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

RE: [802.3af] quick questions

See my answers below.

> -----Original Message-----
> From:	R karam [SMTP:rkaram@xxxxxxxxx]
> Sent:	ω, ιεπι 16, 2001 3:27 AM
> To:	stds-802-3-pwrviamdi@xxxxxxxx
> Subject:	Re: [802.3af] quick questions
> Hi  yair 
> when we want to know when you wake up so here:
> 1- what is the incremental cost of the PD inrush if we add it?
	[Yair Darshan]  Depends on application and how much you need every
watt. I would say that the worst case incremental cost in discrete
implementation will be 
	1$ max.  including timing circuits and two level thresholds.
	For low power PD, 0.3-0.5$ max.

	For chip integration, it depends if you want the Mosfet inside. You
can get better answers from the chip vendors. 
>    and there is sure no shortage of parts needed in the PD now, so
> Integration
>    will happen, what is the added cost of the inrush relative to the rest
> of the needed circuitry?
	[Yair Darshan]
	For PD without inrush current limiter  
	a- signature resistor
	b- Isolating switch (small and low power since it is only a switch)
with few resistors and 1-2 small caps.
	c- Standard DC/DC converter.

	For PD with inrush current limiter
	a- signature resistor
	b- Isolating switch (bigger / higher power - pending on application)
with few resistors and 1-2 small caps.
	b1- sense resistor
	b2 - operational amplifier
	b3- Timing circuits
	b4- Reset circuit
	b5- Two level current limit threshold
	b6- Heat sink if it is used as current limiter in case of shorts and
faulty PD. Or thermal protection as an alternative.
	c- Standard DC/DC converter.

	Add to the above the art of specifying it to the PD designer...
> 2- May be for a class of devices with very low power we do want to rely on
> the inrush
>     from the switch alone .... I am not sure that we are advocating Inrush
> in the PD 
>     at all classes and under all circumstances? Dave do you belive this to
> be the case?
	[Yair Darshan]  The reason for inrush limiting is the PD cap.
	If you don't have a PD with input cap, you don't need inrush current
limiter in the PD.

> 3- Peter Raised a point where he was worried about the next port turn on
> time,
>    I am not sure that I got an answer to my question, clarifiying Peter's
> concern
>    the 350uf cap, if we turn on  port A at time T, how long will we have
> to wait to
>    Turn the Next port on if we make an octal part say? and what if we can
> turn on
>    every 3rd port at the same time or something ...
	[Yair Darshan]  You have two cases:
	Case A: When the system is off. all ports are loaded with PD's. The
system turn ON.
	In this case the time is not important. The system is turn each port
in orderly manner. It is more reliable and keep low stress on the system.
	Case B: The system is on, and Roger wants to connect new PD to a
port. In this case we need to have within 1 sec, 44V-57V across the port.
	No problem. Can be done easily.

	Now, what is the chance that at the same time Roger and Yair will
connect 2 PD's to the same Switch/Hub. I thing zero. and if it is more than
zero the cost is not worth it. In this case the 2nd port will be needed to
wait 1sec. (According to the spec, we need to serve single port...)

	This is a famous question in the Telecom field when we want to
specify the Ring Generator output power.
	The question is: How many phones are ringing now together. Answer.
Now zero. 1minute later may be one etc.

	Here we have much easier condition. What is the chance that two PD's
will hooked together to the same system? to the same 8 port chip out of 24
ports system, 48 ports system? probably zero. and if it is zero+sigma, the
sigma don't worth additional cost. Good statistics always keep costs down..

> did we calculate to accomodate a single PSE part?
	[Yair Darshan]  Yes. 
	Ip=0.4A for 50mSec. min. period of 1 sec. we can reduce the period
to 0.5sec. should not be a problem. I like the 1sec since it is the same
numbers as the normal powering mode and it will help to simplify the spec.
	As soon as you connect the PD to the port the process will start and
within 1 sec, you will see something a live.
	Now if you will disconnect this port and try to connect it again,
you need to wait 1 sec. which is O.K.
	We can play with the numbers anyway you want it.
> thank's
> yo all have a nice week end, Hurry and solve all this, Rick will be back
> next week.
	[Yair Darshan]  Thanks, now I'll am really closing my laptop and let
you have quiet weekend. 
> roger