RE: current limiting of 802.3 power
Hi Ed,
There should be a law against using HTML for a reflector. It makes it very
difficult to respond back in readable form. Notice how the numeric table
below gets modified when editing... :)
I have some comments to address some points you made.
-----Original Message-----
From: Walker, Ed [mailto:ed_walker@ti.com]
Sent: Thursday, July 20, 2000 7:42 AM
To: 'Mike_S_McCormack@ne.3com.com'; stds-802-3-pwrviamdi@ieee.org
Cc: Hemmah, Steve; Pavey, Trevor
Subject: current limiting of 802.3 power
You must limit the current at the source. Otherwise if someone were to short
out the cable you could have quite a safety problem. The 350mA limit came
from the limits of the PCB traces on the patch panels.
Current must be limited at the source.
Agreed.
I would be very leery of using PTC's. They are very slow, have a wide
tolerance, are affected by temperature, and characteristics change over
repeated use.
This is an implementation issue that will be determined by how we spec the
current vs time characteristics of the PSE.
The control the committee wants to have over the power delivery dictates an
active circuit that can:
1. provide a switch in the power path to keep power off the port until
discovery has been made.
Implementation issue.
1. Slowly apply the power to the load, controlled ramp of current and
voltage to prevent ringing due to the inductance of the cable length and to
prevent surges being sourced from the PSE. Could effect other units power
when connected in parallel if port to port isolation is not required.
I think we should control current ramp and overshoot at the PD. The PSE can
sense over current and shut down quicker if a cable is shorted if we
approach the problem
that way.
1. Look at the current, once applied, and make sure it stays above a
minimum level to detect a device that has been removed and then disconnect
power.
Agreed.
1. Sense overcurrent conditions and provide protection.
Agreed. This may be stated by limiting the amount of allowable current, and
defining
that for normal operating conditions,(instantaneous and steady state) and
fault conditions.
1. Possibly do the detection also, if a simple resistive or diode
method is agreed upon.
To sense a minimum current you will already have a current sense circuit at
the PSE. Utilizing that same circuit to control maximum output current is
very reasonable.
Again, that is an implementation issue.
I realize that we are going to base our decision on which detection method
to use from implementation examples, but we should be careful not to create
a specification that drives specific implementations.
Regards,
Dan Dove