Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [8023-POEP] Solved another problem with DS



Derek,

 

Thanks for your interesting thoughts.  See my comments below, in curly braces.

 

Steve

 


From: owner-stds-802-3-poep@xxxxxxxx [mailto:owner-stds-802-3-poep@xxxxxxxx] On Behalf Of Derek Koonce
Sent: Wednesday, May 31, 2006 11:45 AM
To: STDS-802-3-POEP@xxxxxxxxxxxxxxxxx
Subject: Re: [8023-POEP] Solved another problem with DS

 

Steve,
1. I like your slide 5. With the switch in the midspan, the midspan can actually enable the endspan if the midspan has run out of power budget.
{An interesting thought.  Perhaps with the proper software this could work.  But with L1 there is no way for the midspan and endspan to be aware of each other, so you couldn’t have the midspan hand-off any loads to the endspan intelligently.}

 

Can the resistor be higher such that it interferes with the "signature" detection phase and thus the endspan would never really win because the signature is out of spec? {See page 6.  The circuit presents the endspan with an invalid signature, which is <Rgood.  If we made it Rinhib>Rgood then we might not reverse-bias D1 and D2, and that might not give the midspan the window it needs to detect the PD.}

 

{BTW, the switch isn’t intended to lock out the endspan indefinitely until the midspan detects a PD.  Both endspan and midspan attempt detection concurrently.  The switch simply creates a window for the midspan, to guarantee it has a chance to detect. Under 802.3af, there is no backoff time for an endspan, so a midspan might never even get a chance without this switch.  Therefore the switch by itself isn’t enough; the PD has to help.  (See page 8.)  Basically, a MP PD would force the detection process to repeat until the midspan won.  But an Af-PD would be happy with either endspan or midspan.}

 

The midspan can detect faster and avoid the delay in order to interfere with the detection for the endspan. Once PD is powered, the midspan could release the switch and conserve power on the endspan's behalf.  {Yes, I forgot to mention this in my presentation: The switch only closes during the midspan detection cycles.  After the midspan turns on the PD, it opens the switch.  So if the PD is DS, the endspan would be free to detect and turn on the second half of the DS PD.}



2. Slide 8. I thought the objective 7 was not to draw JUST enough to light an LED, but to operate at a lower power while providing the indication in some form.
  {Yes, a MP PD which is capable of operating at some lower power level may stay up with an Af-endspan, but if you want it to get full power it has to turn itself off so the MP midspan eventually wins.  But you’re right, I have to think about this some more.}



Derek Koonce



Steve Robbins wrote:

Guys,

 

I think I’ve solved another problem that’s inherent with DS.  See the attached pdf.  (Only about 10 more tough problems to go!) 

 

I’d sure appreciate some comments.

 

Steve