[SI-LIST] Re: How to connect Chassis ground to DGND
- From: steve weir <weirsp@xxxxxxxxxx>
- To: Chris Cheng <Chris.Cheng@xxxxxxxxxxxx>, cchalmers@xxxxxxxxxxx,si-list@xxxxxxxxxxxxx
- Date: Fri, 27 Feb 2004 14:44:17 -0800
Chris, ground seems to confuse a lot of people. DC isolation is usually
motivated by limiting loop currents in shelf to shelf or equipment to
equipment cables. It can be important. Shields in computer room equipment
cables have been measured carrying many 10's of amperes. A company that I
know of decided to "protect" themselves by only using optical connects
between shelves. Then they forgot to isolate the RS-232 craft maintenance
ports. They managed to set fire in one customer's installation when the
customer connected one PC to two shelves.
I had another case where due to powering requirements, we needed to DC
isolate. Given the opportunity in new equipment that didn't have the same
powering issues to get rid of the DC isolation, I jumped at the
chance. This completely baffled my colleagues who were and remain smart
people.
I can't explain the random goat sacrifices and other rites that some seem
to use with add on cards other than the observation that mythology doesn't
die very easily. PCs have some issues because the power return and logic
ground are not isolated. So, the power return does run through the chassis
metal.
If your EMI scan showed no difference you are lucky. Usually the caps
don't do as well as bonding. At least that has been my experience.
Steve.
At 01:56 PM 2/27/2004 -0800, Chris Cheng wrote:
>Steve,
>
>This is an interesting area I am always curious. If you look at the GBIC
>spec, it requires DC isolation at the connector point between chassis
>and logic gnd. FCAL on the other hand is exactly opposite and requires
>dead short between chassis and logic gnd. I never quite figure out
>why two huge groups of very smart people can come out with specs that
>are completely opposite to each other. It certainly makes my grounding
>scheme looks funny.
>
>To make things worst, a lot of add on cards like PCI cards seems to
>have a grounding scheme based on the phases of moon. And most of
>the disk drivers vendors like to tie their chassis to logic gnd
>internally just to make your integration a little more complicated.
>
>EMI engineers always told us to DC isolate the entire chassis with
>a single shorting point between chassis and logic ground. The rest
>of the system have decoupling caps to AC connect the chassis to
>logic ground. I never quite understand the logic and I have
>experimented with DC shorting the entire chassis with logic ground
>at every point. The EMI scan shows no difference.
>
>-----Original Message-----
>From: steve weir [mailto:weirsp@xxxxxxxxxx]
>Sent: Friday, February 27, 2004 11:57 AM
>To: cchalmers@xxxxxxxxxxx; si-list@xxxxxxxxxxxxx
>Subject: [SI-LIST] Re: How to connect Chassis ground to DGND
>
>
>Chris, you are setting up a cavity that will resonate at a frequency that
>depends on the size of your board. I think the problem here is the
>perception that any large area of metal has a constant AC
>potential. Anything with currents running through it does not. However,
>to make an antenna, there has to be a potential difference with something
>else, like the nearby earth. If we stick a chassis in the middle of a room
>and ground it to earth with long piece of lamp cord that chassis will
>radiate. The chassis is quiet when it has good connections to earth as in
>metal equipment rack. In the ideal case for EMI, everything in the box
>would be shorted dead to chassis. That's not too handy for signals, so we
>don't short them to chassis.
>
>What this means is that we want the DGND to look the same as chassis at any
>place that DGND is exposed. DGND is directly present on top of all of the
>signals in your cable. The potential difference is going to be the
>potential of the chassis at the shield bonding versus whatever is on DGND
>at the signal sources, unless you place a series CM impedance, or shunt to
>chassis.
>
>The two end connections of that floating plane to the DGND plane does
>little that helps and actually will create problems. If you don't need an
>extra DGND layer, your best bet is to remove that floating plane altogether.
>
>
>Steve.
>
>At 04:42 PM 2/27/2004 +0000, Chris Chalmers wrote:
> >Thanks for your reply Steve. You are correct steve in the assumption
> > that I have a plane of chassis ground next to a plane of ground. I
> > have an 0603 0ohm resistor connecting the chassis ground to digital
> > ground next to the connector that I want to stop radiating.
> >
> >At low frequencies the chassis ground will not be floating since it
> > is connected to Dgnd through a 0 ohm resistor. However, at higher
> > frequencies, the lead and mounting inductance of this resistor
> > will cause an impedance between the two planes causing problems.
>However,
> > the plane to plane capacitance between CHGND and the DGND due to the
> > stackup will take over a some higher frequencies effectively shorting out
> > the two planes. Would I be correct in assuming that in the lower
> > frequencies and at the higher frequenices that DGND at at the connector
> > would be pinned to CHGND thus avoiding radiated noise? What happens at
>the
> > cross over or does one take over before the other?
