… part II

The working predictive sizing table For the sake of simplicity, and therefore ease of communications and understanding, a simplified predictive PDN table was devised which provides the network designer with a more concise rules set (Figure 4).

The "idealized" predictive table appeared to have node sizes which appeared to fall into six distinctive size groups (in terms of passings), which were labeled A through F and established in steps of 50 passings, starting from Group A's 100 passings.

Figure 4: PDN design rules summary.

In recognition of the need in lower-density plant to have extended distribution feeds, the cascade rule (and therefore operating levels) for Groups A, B and C was established at Node+4 actives. The cascade limit for Groups D through F was set at Node+3, which therefore allows the designer the flexibility to add the occasional third line extender where needed (the numeric suffix for each group identifies the cascade rule), even though Node+2 is often the practical limit. The absolute size limit for the node was set at 400 passings, even if the power supply is capable of a higher loading.

In MediaOne's current practice for PDN design, node and amplifier operating levels for Groups A–C (Node+4) have been set at 48 dBmV at 750 MHz with a 12.5 dB tilt. Groups D–F (Node+3) operate at 49 dBmV at 750 MHz with a 12.5 dB tilt. The levels chosen leave a comfortable margin of safety before reaching the compression point of silicon-hybrid 750 MHz amplifiers.

The RF power loading is based on an assumption of 80 channels of analog NTSC video to 550 MHz, with 200 MHz of digital operating at -10 dB relative to the visual carrier level. The expected end-of-line performance (including analog headend-to-hub optical transport and RF combining amplifiers) is 48 dB/-53 dBc/-53 dBc for C/N, CSO and CTB, respectively. While the operating levels chosen work well on analytical models, they are regarded as tentative pending the outcome of empirical testing.

The high operating levels might imply the use of high-value taps (29 dB and higher) immediately out of the amplifier, but the highest tap value to be used has been set at 26 dB to avoid excessive passive loss on the sub-band return. The high tilt also obviates the use of 29 dB and higher taps because of the low levels at Channel 2.

The higher operating levels of PDN designs will result in improved amplifier "reach," thus increasing the number of passings per amplifier. Based on MediaOne's early results with PDN trial designs, the number of actives per mile can be expected to decrease by 15 to 20 percent, as compared with HFC-500. Again, the "idealized" and working predictive tables reflect an estimated reduction in the number of actives-per-mile of just 10 percent.

Additionally, it is expected that the proportion of line extenders will increase, further moderating the power load. The initial PDN sizing tables, therefore, are intended to serve only as a starting point from which to begin initial PDN design work. As fresh BOM data from PDN designs is assimilated, the predictive tables will be revised to reflect more aggressive sizing targets.

How well does PDN work?

The PDN design concept takes a little time and some trial-and-error for the network designer to master. Despite the availability of the predictive table to determine approximate node size, it often takes designers several attempts at nodes of varying densities to achieve acceptable and repeatable efficiencies.

Very simply, the following are the new design rules and goals that MediaOne communicates to the HFC design engineer engaged in PDN design:

  • As a prerequisite to use of the predictive node sizing table and the fencing of node service boundaries, the general density in passings per plant mile must be determined;
  • The node, regardless of density, should be large enough in passings and mileage to effectively consume the 75 percent target load factor limit of a single 90 VAC, 15-ampere power supply (with the exception that the node must not exceed 400 passings, regardless);
  • The cascade limits have been set at Node+4 or Node+3, depending on the density group;
  • As a result of the cascade reduction, the forward operating levels have been increased to 48 or 49 dBmV at 750 MHz (with a 12.5 dB tilt) to stretch the "reach" of each feeder amplifier;
  • The use of new overlaid coax cables for express runs and back-feeds should be ruthlessly minimized, particularly in underground areas. Compared to conventional HFC-500 design, these are some of the changes in design metrics that MediaOne expects to realize in PDN designs, all of which have been borne out in early node designs for medium- and high-density plant:
  • An approximate 15 to 20 percent reduction in the number of actives per plant mile, and a substantial increase in the proportion of line extenders;
  • An increase in the number of nodes of about 50 to 75 percent, with a higher proportional increase in areas of low density;
  • A reduction in the number of standby power supplies by approximately 20 to 30 percent;
  • An improvement (increase) in the average power supply load factor of approximately 10 to 15 percent (e.g.: increase from 60 percent average load factor to 70 to 75 percent);
  • A reduction in the amount of new coax cable overlay (new coax cable feet as a proportion of cable-bearing strand/trench feet) from the current range of 15 to 25 percent to less than five percent;
  • An increase in the amount of fiber overlay (fiber route mileage as a proportion of coax plant miles) of less than five percent (e.g.: increase from 22.5 percent to less than 27.5 percent);
  • A reduction in the average cost per mile for outside plant upgrade of approximately seven to 12 percent.

