Tying Chassis Ground to Signal Ground

Sep 20, 2010 69 Replies

What is the preferred way to tie Chassis Ground to Signal Ground? I've seen a number of different ways.



Surround the edge of the board with a chassis ground trace and connect through caps to signal ground every couple of inches.



Tie the mounting holes to signal ground with capacitors.



Bring a chassis ground wire from the power supply and tie with a cap to the signal ground.



Just tie the chassis ground and signal ground together at one point in the power supply.


For audio and composite video: Don't. If you must, use a resistor to bleed charge, and a small cap for RF bypass if appropriate. Anything more invites ground loop.

Other devices, especially stand-alone equipment that doesn't need to be connected with other grounded devices:

My preference is to use a screw as chassis grounding (or several for RF), using at least a perimeter ground like so,

formatting link
which is good enough for single sided, low speed circuits. The perimeter is more for convienience, as you can return grounds to it as you go.

A full ground plane is better:

formatting link
(Same board, top side. This serves for illustration purposes.)

Better still, both sides (as above, or):

formatting link
formatting link
Note the screws in four corners, providing extra RF grounding (not visible are the aluminum standoffs grounding the bottom side -- I don't have PTH boards).

Since signal ground is chassis ground, ground loop, RFI and so on are minimized (assuming you didn't royally screw up something else). The circuit is doubly safe, because loose wires short to ground, which is ground ground, which is also safety ground (assuming you go on to use a three wire cord).

Tim

Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms "Wanderer" wrote in message news:3f577f82-8541-415d-828c-a5cad7dfc0c9@c16g2000vbp.googlegroups.com... > What is the preferred way to tie Chassis Ground to Signal Ground? I've > seen a number of different ways. > > Surround the edge of the board with a chassis ground trace and connect > through caps to signal ground every couple of inches. > > Tie the mounting holes to signal ground with capacitors. > > Bring a chassis ground wire from the power supply and tie with a cap > to the signal ground. > > Just tie the chassis ground and signal ground together at one point in > the power supply.

Use a single solid ground plane in a multilayer PCB. Nail it to chassis ground at spacers/connectors/brackets as many places as possible.

Joerg will rag me for giving this away for free, impacting his consulting revenue.

John

You are both saying to tie the chassis ground directly to the signal ground. Don't you end up with DC currents in the chassis?

xt -

I think it depends on the frequency and your circuit. My old mantra was a single ground point. The Star ground concept. At higher frequencies (maybe 1MHz and above?) this doesn't work as well and as John says things seem to work better if you tie the ground plane and chassis ground together at as many points as possible. Oh here is perhaps a rule of thumb. If your circuit doesn't need a ground plane then a single point ground may be OK.

George H.

Sure. The chassis helps stiffen up your ground plane.

John

ve

ct

in

text -

I think you want to accomplish a few things with the chassis ground.

  1. You don't want DC currents in the chassis.
  2. You do want the chassis ground and signal ground to be at the same potential.
  3. You do want a low inductance RF return.
  4. You want to keep the user safe from dangerous voltages.

You want the ground plane for low inductance paths for circuit traces, but you don't want those signals on the cables so you have beads and filter networks to keep the current in the cables DC. I don't think using a ground plane is a reason for tying it the chassis.

T'is how it's done. Opening some old surplus military electronics is a real eye opener for people not used to this concept. That stuff was designed so it'll still work should the Russian ignite the big one high up in the sky.

Yeah, pretty soon everyone knows how to make apple pie and I'm out of revenue here. I might have to come panhandling at the corner of Otis and Van Ness.

Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.

Latest with the advent of cell phones and in particular GSM the star ground concept is pretty much dead. Ok, for tubes it'll work. Most of them time.

Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.

You'd have to join the union. The panhandlers are organized and work in shifts. Seriously.

John

You've got to be kidding, right? Is San Francisco still in America?

Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.

Most of the time, yeah right!

Did you see, GSM phones screw up the Fluke 87?

I'm amazed something so disasterously awful as GSM was ever approved as a standard.

Tim

Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms

You've got to be kidding, right? San Francisco is in... California ;-)

Tim

Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms

Barely. It's sort of falling off the edge.

John

Technically, but who really knows anymore?

Tim

Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
[...]

I've been waiting for just the right moment to post this link.

formatting link

Heh.

I've

nect

ap

t in

d text -

ext -

Hi Wanderer,

OK lots of times I'm just making stuff up. Just becasue you put on a ground plane does not mean that you need to tie to the chassis at multiple points. But if your circuit needs a ground plane, then it may benifit from the the above idea.

1.) I guess I don't care too much if some DC current flows through the chassis. 2.) This is the critcial idea. With a single point ground you've got the chassis and ground plane tied together at one point. Away from this one point, the ground plane and chassis are at a slightly differernt potentials. You can then get capacitve coupling between the chassis and ground plane or signal traces. And that's the problem. 3.) yup 4.) and yup.

George H.

Silly, in a bad sort of way, and very badly written.

John

Maybe in SF, but here in Whittier, it's dog-eat-dog, albeit usually a guy will yield a spot if he's met his goal for the day. But there's no "organization," merely peaceful coexistence.

Cheers! Rich

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required