Communications Technology: Archive

Hum Got You Down?
Block Capacitors Fix Reverse Path Woes
By Ron Hranac

It seems the more we learn about two-way cable system operation, the less we know. As if ingress, common path distortion (CPD), group delay and even basic network alignment arent enough to keep us occupied, there is yet another piece to this puzzle.

Shaw Cablesystems Allan Hamilton posted an interesting message on the SCTE-List describing a reverse path interference problem that manifested itself as upstream data packet loss. In addition to identifying the problem, he described an effective fix. The accompanying sidebar includes his original List message.

In a nutshell, Hamilton and his staff found that the usual neutral fault currents that exist on the outer surface of cable shielding were entering a subscribers drop wiring via the toroidal transformer found in the drop splitter. This current saturated the transformers ferrite core, which caused fairly nasty hum modulation that appeared to affect only the reverse path. The result was up to 40-percent packet loss, which obviously did a number on data throughput.

Take a look at Figure 1. Its a schematic diagram of a typical two-output drop splitter. Transformers T1 and T2 are toroidal devices; that is, they are wound on small ferrite cores.

All of the components shown in the schematic are installed in a metallic housing that provides suitable RF shielding, which in theory keeps cable signals inside the splitter and over-the-air RF signals outside the splitter.

But what happens when the cable shielding, and consequently also the splitter housing, is bonded to the buildings "grounding electrode" as specified by the National Electrical Code (NEC) or other requirements? The shield and splitter housing are placed at the same ground potential as the buildings electrical power neutral conductor.

This is done to minimize potential differences between cable and power grounds. It also means that power company or in-house neutral fault currents will be shared between the neutral conductor and cable shield.

Thus, some of those neutral fault currents will be present as sheath currents on the outside of our cable. So far, so good. A properly bonded and grounded plant can in most cases deal with this situation.

Now look at Figure 2. This is the same splitter schematic shown in Figure 1, with a path highlighted where low-frequency electrical interference can get inside the splitter via transformer T1s grounded end.

While RF cant follow the same ground-to-center conductor path, the transformer windings present a very low resistance to DC and low-frequency AC.

Out of curiosity, I connected an ohmmeter between a drop splitters input connector center conductor and the housing, and I measured only 0.1 to 0.2 ohm.

Heres where things get interesting. If the subscriber tap at the pole or pedestal is a type without blocking capacitors in series with the center conductor, some of the neutral fault current present on the drop shielding at the side of the house will flow from the splitter ground (housing) through T1s windings and onto the center conductor.

From there it will flow through the drops center conductor to the toroidal transformer in the tap, completing a "circuit." If the current is high enough, it will saturate the splitter ferrite material and possibly also the ferrite material in the tap port circuit, causing hum modulation.

Likewise, if there is a similar path present in a matching transformer or TV or videocassette recorder (VCR) input inside the home, an electrical current could flow from T1 through T2 and through the house side of the drop wiring, causing the same hum modulation problem.

This is by no means a new phenomenon. Nearly 10 years ago, Ron Hepler, then group engineer with Prime Cable, wrote about this in an article that appeared in the May 1990 issue of Communications Technology ("Hum Modulation in Drop Passives").

Heplers focus was how this problem affected addressable converter data carriers in one-way systems. His article described an example where a voltage drop of just 3 V across a neutral conductor could produce a current flow of 5.53 amperes on a drop, 17 percent of which would appear on the center conductor. Thats 0.94 ampere, clearly enough to saturate a drop splitters tiny ferrites.

The problem went away if you temporarily disconnected the neutral bond wire from the drop, but this created a safety hazard and noncompliance with the NEC. Prime Cables permanent fix was to buy drop splitters with blocking capacitors on the input port.

Figure 3 shows blocking capacitors added to the original splitter schematic. The capacitors will allow RF to pass, but not DC or low-frequency AC.

The capacitors eliminate the DC/low-frequency AC path through the cable center conductor, preventing current from flowing through any of the splitters transformer windings.