The last point, while critically important, is something of a moving target. MediaOne is confident that such cost reductions will be commonplace in medium- and high-density plant, but realizing similar reductions in lower-density (<75 passings per mile) plant will prove difficult to achieve because of the higher proportions of fiber overlay needed to reach nodes of 100 to 250 passings. As a practical matter, however, few low-density nodes in today's HFC-500 architecture approach 500 passings in any event because of cascade limitations.

Figure 5: Hypothetical power domain node physical layouts.

Another important caveat involves suburbia. In today's HFC-500 designs, it is commonplace for the node receiver to be placed outside underground subdivisions on the aerial right-of-way. The move toward smaller nodes (regardless of the specific topology) will certainly force the extension of fiber into areas of underground utilities, with the obvious increase in labor and material (conduit) costs.

To moderate costs even further in PDN design, an operator might consider these options:

  • Reduction of the fiber count from six to four, or less. It is then possible that the overall fiber requirements in higher-density areas might actually decrease, despite the increase in the number of nodes;
  • Use of higher-powered laser transmitters with increased optical splitting;
  • Transition to a less-complex and lower-cost node receiver, especially in low-density areas where there is a lesser need for the multiple RF outputs of the more elaborate receivers;
  • Use of a centralized powering architecture in high-density areas where the costs of a "power node" type power supply might be efficiently shared over multiple nodes;
  • Avoidance of AGC amplifiers in favor of manual- or thermal-control only.

As with other architectures that feature smaller nodes, the vulnerability to upstream ingress is not necessarily improved unless some advantage is taken from the greater node granularity by providing more communications ports for each interactive service. Naturally, the increase in the number of nodes will increase the complexity and expense of the downstream and upstream combining and splitting networks, even if the extra node granularity is not fully exercised.

As a final thought regarding costs, MediaOne expects that the smaller nodes will prove to be much easier to activate, sweep and certify for advanced services. The "time to market" period between initial node activation and final certification for carriage of advanced services (RF-IPPV, HSD & telephony) might well be reduced by half of its current period (of course, the number of nodes to certify will increase as well).

Growth strategies

An important consideration in the PDN topology is determining how to handle growth in node power loading in an efficient and non-disruptive manner. Power loading increases can come from minor plant extensions or growth in telephony service penetration, or both.

MediaOne is presently soliciting the power supply manufacturers for standby power supplies (some with integral gensets) that can be modularly "grown" beyond the initial 15-ampere output capacity to 18, and ultimately, 21 amperes. Because the network is likely to carry "lifeline" telephony traffic and high-speed data, the upgrade must be accomplished in a "hot swap" manner and be totally non-service-interrupting.

Because active and passive components have a limited current-passing capability, an output power bus is needed on the power supply that will provide at least two current-limited outputs to remote insertion points via relatively short power feeder cables, which are best placed at the time of initial plant upgrade.


From the Power Domain Node designs completed thus far, MediaOne is confident that this variation on HFC architecture is one of the most effective plant upgrade solutions that has been found so far which positively addresses the issues of first cost of capital, cost of operation, and network performance and reliability. Its attractiveness with regard to initial capital costs in medium- and high-density areas with aerial plant is clear. In areas of especially low density and underground utilities, its economics are still open to question.

Even if the first-cost economics fail to match those of HFC-500 in some cases, the Power Domain Node upgrade architecture provides clear benefits in network performance, reliability and operational efficiency. It may prove to be the preferred architecture in some "greenfield" applications as well. Indeed, in most design cases examined by MediaOne so far, PDN's motto appears to be, "More gain, less pain!"