ground that is able to handle your circuit? Do you add more ground metal like metal sheet, larger PCB traces, an elaborate PCB ground-plane? How do you determine/calculate what is adequate?

\$\endgroup\$
5
  • \$\begingroup\$ I think you are mistaking earth for ground \$\endgroup\$ – DKNguyen May 16 '20 at 2:11
  • \$\begingroup\$ I thought that in some cases those are the same, and larger ground can absorb noise and help with circuit stability if no earthing is available (low voltage mAh circuits or with higher frequency ). \$\endgroup\$ – TommyS May 16 '20 at 2:19
  • 1
    \$\begingroup\$ Not a larger ground, but one with minimum impedance to everything it is connected to. Tight, small loops like a plane. if it has to go everywhere. \$\endgroup\$ – DKNguyen May 16 '20 at 2:21
  • \$\begingroup\$ I think I see it now. So if the impedance is too high that could be seen by the circuit as no ground or may not be adequate enough to provide a negative return path. \$\endgroup\$ – TommyS May 16 '20 at 2:22
  • 1
    \$\begingroup\$ More like the currents makes a voltage drop via the parasitic inductance which makes noise and EMI. \$\endgroup\$ – DKNguyen May 16 '20 at 2:24
1
\$\begingroup\$

Usually it comes down to breaking down the current loops of your circuit, this generally only comes in to significant play if your circuits are sensitive to small shifts, or very high speeds, where cross talk is significant.

If your circuit is sensitive, e.g. a high resolution ADC, you want to make sure other parts of the circuit can't effect the value your reading, this may mean isolating a portion of the circuits ground, and connecting it to the rest of the circuit at a single point, or by putting a slot in a ground plane to keep larger / noisier currents flowing around that part rather than through,

Traces on a PCB has resistance, so any current flowing through it will create some voltage, depending on how you have routed things, this can cause issues,

In the case of higher speed signals, when the gap between the signal and its return path (sometimes ground, sometimes a differential signal line), changes, it acts a bit like an antenna, either emmiting signals, or risking coupling in others, to this end you try and make sure it has a ground return path under or adjacent to that trace for the entire length of the signal to prevent this.

Personally to reduce some of this when doing layout, I will begin by grouping up parts into functional blocks, if its say some precise analog block, if I can keep most of them sharing the same ground / reference, then it does not matter how that block shifts in regard to the rest of the ground, it just means I need what is measuring the output of that analog block to not vary too much,

To this end, you can mix and match planes, start grounds, and trees of ground traces, its about keeping one part of the circuit from interfering with another, and there is no hard and fast rule of thumb about which is the best in all cases.

\$\endgroup\$
5
  • \$\begingroup\$ You mentioned: "...if I can keep most of them sharing the same ground / reference, then it does not matter how that block shifts in regard to the rest of the ground.." Question: on the actual circuit board, or in the circuit design phase what would be considered the same ground? For example, if I have an analog and a digital side of the circuit, don't all grounds return to the battery negative side into one common ground point? \$\endgroup\$ – TommyS May 16 '20 at 2:31
  • 1
    \$\begingroup\$ Lets say I have a block of 4 chips, e.g. some op amps, they measure some tiny signals from an external connector, so any shift in there reference point, ground for this example would offset those tiny signals and make it useless, each of those chips have almost no supply current, so over a few centimetres of trace will only create offsets in nanovolts. the connector and the op amps all have there grounds connected to a single point in that block of the circuit, and then that point is connected to the main ground plane, \$\endgroup\$ – Reroute May 16 '20 at 2:38
  • 1
    \$\begingroup\$ Everything connected to that point may shift up and down by what is passing through the plane below, but relative to the signals, it remains the same. \$\endgroup\$ – Reroute May 16 '20 at 2:39
  • \$\begingroup\$ So given all the sub-circuits are equal but their input levels differ, it's important to make sure that the signal from each separate sub-circuit's load to the common ground has the least amount of impedance in order to make sure their individual singal-levels are not skewed? \$\endgroup\$ – TommyS May 16 '20 at 2:55
  • 1
    \$\begingroup\$ In this case its more to prevent 1 part of the circuit shifting the DC value of other parts of the circuit. say an op amp circuit with one of the input pins tied to ground, what it measures is relative to that grounded pin, if what its measuring is also relative to that pin, then nothing changes on the output, if you have say 5A being switched to ground elsewhere in your circuit, that might cause the ground at that part of the circuit to shift up and down by tens to hundreds of millivolt as that ground path has resistance. which changes the absolute value, but not the relative. \$\endgroup\$ – Reroute May 16 '20 at 3:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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