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

RE: [10GBASE-T] clarification on voltage level




When you performed the tests on cables, were there any connectors in to mimic patch panels and wall outlets? In the tests I've done in the past, a plain cable run tests really well but the DM to CM conversion caused by small imbalances/discontinuities occuring in connectors on the span increase the emmissions significantly.

Pat

-----Original Message-----
From: Larry Cohen [mailto:lcohen@solarflare.com]
Sent: Wednesday, January 15, 2003 12:43 PM
To: stds-802-3-10gbt@ieee.org
Subject: RE: [10GBASE-T] clarification on voltage level



There are many issues that determine whether a cable will meet FCC EMI requirements for a given line code.  The emissions from a cable span depend not only upon the cable characteristics, but on the balance of the line interface, the spectrum of the launched signal, and even the peak-to-average energy of the signal.  

The most important radiation mechanism from the cable is due to common-mode currents flowing on the cable.  The common-mode current sees the cable as a long wire antenna.  The level of common-mode current depends not only on the inherent cable balance, but more importantly on the line interface balance.  Other issues include common-mode discontinuities along the cable span (from connector interfaces and cable imperfections), proximity of the cable specimen to a ground plane, and the angle of orientation of the receiver with respect to the cable span.  Differential radiation is extremely small because of the tight twist pitch of the wire (with respect to the signal wavelength).  Emissions from the differential signal become important when the common-mode emissions are very small (and well below FCC limits).

We performed several measurements on various cable specimens at an FCC qualified test site to determine if extended bandwidth signals (up to 500 MHz) would cause potential problems.  The recorded emission level at each frequency represents the maximum observed level over a set of various antenna orientation angles as usually performed for FCC compliance tests.  Our results show the emissions below 500 MHz are below the FCC Class A limits.  The FCC Class A limits are specified in FCC Part 15.

Discussions of FCC compliance are only truly valid when considering an actual physical implementation and the physical configuration of the final product and attached cable.  For the purpose of this study group we only need to determine if transmission of high bandwidth signals on the cable creates emission problems.  Our stand-alone measurements indicate proper design (good balance) of the line interface should reduce emissions from the cable below compliance limits (FCC Class A) for signal bandwidths below 500 MHz.  Above 500 MHz emissions increase substantially.  Our results are less conclusive as to whether this increase is due primarily to the cable or the line interface.  


Regards,

Larry


Larry Cohen
SolarFlare Communications


-----Original Message-----
From: Rao, Sailesh [mailto:sailesh.rao@intel.com]
Sent: Wednesday, January 15, 2003 8:38 AM
To: 'Yossi Erlich'; 'stds-802-3-10gbt@ieee.org'
Subject: RE: [10GBASE-T] clarification on voltage level



Yossi,
	
	The DM-CM conversion characteristics of twisted-pair cabling
typically increases at a 20dB/decade slope as a function of frequency. This
is why the transmit PSD of 100BASE-Tx and 1000BASE-T exhibit a ~20dB/decade
decrease in the 30MHz-100MHz band to optimize their emissions
characteristics. 

	Yes, even with the 3V ptp signaling, the proposed average energy in
the 30MHz-100MHz band is less than what 1000BASE-T uses (the peak energy
could be a different story, depending on the signaling used in the proposed
system), but it has traded it off for energy in the 100MHz-833MHz band, and
also roughly doubled the total energy. 

	Even if we assume that the cable is well-behaved in the
100MHz-833MHz band, we should expect an additional 20dB/decade rise in the
DM-CM conversion, and therefore the tradeoff is not a good one.

Sailesh.

-----Original Message-----
From: Yossi Erlich [mailto:Yossi.Erlich@tiaris.com] 
Sent: Wednesday, January 15, 2003 2:34 AM
To: 'Rao, Sailesh'; 'stds-802-3-10gbt@ieee.org'
Subject: RE: [10GBASE-T] clarification on voltage level

Sailesh,
Why are you worried with the 50% voltage increment?
Using wider spectral band enables larger transmitted power
without exceeding the same PSD (about 158% voltage 
increment for the stated bands).
Regards,
Yossi

-----Original Message-----
From: Rao, Sailesh [mailto:sailesh.rao@intel.com]
Sent: Tue 14 January 2003 19:30
To: stds-802-3-10gbt@ieee.org
Subject: RE: [10GBASE-T] clarification on voltage level




Let's compare 1000BASE-T with the proposed system:

                          1000BASE-T        Proposed 10GBASE-T
Symbol Rate:               125MHz               833MHz
Transmit Shaping Filter:   0.75+0.25D           None
Launch Voltage:            2.0Vpp               3.0Vpp

In comparison with 1000BASE-T, the proposed system is using the cable well
beyond the specified frequency range, amplifying the launch voltage by 50%
and then eliminating the spectral filter. 

Even if we assume that the DM-CM conversion of the cable behaves reasonably
over the unspecified frequency ranges of the cable, this doesn't look good. 

Sailesh.         

-----Original Message-----
From: Vivek Telang [mailto:vivek@cicada-semi.com] 
Sent: Tuesday, January 14, 2003 9:59 AM
To: stds-802-3-10gbt@ieee.org
Subject: RE: [10GBASE-T] clarification on voltage level


This is true. Moreover, the emission levels can be meaningfully measured
only at the "system" level and not at the transceiver level. For example, a
prototype of a switch or a NIC is used to make sure that the final shippable
product meets the FCC requirements.

So it is extremely difficult (if not impossible) to answer the question :
"does this line code meet FCC requirements". But what you can do, is compare
the transmitted spectra of two line codes, and *assuming all other things
being equal*, make the claim that one line code will have XdB more margin to
the FCC requirement, if its transmitted spectrum is XdB lower than that of
the other line code.

Note that the shape of the transmitted spectrum, and the peak transmitted
power are as important (and IMHO more important) than the average
transmitted power.

Regards,

Vivek

Vivek Telang
Cicada Semiconductor
811 Barton Springs Road, Suite 550
Austin, TX 78704.
(512) 327-3500 x114
vivek@cicada-semi.com
http://www.cicada-semi.com



> -----Original Message-----
> From: owner-stds-802-3-10gbt@majordomo.ieee.org
> [mailto:owner-stds-802-3-10gbt@majordomo.ieee.org]On Behalf Of John
> DeAndrea
> Sent: Monday, January 13, 2003 8:56 PM
> To: xichen@marvell.com; stds-802-3-10gbt@ieee.org
> Subject: RE: [10GBASE-T] clarification on voltage level
> 
> 
> 
> We will need more information on cable balance and shielding to answer
> this question. It is not a simple calculation of pp voltage to FCC
> emission levels, but related to differential to common mode conversion
> in the cable and integrity of the grounding of the cable to the MDI.
> 
> John DeAndrea
> iTerra Communications
> 
> -----Original Message-----
> From: xichen@marvell.com [mailto:xichen@marvell.com] 
> Sent: Monday, January 13, 2003 6:16 PM
> To: stds-802-3-10gbt@ieee.org
> Subject: Re: [10GBASE-T] clarification on voltage level
> 
> 
> 
> 
> Can anybody who is familiar with FCC requirement give a careful
> calculation on the highest possible voltage and launch power that meet
> the FCC regulation.
> 
> Xiaopeng Chen
> Marvell Semiconductor Inc.
> 
>