> >
> > >From your comments Steve it appears that a simple resistor connection
> > between the two is not enough but instead you need lots of vias between
> > them. Do I really want to connect the chassis ground to multiple sites
> > over the whole of the DGND plane as it will just look like DGND and I
> >will have lost the quiet effect of the chassis ground layer.
> >
> >I am not that wised up on this chassis layer stuff, please be gentle.
> >
> >Best Regards
> >
> >Chris
> >
> >
> >
> >
> >
> >My first quesion is, in this ov
> >
> >
> >-----Original Message-----
> >From: si-list-bounce@xxxxxxxxxxxxx
> >[mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of steve weir
> >Sent: 27 February 2004 11:04
> >To: cchalmers@xxxxxxxxxxx; si-list@xxxxxxxxxxxxx
> >Subject: [SI-LIST] Re: How to connect Chassis ground to DGND
> >
> >
> >Fun with grounds. If I read you correctly, you have put a plane into your
> >board that you call chassis ground that is for all intent and purposes a
> >big piece of floating metal as far as DGND is concerned. If so, this
> >should give you no end of grief. Either stitch that plane to DGND making
> >it just another DGND layer, or get rid of it.
> >
> >What you want to get rid of is common mode on your signals relative to the
> >potential on the chassis near the exit point of the signals. There are
> >several steps you can take to go about this:
> >
> >1) Provide good bonding between the DGND and the chassis ground in the area
> >surrounding each connector.
> >
> >2) Stitch DGND and the chassis itself over the area of the board.
> >
> >3) Insert CM rejection filters in series with signals leaving the
> >box. Some people do this with discretes, others use selective moating (
> >don't try if you don't fully understand it ) with or without other CM
> >suppression. The idea of the moat and the drawbridge is to put the signals
> >traversing the drawbridge and the island on the other side at the same CM
> >potential and to anchor that potential as the chassis at the exit point.
> >
> >Steve.
> >At 10:30 AM 2/27/2004 +0000, Chris Chalmers wrote:
> > >Folks,
> > > Sorry to bring this old chestnut back to life. I have a question in
> > >relation
> > > to Dr Howard Johnson's note on radiated ground noise (on sigcon
>website).
> > >
> > > I have a board that has a graphics DVI connector on it. I have
>connected
> > >the grounds on the
> > > connector to DGND on my board. I have also put a chassis ground layer
>in
> > >my stackup
> > > next to a solid ground layer. From Dr J's note, I have connected DGND
>on
> > >my board
> > > to the chassis layer at the DVI connector. This is to pin the DGND to
> >the
> > >same potential as the
> > > chassis ground (which should not be bobbing about like the DGND) at
>this
> > >point so that
> > > noise on my DGND is not radiated on to the graphics cable. Currently I
> > >have made the
> > > connection between CHGND and DGND with an 0603 resistor. Is this OK or
> > >should
> > > I make multiple connections with vias instead of the resistor?
> > >
> > > I have a second question as well if I may. I have a connector at
>the
> > >other end of the
> > > board which I want to implement a similiar strategy. Can I do exactly
> >the
> > >same at this
> > > other connector or will I start to have problems with ground loops
> >between
> > >the two.
> > >
> > >Thanks in advance
> > >
> > >Chris
> > >
> > >
> >
> >
> >
> >
> >
> >**********************************************************************
> >This communication contains information which is confidential
> >and may also be privileged. It is for the exclusive use of the
> >intended recipient(s). Please note that any unauthorised
> >distribution, copying or use of this communication, or the
> >information in it, is strictly prohibited. If you have received
> >this communication in error, please notify us by email, by
> >telephone: +44 1383 828800, or fax: +44 1383 828801, then
> >delete the email and any copies of it.
> >
> >This communication is from CRL Opto Ltd., whose registered
> >office is at 1 St David's Business Park, Dalgety Bay,
> >Dunfermline, KY11 9PF, Scotland.
> >
> >This footnote also confirms that this email message has been
> >checked for the presence of computer viruses.
> >**********************************************************************
> >
> >------------------------------------------------------------------
> >To unsubscribe from si-list:
> >si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
> >
> >or to administer your membership from a web page, go to:
> >//www.freelists.org/webpage/si-list
> >
> >For help:
> >si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >
> >List technical documents are available at:
> > http://www.si-list.org
> >
> >List archives are viewable at:
> > //www.freelists.org/archives/si-list
> >or at our remote archives:
> > http://groups.yahoo.com/group/si-list/messages
> >Old (prior to June 6, 2001) list archives are viewable at:
> > http://www.qsl.net/wb6tpu
> >
>
>
>------------------------------------------------------------------
>To unsubscribe from si-list:
>si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
>or to administer your membership from a web page, go to:
>//www.freelists.org/webpage/si-list
>
>For help:
>si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
>List technical documents are available at:
> http://www.si-list.org
>
>List archives are viewable at:
> //www.freelists.org/archives/si-list
>or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
>Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.org
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
