According to an artical by Lee Goldberg in Electronic Design November 16, 1998, Giga bit ethernet uses"PAM-5" modulation scheme with a partial-response spectrum shaping of the form 0.75 + 0.25z(-1) to limit emissions to within the FCC limits. This simple filter shapes the spectrum so that its power spectral density falls below that of existing 100Bast T ethernet. Since 100BaseT is compliant, it is likely that Gigabit will be compliant.
The magic is in the spreading algorithim which spreads the energy uniformly over a very broad range while maintaining the bipolar average. As far as I know, all ethernet standards incorporate these modulation schemes in order to comply when transmitted over unshielded cables. Since emissions are measured with a 120kHz bandwidth (from 30-1000MHz), you can pump out a lot of power if you spread it over a large frequency (say 100MHz). If the other end is intellegent enough to decode your scrambled bits, and you code for emissions reduction/spreading the result is communications at a high rate with emissions in compliance with the FCC limits. Donald Kimball wrote: > The IEEE 802.3ab defines the new Gigabit Ethernet Standard (i.e. > 1000Base-T). This Local Area Network (LAN) can use 4 twisted pairs of > unshielded copper cable (Category-5) at 100m maximum operating at 250Mb/s > per pair in full-duplex bi-directional mode. This standard is designed to > utilize existing LAN cables such as older 10Base-T and 100Base-T networks. > The signaling (i.e. baud) rate is 125MHz per pair using 5-level Pulse > Amplitude Modulation (PAM) . This probably results in the fundmental energy > at 62.5MHz given that the signal must be bipolar to be compatible with > transformers. Vendors such as Broadcom Corp. have developed single chip > copper cable interfaces for this new standard. > > In the past, standards using unshielded cables, such as 100Base-T, > 10Base-T, T1, E1, etc, have had signaling rates less than 30MHz, so that > the fundamental frequency was below the 30MHz FCC and CISPR starting > frequency for radiated emissions. However, 1000Base-T has a 125MHz > signaling rate. A common mode current of less than 10uA at 30MHz at 1/2 of > wavelength can yield an emission level equal to or greater than the Class B > level. In addition, the 4 twisted pairs are all phase locked to each other. > The intentional differential mode current is about 10mA, so the trans > hybrid balance needs to be better than 60dB. This is achievable with the > hybrid at the component level, but not at the system level with 100m of Cat > 5 cable attached. > > Conclusion, I think that 1000Base-T (IEEE 802.3ab) on unshielded Category > 5 cable is doomed to fail EMI. Anybody ever try a test? Any other opinions? > > Don Kimball > > --------- > This message is coming from the emc-pstc discussion list. > To cancel your subscription, send mail to [email protected] > with the single line: "unsubscribe emc-pstc" (without the > quotes). For help, send mail to [email protected], > [email protected], [email protected], or > [email protected] (the list administrators). -- Jon D. Curtis, PE Curtis-Straus LLC [email protected] Laboratory for EMC, Safety, NEBS, SEMI-S2 and Telecom 527 Great Road voice (978) 486-8880 Littleton, MA 01460 fax (978) 486-8828 http://www.curtis-straus.com --------- This message is coming from the emc-pstc discussion list. To cancel your subscription, send mail to [email protected] with the single line: "unsubscribe emc-pstc" (without the quotes). For help, send mail to [email protected], [email protected], [email protected], or [email protected] (the list administrators).
