Thread Links |
Date Links |
||||
---|---|---|---|---|---|

Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |

*To*: "Roger Karam" <rkaram@cisco.com>*Subject*: Re: [10GBASE-T] The channel model*From*: "Chris DiMinico" <cd@mohawk-cdt.com>*Date*: Tue, 14 Jan 2003 16:46:21 -0500*Cc*: <stds-802-3-10gbt@ieee.org>, <xichen@marvell.com>*References*: <OF1A78FDFD.106B6A85-ON88256CAE.006B43EF@marvell.com> <4.3.2.7.2.20030114131803.01a0a2a8@fedex.cisco.com>*Sender*: owner-stds-802-3-10gbt@majordomo.ieee.org

Roger, TIA requires that measurements be performed on cable samples removed from the reel or packaging and laid out along a nonconducting surface, loosely coiled, or supported in aerial spans, with all pairs terminated. Modeling applies to this test configuations. Modeling "coiled" cable is as easy as power supply engineering. Regards, Chris ----- Original Message ----- From: "Roger Karam" <rkaram@cisco.com> To: "Chris DiMinico" <cd@mohawk-cdt.com> Cc: <stds-802-3-10gbt@ieee.org>; <xichen@marvell.com> Sent: Tuesday, January 14, 2003 4:18 PM Subject: Re: [10GBASE-T] The channel model > > Hi Chris, > > Just how much fun is it to model the coiled and the 'not coiled' cables? > roger > > At 12:38 PM 1/14/2003, Chris DiMinico wrote: > > >Xiaopeng, > > > >Worst case channel models for category cables are specified in TIA and ISO > >cabling standards. > >Given that one of our objectives is to operate over ISO 11801 cabling we > >should use the cabling > >standards (ISO and TIA) channel specifications as our 10GBASE-T model > >baseline (whenever > >possible). > > > >For 1000BASE-T, the channel models we used in the matlab simulations > >consisted of measurements > >(magnitude and phase) of cabling channels scaled to the envelope of the > >worst case function specified. > > > > > >Regards, > > > >Chris > > > > > > > >----- Original Message ----- > >From: <xichen@marvell.com> > >To: <stds-802-3-10gbt@ieee.org> > >Sent: Tuesday, January 14, 2003 2:44 PM > >Subject: [10GBASE-T] The channel model > > > > > > > > > > I would like to propose the following channel and noise model. Then we > > > can work together to come up with a model that all of us agree on. > > > > > > 1. Channel insertion loss model: refer to the RLCG model on page 47 of > > > T1E1.4/2000-002R6. Then calibrate the insertion loss (100m) at 100MHz to > > > 21dB for CAT-6 and 24dB for CAT-5/5E. > > > > > > 2. NEXT model: ~f^(1.5)n^(0.6) where n is number of disturbers. In our > > > case, n=3. Then calibrate the NEXT at 100MHz to -31dB for CAT-5E > >and -40dB > > > for CAT-6. > > > > > > 3. ELFEXT model: ~f^(2)n^(0.6) where n is number of disturbers. In our > > > case, n=3. Then calibrate the ELFEXT (100m) at 100MHz to -19dB for CAT-5E > > > and -24dB for CAT-6. > > > > > > 4. ANEXT model: > > > CAT-5E: max[-65, -38+15log10(f/1e8)] dB > > > CAT-6: max[-65, -41+15log10(f/1e8)] dB > > > > > > 5: Tranformer model: please contribute > > > > > > Xiaopeng > > > > > > > > > > > > > > > > > >

**References**:**[10GBASE-T] The channel model***From:*xichen@marvell.com

**Re: [10GBASE-T] The channel model***From:*Roger Karam <rkaram@cisco.com>

- Prev by Date:
**RE: [10GBASE-T] Auto-Negotiation Support** - Next by Date:
**Re: [10GBASE-T] The channel model** - Prev by thread:
**Re: [10GBASE-T] The channel model** - Next by thread:
**[10GBASE-T] The channel model** - Index(es):