Ferrite saturation cannot occur, and neither will hum modulation. In practice, this fix requires more than just adding the capacitors; the capacitors by themselves will affect the splitters RF performance somewhat, so a few other components are necessary. Ive left them out to simplify the diagram.

Theres another benefit to using splitters with blocking capacitors. The capacitors are one way to help to keep the splitter ferrites from developing residual magnetism, which can cause passive device intermod. (See my column in the September 1998 issue of Communications Technology for more on this subject.)

Several manufacturers now have available drop splitters that come with blocking capacitors built-in.

If youve been pulling your hair out trying to solve reverse path data transmission woes that occur despite an apparently clean reverse path, you might want to look at this as a possible solution. Its a cheap and effective tool for your two-way toolbox. - CT

How It All Started

The SCTE-List is a great place to share your success stories as well as your engineering challenges. Shaw Cablesystems Allan Hamilton posted the following message on the SCTE-List.

Hello, Listers. I am a new subscriber to the List and thought you might be interested in a cable modem service call problem we recently had.

A new modem install had trouble getting dynamic host configuration protocol (DHCP) to get Internet protocol (IP) connectivity. It eventually got an address, but our tech found that pinging the gateway showed 40-percent packet loss.

He found that removing the ground at the entrance splitter totally cleaned up packet loss. We measured 4.5 VAC from the disconnected shield to building ground. Connecting an ammeter from shield to ground showed 1 amp of ground fault current (barely visible spark). Our underground plant is bonded to electrical utility service about six poles away on a different power distribution line.

My experience is that small potential differences in grounds is very common.

We installed a voltage block coupler (VBCan inline "F" device with a 0.001 µF capacitor in center conductor) at the entrance splitter. Pings gave 100-percent return; we removed the VBC, and pings immediately began to miss.

To gain a better understanding, we injected an 18 MHz continuous wave (CW) carrier at the splitter and measured hum back at the headend. With ground and no VBC, we measured 20-percent hum (2 dB peak-to-peak). Without ground or VBC, we measured 2 percent with erratic jumps to 20 percent. With ground and VBC, the hum was stable at 2 percent.

We feel that the ground fault current on the shield also is carried on the center conductor because of the small ferrite transformers at the multitap port and house splitter input causing the center conductor and shield to be in parallel. The current through the ferrite transformer causes saturation of the ferrite with hum modulation resulting, particularly at lower return frequencies. The TV picture quality was good with or without the VBC. The VBC does not stop the ground fault current; it just prevents the center conductor from carrying it.

Ground fault current is not new, but we believe it may be just as serious a return path problem as common path distortions (CPDs) and ingress in our return plant. It can vary with earth moisture, time of day and electrical loads as well as the condition of our plant and the electrical utility. It cannot be reduced by adding more fiber nodes, and it cant be detected by our signal leakage programs or spectrum analyzer noise monitoring.

It can be fixed with a simple VBC device, although a still better solution is indoor passives with built in blocks.

Allan Hamilton, A.Sc.T
Technical Supervisor
Shaw Cablesystems
Kelowna, British Columbia

Bottom Line

When Hum Modulation Strikes

Neutral fault currents that exist on the outer surface of cable shielding can enter a subscribers drop wiring via the toroidal transformers found in drop splitters.

This will, under some circumstances, cause an electrical current to flow on the cables center conductor. If the electrical current is large enough, it will saturate the splitters toroidal transformer ferrite core and cause hum modulation.

Shaw Cablesystems personnel in Kelowna, British Columbia, found instances of this that appeared to affect only the reverse path. The result was up to 40-percent data packet loss, which degraded cable modem performance.

The initial fix was to install a voltage blocking coupler (VBC) on the splitters input port, which prevented the unwanted electrical current from flowing in the drop. This success led to the use of drop splitters with built-in blocking capacitors on all of the splitter ports, a low-cost and effective fix to this potentially serious problem.

Ron Hranac is vice president of RF engineering for High Speed Access Corp. in Denver. He also is senior technical editor for "Communications Technology" He can be reached via e-mail at .