Re: [8023-POEP] What's the max voltage drop thru a midspan?
Everyone,
Are there any midspan makers out there who think that their board layout may
have used thin traces where the Ethernet data from the endspan is passed
thru to the PD? By "thin" I mean that you think your product can't take at
least 500mA per trace reliably. Please check with your layout people.
If you don't want to reply to the reflector because you don't want bad
publicity then please reply to me directly. I'm an independent consultant,
not affiliated in any way with any PSE manufacturer. I promise not to
reveal anyones name or company. I just want to know if this is a real
problem. (See the email chain below.)
Steve
-----Original Message-----
From: owner-stds-802-3-poep@xxxxxxxx [mailto:owner-stds-802-3-poep@xxxxxxxx]
On Behalf Of Chad Jones (cmjones)
Sent: Monday, June 12, 2006 10:43 AM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: Re: [8023-POEP] What's the max voltage drop thru a midspan?
In that case it will be a drastic temp rise.
I was using this tool:
http://www.geocities.com/capecanaveral/lab/9643/TraceWidth.htm
In 'reverse mode'; change the temp rise to find the desired trace width.
I'm not sure how accurate the tool is outside of a 'normal' temperature
range but with 800mA and .5oz Cu, I had to get to about 300C to get
under 5 mils trace width for internal layers.
-Chad
-----Original Message-----
From: Steve Robbins [mailto:steven_robbins@xxxxxxxxxxxxx]
Sent: Monday, June 12, 2006 1:37 PM
To: Chad Jones (cmjones); STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: RE: [8023-POEP] What's the max voltage drop thru a midspan?
Chad,
Unfortunately, there was no reason for midspan designers to route these
traces on outer layers, since they were not told they had to carry
current on them.
I wouldn't be surprised if some routed these traces on internal layers,
which have 0.5oz weight (0.7mils thick). Furthermore, if they routed
them as closely-coupled pairs then the widths may be as narrow as 4 mils
(depending on stack up).
What would be the temp rise in this case? I don't have the graphs,
since I don't work at Ixia anymore.
But speculation about temp rise in the traces may be pointless. We
didn't spec a current rating in 802.3af, so I don't see how we can write
802.3at assuming this path can carry significant current.
Steve
-----Original Message-----
From: owner-stds-802-3-poep@xxxxxxxx
[mailto:owner-stds-802-3-poep@xxxxxxxx]
On Behalf Of Chad Jones (cmjones)
Sent: Monday, June 12, 2006 10:26 AM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: Re: [8023-POEP] What's the max voltage drop thru a midspan?
Martin summarized it well by stating that the midspan would have to
carry the endspan currents in one of these (not recommended)
installations. Unfortunately, that means that legacy midspans may have
only been tested to carry 350mA - if this was tested at all.
On the flip side, 800mA flowing through a 5 mil trace of 1oz Cu on an
external layer will only result in a 17C temperature rise. I am
assuming that these traces are on an outer layer and that most people
use 1oz Cu on outer layers and that most traces are 5 mils or larger.
-Chad
-----Original Message-----
From: owner-stds-802-3-poep@xxxxxxxx
[mailto:owner-stds-802-3-poep@xxxxxxxx] On Behalf Of David Law
Sent: Monday, June 12, 2006 11:35 AM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: Re: [8023-POEP] What's the max voltage drop thru a midspan?
Hi Steve,
I see your point but the text states that the Mid-span 'shall not alter
the transmission requirements of the permanent link.' so it really
depends if the DC Current carrying capacity and the Resistance are
considered 'transmission requirements' or not.
Regards,
David
owner-stds-802-3-poep@xxxxxxxx wrote on 12/06/2006 16:00:15:
> David,
> Unfortunately I don't think this text solves the problem. My
interpretation
> is that it does the following:
> 1. The midspan must pass data thru it without degrading data
integrity.
> 2. As the midspan injects power on the spare pares, it must block
> that power from going back to the endspan.
> So it doesn't say anything about passing power thru on the data pairs.
> Steve
> -----Original Message-----
> From: owner-stds-802-3-poep@xxxxxxxx
[mailto:owner-stds-802-3-poep@xxxxxxxx]
> On Behalf Of David Law
> Sent: Monday, June 12, 2006 3:46 AM
> To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
> Subject: Re: [8023-POEP] What's the max voltage drop thru a midspan?
> Hi Steve,
> Do you think the following text contained in the third from last
paragraph
> of subclause 33.4.8 'Midspan PSE device additional requirements' of
> IEEE Std 802.3-2005 covers what you are looking for:
> ---oo000oo---
> Configurations with the Midspan PSE in the cabling channel shall not
alter
> the transmission requirements of the "permanent link." A Midspan PSE
> inserted into a channel shall provide continuity for the signal pairs.
> A Midspan PSE shall not provide DC continuity between the two sides of
> the segment for the pairs that inject power.
> ---oo000oo---
> Regardless, I agree that the IEEE 802.3at specification will need to
> be written carefully to ensure that, for example, it doesn't
retrospectively
> place a higher current carrying requirements on existing IEEE Std
> 802.3-2005 compliant Mid-Span PSEs.
> Regards,
> David
>
> owner-stds-802-3-poep@xxxxxxxx wrote on 10/06/2006 05:34:39:
> > Guys,
> >
> > As you all know, we're planning to run current from the endspan thru
the
> > midspan in a 4P system.
> >
> > Although I think its unlikely that we'll see Af-midspans with fused
> traces
> > or delaminated boards, technically we can't prove that it won't
> > happen because 802.3af doesn't specify a min current rating for this
path.
> There
> > seems to be no max limit for dc resistance either. (If these specs
are
> in
> > 802.3af and I've missed them, then someone please point me to the
> applicable
> > table or paragraph.)
> >
> > We should make sure 802.3at specifies both these parameters. (But
only
> for
> > midspans that output >15.4W so we're not retroactively putting new
> > requirements on Af-midspans.)
> >
> > Any comments?
> >
> > Steve
> >
> > [attachment "C.htm" deleted by David Law/GB/3Com]