RE: [802.3af] Late comment
10 meg resistor. Very good point.
Analog Product Specialist
Texas Instruments Incorporated
HC66 Box 203
Mountainair, NM 87036
WEB SITE: <http://www.ti.com>
Office = 505-847-0576
Fax = 413-280-0812
From: bachand@xxxxxxx [mailto:bachand@xxxxxxx]
Sent: Wednesday, January 15, 2003 3:16 PM
To: Steven Murray
Cc: Jack Andresen; cmjones@xxxxxxxxx; 'Geoff Thompson';
Subject: Re: [802.3af] Late comment
Although it may make sense to tie the common to earth ground, the existing
IEEE-802.3 standard has a nasty little requirement that it must be floating.
In fact it's further specified that the system should be able to have all
conductors of all ports tied together, then hi-pot tested at 2250Vdc,
the wires and the chassis.
Therefore it was decided that this group will not make any changes to the
existing 802.3 standard.
I would however suggest a high resistance (possibly 10M) or so, between the
common and chassis, in order to drain static charge.
Gerard E. Bachand
Power Systems Engineer
7 Atwood Terrace
Cherry Valley, MA 01611
> "Jack Andresen" <jandresen@xxxxxxxxxx> wrote:
> > Any common point needs an impedance to ground.
> I would agree with this comment, and add:
> It is unusual practice to distribute a commoned
> conductor to a number of physically seperate
> locations without providing an earth return, somewhere.
> If a fault should happen that causes a hazardous
> voltage to be applied to the commoned conductor,
> that hazardous voltage will then be present in a
> number of physically separate locations. This could
> create a hazard for a person handling a connector,
> or a hazard for the electronics in certain circumstances.
> If the common point were grounded, the hazardous voltage
> would normlly only affect a smaller number of devices:
> Particularly, just the devices on the cable to
> which the hazardous voltage was applied.
> In the situation where a fault causes a hazardous voltage
> to be applied to the commoned (but ungrounded) conductor
> for an extended period of time, and the electronics is
> not affected, the fault is likely to remain undetected
> until such time as it causes a failure or a person comes
> into contact with a conductor somewhere in the system.
> A danger of electrocution might exist in this circumstance,
> the issue being that the fault may remain undiscovered until
> such time as it harms a person. In point-to-point data wiring
> (eg a LAN Hub and a PC) such a fault condition would really
> only affect the two endpoints. However with commoned
> (but ungrounded) POE wiring, the hazardous voltage could
> well be present at many physically seperated locations.