| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
Yair,
there are a few comments below.
In general, I am making this proposal so that the 802.3af spec will not run directly against
an existing patent not matter how fond we are of DC minimum current.
I am not judging the patent at all, except to comment that it appears to have
some validity pertaining to UTP data networks. It would be unwise to ignore it.
If anyone can find prior art, then great. In the meantime we cannot and should not discuss any legal
or patent validity issues. I suggest we simply consider the alternatives in case we must use them.
The sole reason to detect an unplug is to shut down power so that other RJ-45 devices will not
be damaged. Therefore, we could allow a few different detection methods for an unplug event and still meet this goal.
-----Original Message-----
From: Yair Darshan [SMTP:YairD@powerdsine.com]
Sent: Saturday, April 28, 2001 5:00 PM
To: Brooks, Rick [SC5:321:EXCH]; stds-802-3-pwrviamdi@ieee.org
Subject: RE: detecting a PD disconnection event
Rick,
For 1) The PSE should disconnect power from the port within 400mSec.
Reason: Lab tests shows that fast technician can not disconnect PD
with non-PD device and cause damage.
It is also suggested that within this time, the average port
voltage will be 5V max. (to limit PSE output port capacitor energy)
[Brooks, Rick [SC5:321:EXCH]]
That sounds good to me. The definition would be that the maximum time from an "unplug" action
until the PSE out goes to 5V or below is 400ms. Of course, we all need to discuss these numbers.
For 2) The easiest way is to measure the current and disconnect the port
voltage if the current is less than 10mA for more than TBD msec.
Actually the PSE should disconnect in less than 5mA, and may or
may not disconnect between 5 to 10mA. These values allows simple sensing
circuit
which we need any way for the overload protection.
The other method like checking the Ethernet link has few drawbacks:
1. It can not be supported by Midspan.
2. What will happen if the PD is connected to external source
voltage and you still have link, however the PSE is still delivering power
and it should not.
[Brooks, Rick [SC5:321:EXCH]]
These are good points.
See more comments below.
Yair.
> -----Original Message-----
> From: Rick Brooks [SMTP:ribrooks@nortelnetworks.com]
> Sent: ו, אפריל 27, 2001 5:58 PM
> To: stds-802-3-pwrviamdi@ieee.org
> Subject: detecting a PD disconnection event
>
> Just to start up another discussion, I wanted to find out how many people
> would support
> the following concept for detecting when a PD receiving power becomes
> disconnected.
>
> the concept:
> $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
>
> 1) The PSE is responsible for detecting when a PD receiving DTE power
> becomes disconnected.
> If that event occurs, the PSE must turn off the power within some time
> frame (100ms to 500ms ballpark?)
>
> 2) The PSE may use any of the following methods to detect this PD
> disconnect event.
>
> a) the DC load current drop below "X" ma (10ma?)
[Yair Darshan] See above.
> b) the ethernet link has dropped out
[Yair Darshan] See above.
> c) another proven method (pending discussion and feasibility
> proof)
> some suggestions:
> 1) the AC probing circuit detects when the AC load changes
> (opens)
[Yair Darshan] It is sensitive to noise. DC methods are more
reliable.
[Brooks, Rick [SC5:321:EXCH]]
The possible use of something like synchronous detection could help here.
> 2) the PD modulation heartbeat current ceases
[Yair Darshan] Not clear?
[Brooks, Rick [SC5:321:EXCH]]
This means that the PD would have a small circuit that produces a periodic current modulation.
The PSE has a watchdog timer that must always see the periodic current or the power shuts off.
> 3) the PD response to a given PSE code is wrong or absent
[Yair Darshan] This is related to false detection, and we need some
thing unique for disconnecting.
[Brooks, Rick [SC5:321:EXCH]]
I am assuming that this is independent of discovery, a different mechanism, and therefore not related
to false discovery.
> 4) other ideas
>
> $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$
>
>
> Please voice your opinions on this concept.
> The motivation here is to lower the cost of delivering DTE power.
>
> thanks,
> - Rick
>