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

RE: comments on Clause 47




Mike, Dawson,

The proposed limits on XAUI transmitter terminal voltages were chosen
such that the ESD protections would not be forward biased on a 1.8V 
supply receiver without coupling capacitors. If you plan to use a
XAUI transmitter with 50ohm resistors to a 3.3V supply and an output
amplitude of 800mVpp is generated, the nominal transmitter terminal 
voltages will be 3.1V and 2.7V. (2.9V and 2.1V for 1.6V amplitude)
I don't see the reason for going higher than 3.3V if you use a 3.3V
+ 5% supply and return loss compliant precision 50 ohm resistors.
Remember that the spec point is meant to apply for a nominal 100 ohm
test load resistor.

Because of fewer pins than XGMII, XAUI can be a preferred choice also
for short on-board links. Then, AC coupling capacitors are an
un-necessary
burden creating board layout and transmission problems. Already at
10Gbit,
a 24 port switch can have 768 XAUI links with 1536 capacitors. With some
foresight, XAUI specifications can be geared towards the anticipated
trends
in supply and signal voltages rather than towards older style
signal-voltages
that might be used. 

The anticipation when the 0V and 2V limits(-0.3 and 2.3 today) were 
first proposed, was to get feedback from implementers using voltages 
other than today's mainstream technology, such that the limits would 
not be made wider than what the implementations considered today demand.


Best regards  Tord.



-----Original Message-----
From: Kesling, Dawson W [mailto:dawson.w.kesling@xxxxxxxxx]
Sent: Tuesday, January 02, 2001 19:59
To: Reflector, Higher Speed Study Group
Subject: RE: comments on Clause 47



Mike,

Some comments on your first point.

The 2.3V upper limit was reached by compromise in November at Tampa.
Approval was overwhelming, but not unanimous. Several parties other than
yourself have verbally suggested the same increase that you are
suggesting,
though none have submitted comments. The motivation behind the limit was
mainly to put some reasonable limit on the DC voltage that the AC
coupling
must withstand. Some parties were concerned that a 3.3V limit might rule
out
certain AC coupling approaches. The main proponents of a lower number
have
since agreed that 3.3V would probably be acceptable. However, there may
be
some parties who are designing receivers capable of tolerating the 2.3V
limit without requiring coupling capacitors; these parties might object
to
raising the number to 3.3V. If the goal is to allow 3.3V operation, why
not
propose 3.3V + 5% for power supply variation? One comment has been
submitted
proposing 3.4V. On the low side, I am not aware of any objections to
reducing the lower limit below -0.3V. But again why not allow margin for
termination impedance variations and other factors that could increase
the
swing slightly?

-Dawson

-----Original Message-----
From: Mike Jenkins [mailto:jenkins@xxxxxxxx]
Sent: Thursday, December 28, 2000 1:45 PM
To: Reflector, Higher Speed Study Group
Subject: comments on Clause 47


All,

I have two comments on the XAUI driver electrical specs in clause 47.
Ballot comments will be submitted, but I want to air them here on the
reflector in case I'm missing some piece of history on these issues.

1) Section 47.3.3  Driver characteristics:  (Table 47-1)
	Present maximum driver voltage is 2.3 volts.  I'm unaware
	of the origin of this number, but it seems to pointlessly(?)
	exclude devices with higher supply voltages.  I propose 
	this spec be changed to 3.3 volts.

	Also the minimum driver voltage is listed as -0.3 volts.
	I propose that this value be changed to -0.4 volts.  The
	motivation is to not preclude any driver that swings
	around ground.  The max dif'l swing of 1600 mVp-p implies
	a max single-ended zero-peak voltage of 400 mV, which
	would be the max swing below ground.

2) Section 47.3.3.5 Driver template:  (Table 47-3)
	Present values for X1 and X2 are 0.325 and 0.450 UI,
	respectively.  I understand these are typos and comments
	have likely been submitted.  I propose (as I assume other
	comments will) that X1 be 0.175 UI, half the peak-peak TX
	jitter spec.  Several values have been proposed for X2,
	including X1+0.19 UI (the Fibre Channel value), as well as 
	a larger value.  I propose X2 = X1+0.205 UI = 0.380 UI.
	The derivation for this is attached, and is the closest
	I can come to the original Fibre Channel methodology.
	(The difference between X2=X1+0.19 and X2=X1+0.205 is
	about 5 ps, so this is not much of a big deal either way.)

Thanks in advance to anyone who can provide (helpful) feedback.

Regards,
Mike
-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 Mike Jenkins               Phone: 408.433.7901            _____     
 LSI Logic Corp, ms/G715      Fax: 408.433.7461        LSI|LOGIC| (R)   
 1525 McCarthy Blvd.       mailto:Jenkins@xxxxxxxx        |     |     
 Milpitas, CA  95035         http://www.lsilogic.com      |_____|    
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~