Ground plane or star ground?

[yakideo]

It seems that there are many different opinions of whatever to use ground planes (split or not) or star ground. The eb-ta2020 datasheet recommends splitting the ground planes. However after more research I found this site (http://www.hottconsultants.com/techt...gnd-plane.html) claiming the opposite. I must say that the more I look into this issue the more confused I get.

Could someone clarify this for me?

[AndrewT]

Hi,
simply asking the question shows you are not ready to design a ground planed anything.

Stick with good star grounding practice.

[Sebastiaan]

Dear,

It is always hard to say one is better then another. There is always a catch, and some upsides create downsides on the other side of the bridge.

That said, I am totally pro ground-plane. However what most people think of as ground-plane isn't often the right way. In many applications I see a ground-plane performed wrong or poorly.

If you think higher bandwidth and inductive wise, ground planes are superior to any star ground.

In my current commercial designs I use ground-planes all over, with high and lower power all mixed on one board. Never had any better performing amplifier with such a low self noise. You just must know 'how" to do it.

I collected many knowledge over the years from different "guru's" according this subject end eventually bite the bullet and start to design this route myself.

I just wanted to make a quick post now, but if there is a interest I am more then willing to share some insight and concepts that make designing with a ground-plane more likely to be successful.

Ps. And indeed avoid spitting ground-planes. If you have a good routing and component placing discipline, and you can virtually imagine where and how the currents will flow, this is not needed at all. If in any case you must split the plane for whatever reason. Never run traces over the "split" but make a "Bridge" where the two ground-planes join, and only run traces over this split. This way the Rf current loop will remain local and can find its return path underneath. That is the whole idea and meaning of a ground-plane to start with.

With kind regards,
Bas

[yakideo]

Originally Posted by Sebastiaan

Dear,

It is always hard to say one is better then another. There is always a catch, and some upsides create downsides on the other side of the bridge.

That said, I am totally pro ground-plane. However what most people think of as ground-plane isn't often the right way. In many applications I see a ground-plane performed wrong or poorly.

If you think higher bandwidth and inductive wise, ground planes are superior to any star ground.

In my current commercial designs I use ground-planes all over, with high and lower power all mixed on one board. Never had any better performing amplifier with such a low self noise. You just must know 'how" to do it.

I collected many knowledge over the years from different "guru's" according this subject end eventually bite the bullet and start to design this route myself.

I just wanted to make a quick post now, but if there is a interest I am more then willing to share some insight and concepts that make designing with a ground-plane more likely to be successful.

Ps. And indeed avoid spitting ground-planes. If you have a good routing and component placing discipline, and you can virtually imagine where and how the currents will flow, this is not needed at all. If in any case you must split the plane for whatever reason. Never run traces over the "split" but make a "Bridge" where the two ground-planes join, and only run traces over this split. This way the Rf current loop will remain local and can find its return path underneath. That is the whole idea and meaning of a ground-plane to start with.

With kind regards,
Bas

The way I see it, it's easier to interface (read: connect) SMD components such as bypass caps and alike when using a ground plane. Having a two layer board with SMD and through hole components I was thinking of having the ground plane on the top layer.

If you're willing it will be much appreciated if you can share some of your tips to a successful ground plane.

[Speedskater]

Henry W. Ott, who is a recognized authority, has several pages on the subject:

http://www.hottconsultants.com/

[stratus46]

Originally Posted by yakideo

It seems that there are many different opinions of whatever to use ground planes (split or not) or star ground. The eb-ta2020 datasheet recommends splitting the ground planes. However after more research I found this site (http://www.hottconsultants.com/techt...gnd-plane.html) claiming the opposite. I must say that the more I look into this issue the more confused I get.

Could someone clarify this for me?

I've done both in video designs. Now you'll say video is different but I think of it as audio on a massive dose of steroids. Instead of 20 KHz, make it 30 MHz. Star is very difficult to do well on a PCB while the ground plane in the board effectively randomizes the noise. If it makes the high frequency noise better in video, it will do the same in audio. Beware however that there are nodes in the circuit that don't like capacitive loading so 'holes' need to be placed in the ground plane at those nodes. Those are usually the '-' input of an opamp running as an inverter.

G²

[Speedskater]

Jim Brown (AES committee chair on EMI/RFI) wrote this about Mr. Ott:

OTT, HENRY W., Electromagnetic Compatibility Engineering,, Wiley Interscience, 2009 –
An absolutely essential book. Henry Ott nailed it, in this definitive text that ties together
both theory and practice in EMC. If you disagree with a single word in this book, you’re
wrong! This is a significant update and expansion of Noise Reduction Techniques in
Electronic Systems, Wiley Interscience, Second Edition, 1988, which has been the standard
for two decades. Ott doesn't "hide behind the math," including just enough to allow
solid numbers to be assigned to each mechanism that affects EMC.

I added the bold hi-light for emphasis.

You might try to find Mr. Ott's new book "Electromagnetic Compatibility Engineering"
in a collage library, as it's rather pricey.
Or his older book "Noise Reduction Techniques in Electronic Systems,"2nd Edition.