RE: floating power, be afraid...
Hi all,
I like the idea of grounding the PSE ports at the PSE, or power supply
feeding 48V to the PSE('s). This removes the need for an additional 2.5kW
converter (200 odd ports) just to provide the port to ground isolation. A
converter this size isn't cheap, and is likely to have about 400W of
losses. I'm sure the people living in California would appreciate the
saving.
The DTE inputs would need to be isolated to avoid any possibility of ground
loops or transients that others have already talked about. This should not
be a problem as you need to convert the 48V input into something more
useful, like 5V, 3.3V, etc, so there is probably a need for a transformer
in the converter.
Simon Third
Invensys Energy Systems (NZ) Ltd.
"Rick Brooks"
<ribrooks@nortelnetworks. To: R karam <rkaram@xxxxxxxxx>, "Schwartz, Peter"
com> <Peter.Schwartz@xxxxxxxxxx>, stds-802-3-pwrviamdi@xxxxxxxx
Sent by: cc:
owner-stds-802-3-pwrviamd Subject: RE: floating power, be afraid...
i@xxxxxxxx
07/03/01 05:21
What Roger says is true.
However, I don't think that is really a safety problem.
This setup would mean that you are testing 512 ports at the same time,
since all 512 RJ-45 ports would be getting the high voltage from the Hipot
tester at the same time.
I think that the "greater than 2 meg at 500V", at the end of the Hipot test
applies to a single MDI.
- Rick
-----Original Message-----
From: R karam [SMTP:rkaram@xxxxxxxxx]
Sent: Monday, March 05, 2001 2:34 PM
To: Schwartz, Peter; stds-802-3-pwrviamdi@xxxxxxxx
Subject: RE: floating power, be afraid...
Hi Peter
the 39k causes a problem for the high pot test,
when 2000v low freq is applied for 1 sec...
roger
At 01:53 PM 3/5/01 -0800, Schwartz, Peter wrote:
>
>Rick:
>
>The concern seems reasonable. My 2c worth:
>
>I'd like to see a single-point ground [as you outline in (1)], but I
doubt
>that will happen. Even if it did/could happen, there is no clear way
of
>saying where in a building all the various switches, hubs, routers,
etc.
>might be - so it might be difficult (pardon me if I am being naïve)
to
>specify just where that single point is.
>
>Therefore, I am given to favor your second plan, wherein each PSE
port gas a
>small bleeder path to its local ground.
>
>Some quick math: (10 Meg / 256 ports) = 39k from PSE ground to earth
>ground. Does this introduce any safety hazard(s)?
>
>Peter Schwartz
>Applications Engineer
>Micrel Semiconductor
>Phone: 408.435.2460
>FAX: 408.456.0490
>peter.schwartz@xxxxxxxxxx
>
> -----Original Message-----
> From: Rick Brooks [SMTP:ribrooks@xxxxxxxxxxxxxxxxxx]
> Sent: Monday, March 05, 2001 12:54
> To: stds-802-3-pwrviamdi@xxxxxxxx
> Subject: floating power, be afraid...
>
> This reflector has been very quiet, so maybe I can stir up
some
>controversy?
>
>A) We are allowing "Env A" PSE's in 802.3af
>
> B) These "Env A" PSE's could have many ports which share the
same
>large DC power source
> therefore there could be many meters of cable that are all
hooked
>together at DC.
>
> Example: 200 ports, 100 meters each is 20 km, 656,000 ft, or
12.4
>miles.
> So, 20 km of cable that is all hooked together at DC, and each
PSE
>port is delivering DTE power
>
> Now, I plug in one more UTP cable and PD into this PSE system.
> I'm thinking that there could be a very large transient as the
new
>port turns on due to all the static
> charge that could be on all those 20 km of cables.
> Now, in reality, there is probably enough leakage to ground to
bleed
>off the charge, but
> can we be sure of this?
>
>
> Proposals:
>
> 1) make DTE power be tied to ground in the same way as telecom
(48V)
>power is, and as most power
> distribution systems in the world are. The connection to
ground
>could only be at one point, of course.
> This would require changing or at least interpreting the
existing
>802.3 spec.
>
> 2) Have each PSE port contain some leakage path, maybe around
10 Meg
>ohm,
> it must be greater that 2 Meg which is the 802.3 spec.
>
> 3) Make this concern that of the large PSE producer only and
not of
>the 802.3af
>
>
> Does anyone out there share these concerns? Any other ideas?
> - Rick
>