|CABLEIQ implements a new method to examine the RF transmission performance of cabling to determine if it is qualified for various network types. The current network specifications provide requirements for NEXT and RL and other parameters in terms of traditional frequency domain parameters. There is no distinction between the effects of either distributed or single point coupling or reflection sources. Cabling which has a single point coupling source will generate higher peak to peak coupling noise spikes than cabling with distributed coupling that has the same relative margin relative to the frequency domain limits. |
The amplitude of noise generated from distributed sources depends on the nature of how the individual noise signals tunnel back to the receiver and combine. In some cases they may add like a rouge wave or even cancel. In general the effective peak noise is always lower than that generated from a single point source since it is unlikely that all the peaks combine in the worst case. The peak noise from a distributed coupling can be predicted assuming that the noise signals combine randomly and can be treated as uncorrelated events. Since the network requirements make no distinction between the nature of the coupling, it must be able to tolerate the effects of the worst case point source and this can be used as the noise limit requirement.
Using these criteria allows the time domain impulse response of cabling to be examined in order to determine if coupling and reflection noise will be within proper operational limits. Since a single point fault with magnitude at the network frequency domain limit is allowed by the specification, it is used to establish the time domain limit for peak noise. The measured NEXT and reflection impulse responses are then processed in a manner in which information from all time bins is combined statistically using power sum methods (RSS - Residual Sum of Squares) to predict the peak noise that will result from the cable response. The computed statistical peak is then compared to the peak from a single point source to determine if the RF performance is qualified for a specific network type.
This method of qualification generally provides more margin to failure than the traditional frequency domain certification process to limits. The time domain analysis will generally predict failure only when the frequency domain response exceeds the frequency domain limits on average. Single excursions beyond the limit which cause failures in certifications instruments will not result in time domain analysis qualification failures.
Simulations were performed to demonstrate the effect of both point and distributed NEXT sources on the network signaling. The first example assumes a single point NEXT source which has coupling equal to the limit as specified in the 100BASE-TXrequirements. The frequency domain response of this coupling source is shown in figure 1.
Figure 1: 100BASETX Frequency Domain Channel NEXT Limit
From Point Source at Channel Limit
A 100BASE-TX signal was applied to this channel and the coupling to the RX channel was observed. The time domain graph of this simulation result is shown in figure 4.
Figure 4: Time Domain Coupling or MLT-3 Signal for Point and Distributed NEXT Conditions
The actual coupled signal from the distributed cabling is shown in red. A reference C5 Limit point source is shown in green. The coupling from the Cat 5 point source is 24 mV peak-to-peak and the noise from the TX signal in the channel with distributed coupling is 15 -17 mV peak-to-peak. Although this channel had 2 dB margin relative to the Cat 5 frequency domain limit, the actual noise generated as a result of a 100BASE-TX signal is only 70 % (-3 dB) of that the worst case point source which is allowed by the specification.
Figure 5: CABLEIQ Time Domain NEXT of 34 Meter Channel and Predicted Peak Noise
The time domain NEXT response was processed to estimate the peak noise. This estimate is shown as a flat straight red line. The estimate peak is 7.5 mV which is very close to the 8 mV peak seen when the actual 100BASE-TX signal is applied to the same channel. The coupling from a single point source at the connector limit is shown in green to provide a reference for the qualification limit. Note the predicted peak noise coupling from the channel with distributed NEXT is significantly less than the point source as demonstrated in the previous simulation results. Therefore the RF characteristics of this channel would be qualified for 100BASE-TX even though the frequency domain response had negative margin in one area of the sweep.