Network power, a natural progression

Advances in the design of power packs in coaxial and fiber-based signal processing equipment have not been paralleled by advances in power source technology. As a result, the potential of modern broadband networks as power distribution systems has not yet been fully exploited.

Research and development in power wave shape control promises significant improvements in area coverage, efficiency, stability and reliability of networks powered by AC sources. Here, we'll examine the characteristics of network element power packs and how to enhance network reliability and performance with the use of a new power source technology called Unity Wave.

Ferroresonant Technology Traditionally, broadband networks are powered by a root-mean-square (RMS) regulating device called a ferroresonant transformer (FR). The FR has been used in networks since the advent of cable television and can enhance network reliability by enabling the operating voltage at the network element power packs to remain constant in response to changing utility line conditions.

The output wave shape of unloaded ferroresonant transformers was named "quasi-square" and has come to represent the industry standard input wave shape for network elements. Interestingly, the output wave shape from these devices actually varies between nearly square, asymmetrically peaked and haystack-shaped, depending on the FR's type, loading and utility input voltage.

Thus, it seems that the term quasi-square is largely undefined. Signal processing equipment for distributed systems is designed to "feed through" this low frequency power. The intended result is conditioned power that is available at any point in the network distribution system.

Over time, the FR does maintain good RMS regulation, but it has a relatively slow response time to changing line or load conditions and tends to temporarily overcorrect in response to transients. The result is network voltage overshoot or undershoot, which can create opportunities for the network to destabilize, trip protective devices or become damaged. In addition, the output wave shape variation leads to variation in network performance, especially at heavy loads.

The same variation makes accurate numerical prediction of network node voltages during the design phase difficult, if not impossible. The common practice to address this is to derate the power supply to between 80% and 85% of its rating, resulting in underused power nodes. System stability problems generally do not surface until the network is operated at or near its "knee," a point where the power required at the first node downstream from the power node exceeds the power that can be transmitted through the network cabling.

In networks where the load is somewhat unpredictable, such as those in which requirements change with traffic, the knee may be approached inadvertently. Reliability in these networks would be enhanced if the knee were to be moved outward, effectively increasing the maximum power transmission capacity of the cabling.

In its favor, the FR is simple to design and has a low parts count in non-uninterrupted power supply (non-UPS) applications. In these applications, the parts count might be made similar between ferroresonant and switching designs. In any case, as long as their limitations are taken into account, networks can be designed to operate reliably using ferroresonant devices.

Network Element Power Packs In the not-too-distant past, all line-powered signal processing equipment was equipped with a series-pass DC power regulator called a power pack. The purpose of this regulator was to supply a fixed DC voltage to the signal amplifier over a stated range of network node voltages. Today, virtually all of the series-pass power packs have been replaced by switching regulators.

The initial migration was made at a slightly higher cost and parts count, but the available benefits could not be ignored. The greater efficiency, wider input voltage range and lighter weight of the switching power packs drove vendors to develop switching designs that became more cost-effective and would eventually supplant the simple series-pass regulator as the network power pack of choice.

Similar to its predecessor, the switching packs use capacitive input filtering, offering an opportunity to control power factor in the network cabling by forcing the power source voltage to be as nearly square as network harmonic limits allow.

How can that be? Poor power factor usually occurs when a load draws most of its power during short portions of a power cycle. If the voltage crest is peaked (as opposed to flat), capacitive filters will concentrate most of their input power consumption at or near the voltage peak. Conversely, if the voltage crest is flat, capacitive filters will average their input power consumption across the crest, drawing power over a greater percentage of the power cycle.

This theory is easily put to the test due to the constant power characteristics of switching power packs. A switching power pack will draw only as much power as necessary to supply DC voltage to its network element load. If the input voltage is increased, the power pack current is decreased proportionally and vice-versa.

Therefore, any difference in measured power (P) or current (I) drawn by a network that is supplied by two different sources having equal RMS output voltage (E) must be due to a difference in power lost by the network cabling. Ohm's law for AC circuits states that any such difference can only be due to a change in power factor (PF) driven by the wave shape of the source: P = I x E x PF.

An indoor network was constructed at the Antec Technology Center in Atlanta for the purpose of verifying this theory. The test bed was constructed with coaxial cable and network amplifiers as presently deployed in actual systems.

An additional programmable constant-power load was employed to enable the test loading to be varied. Once the network was adjusted for a comparative test, power analyzer data was collected for each of the test sources without further load adjustment.

Test Results Plots taken from each source powering a 15-amp load are shown in Figures 1 and 2. Notice the characteristic sloping output voltage (V out) of the FR and the high crest factor of its corresponding current trace (I out) in Figure 1. Compare these two traces with those in Figure 2 and an improvement in power factor becomes apparent, owing to the fact that the switching source delivers a voltage output with a flat crest.

Comparative testing data for RMS current, network input power and network power factor are given in Table 1. Under load conditions producing a current of 15 amps from an FR, the switching source produced a current of 13.2 amps, a reduction in effective network loading of 12.5%.

In the same instance, the power required to drive the network was reduced from 1221 W to 1145 W, a reduction of 6%. As the test loading was increased until the switching source produced 14.7 amps, the FR transformer drove the same network with 17.5 amps. While the switching technology was operating at 98% load, the FR technology was at 116% of load. In both cases, full load is recognized as 15 amps. What fundamental change in network characteristic is responsible for these changes in performance? In all cases, the network power factor was driven significantly toward unity by the switching source wave shape.

Switching Technology To date, switching technology has been applied and proved in nearly every application previously reserved for linear or ferroresonant technology. Proliferating first in DC applications like telecommunications rectifiers and computer power supplies, switching technology has now migrated into AC UPS systems and DC network element power packs as well. All of these industries are sensitive to cost and reliability.

Research and development are ongoing at Power Guard to adapt switching technology to the broadband network power source at a competitive price to deliver the promise of better network reliability and performance. With the use of switching technology, total control of network power is possible. The output shape can be controlled for optimum power factor.

At the same time, the output voltage can be controlled without changing the shape, making the network entirely predictable. Given the ability to synthesize the source output signal, a vista of new opportunities for system enhancement is achievable beyond the scope of this introductory article.

Pros and cons of two methods of AC network powering have been examined in some detail. A few drivers for the evolution of broadband network powering technology have been identified, and a new concept of system powering called Unity Wave has been introduced. Although switching technology is now a mature and well-understood art in electronics, its application in network power sources is only recently being understood and proved.

As was the case for network power packs, the benefits to be gained by its application are too numerous and too significant to be neglected. As technology continues to progress, the choice between the familiar ferroresonant and the new switching power sources will become increasingly easier to make. Is there a switching power source in the future of your broadband network? It would be a natural progression.

Gary Batson is Vice President of Engineering for Power Guard, Opelika, Ala.