Grounding Principles

RE: Grounding Principles

Jens-Michael Gross — Thu, 20 Jun 2013 12:29:53 GMT

Hi Bruce,

thanks for bringing this topic up.

Personally, I always recommend to other people (mainly in the Microcontroller world) to separate digital from analog ground, but also current-carrying analog from reference analog ground.

Even if you have a ground plane, keeping the reference ground separate and free from any significant currents will usually improve signal quality. It is an astonishing and not-so-commonly known fact that even on a ground plane, currents flow not equally distributed. They prefer current paths where a current already flows in opposite direction on another plane/wire above or beneath, causing not evenly distributed voltage potentials on the ground plane and even increased crosstalk. Ths is one of the reasons why a ground plane below a crystal should be isolated form the common ground plane (except for one connection point, of course)

unfortunately, most layout programs only know one ground and joind it where there is the next possibility. Even when doing a manual layout, it is often difficult to trick the program in keeping the traces separate. SO I usually define three different signals (GND, AGND, RGND) and manually connect them near the supply GND reference point by using a special 'connecting device', mainly a tiny piece of manual copper.

You're right, grounding is science, but implementing the scientificly calculated grounding paths is an art of its own.

RE: Grounding Principles

Guenther Klenner — Wed, 19 Jun 2013 13:37:52 GMT

Hello Bruce,

Thank you highlighting another aspect of good grounding. Beside all abstract views additonal two topics impact grounding of bypass capacitors: thermals and vias.

- Thermals on capacitor's grounding pad act like a resistor and inductor. They are needed to ensure good soldering. Routing wires close by may reduce the number of thermals easily from 4 to 3 to 2 to 1 and result in increase resistance and inductance the other way.

- Vias: Two vias of 12mil are better than one via of 24mil. Best is arranging 3 small vias around capacitors grounding pad close to thermals (below, left and right) as there is normaly no space for a 4th via inbetween capacitor's pads.

Good grounding is not easy, not at all.

Guenther

RE: Grounding Principles

Anish Joshi1 — Thu, 30 May 2013 15:07:39 GMT

Thanks Bruce for elaborating.

I think now I understood what you were saying.

I read a few articles and found them really informative and interesting!

Thanks for posting..

RE: Grounding Principles

Bruce Trump — Tue, 28 May 2013 17:58:35 GMT

Anish-- 1) Yes, my discussion was primarily focused on the analog portion of a system or an analog-only system. You could infer, however, that a digital ground connecting at points A, B or C could be a very bad situation.

2) These analog grounds (A, B, C) should come together at a single point. They must necessarily come from some other location to perform their appointed task. As they lead into this common joining point, they should not be tapped along the way with other sources of noisy or distorted ground current. They should make a solo journey to the joining point.

3) Ferrite beads are used to add a small series inductance, allowing DC current to flow freely while offering a series impedance at high frequency. They might be used a variety of ways to control noise or provide "decoupling" between two portions of a circuit at high frequency. One possible use might be in the joining of a digital and analog ground.

Hope that helps-- Bruce.

RE: Grounding Principles

Anish Joshi1 — Tue, 28 May 2013 16:05:47 GMT

Hello Bruce,

Thanks for sharing this grounding concept. To an amateur like me, it does seems like black magic.

1. Firstly I would like to ask-This is all going on inside the analog section of the PCB, right?

2.Do you mean to say that the signal(i/p and o/p) ground should be connected by a single point to

the analog supply ground? Or you mean to say that the trace connecting this point to supply ground should be

physically away from this point?

3. Where exactly should the ferrite beads be used? Only to separate analog and digital grounds?

RE: Grounding Principles

Georg Curnutt — Thu, 23 May 2013 11:09:47 GMT

Thanks for posting this. Its been a long time since my Pops showed me things like this. I keep hearing the word Diode in my head. Diodes were the first things my Pops explained to me in electronics. And Grounding, he tried to show it to me like antenna theories. You did a great job of cramming a whole lot into a short paper.