Above Board

This article was written by UltraCad Design in Bellevue, WA, who specialize in designing high speed boards.
They can be reached at (206)450 9708.

Distributed Capacitance Planes:

Some of our customers have asked about our opinion of the relatively new distributed capacitance planes that can be used in PC board fabrication. Some board manufacturers have licensed this technology from Zycon and offer it as an option to their customers.

Generally, UltraCAD does not promote or otherwise comment on other brands or companies. But we see this as a technological issue that may have relevance to our customers. Therefore, we have decided to offer these observations and opinions in hopes that they will help clarify some issues for our customers.

The distributed capacitance plane is basically a power/ground pair of planes separated by a very thin dielectric material. Typically, instead of designing a board with a single ground plane and a single power plane, one would design the board with two pairs of these planes. The plane-pairs provide a source of distributed capacitance that may provide advantages over a board without this structure.

Our board manufacturer friends have explained the primary tradeoffs regarding this structure to us in these terms:

The added cost of using distributed planes is approximately the same as adding two more layers to the circuit board. But the inherent capacitance of the planes, then, can be used to eliminate up to as much as 80% of the bypass capacitors on the board. The reduction in parts count can have two additional potential benefits:

1. It may mean that SMT parts (principally caps) that had to be mounted on the back side of a board can now be moved to the front or eliminated altogether, resulting in significant manufacturing assembly savings.

2. It may also mean that the reduced parts count allows for routing economies that, in turn, result in a reduced layer count.

Therefore, the cost of the extra layer pair can be (perhaps more than) gained back through reduced parts cost, manufacturing savings, and potential layout savings.

UltraCAD takes no position on these claims. They are legitimate Issues that an engineer should consider on a case by case basis in evaluating the benefits of the distributed plane technology for his or her particular design.

What we do see as a significant potential benefit is the possibility for improved EMI performance with distributed planes. In the Section above, we pointed out that a major problem in quieting boards where very fast logic circuits were used was the inability of even small bypass caps to provide charge to a switching device fast enough.This is because even the smallest bypass caps still have some finite lead inductance. Even very small inductances become important at the high frequency components of very fast rise times.

It would seem that it would be hard to find a source of charge for a switching device with lower inductance than a distributed capacitance plane directly under the device and connected directly to its power and ground pins. The amount of available charge may be very small, and the plane may not be a substitute for the use of bypass capacitors for the larger amounts of stored charge the circuit requires. But the plane may be just what is needed to provide the initial charge required at the very point of switching that lead inductance prevents ordinary bypass caps from providing.

For this reason, we believe a very significant potential benefit of the distributed plane technology is its promise for very quiet circuit switching with respect to EMI emissions. In systems where FCC compliance issues are important, this potential benefit, alone, may offset any additional costs associated with the technology.

(In all fairness we should add that some of the board fabricators we respect question if there are any benefits to distributed capacitance planes, and they worry about dielectric breakdown.)

Power Supply Boundaries:

Usually, the basic power supply either is placed near a corner of the board or is off the board altogether with power supplied through a connector. In the latter case, there may be, and should be, some sort of filtering at the point of entry. In all cases, there is a defined point where the power supply ends and where the power distribution system for the circuit begins. THAT is the singular point where the power supply should tie into the planes.

For example, in Figure 1.8c the typical power supply filtering ends at C2 and the power and ground planes start at that point.

C1 has a large capacitance for bulk filtering and C2 is a small capacitor for high frequency filtering. (Some people may add a third, very small value capacitor for very high frequency filtering.) Figure 1.8d shows alternative ways traces might be routed on the board. Figure 1.8d(a) illustrates normal routing practices, where power ties to the planes at several places instead of a single point, as illustrated in Figure 1.8d(b). Thus, in Figure 1.8d(a), any current flowing between the basic power supply and C1 and C2, or between C1 and C2 themselves, flows on the plane. This is not the case in Figure 1.8d(b). Again using Henke's voice analogy: If there are "voices" on the plane, they can be "heard" by the other circuits. So you want to keep those "voices" off the plane; and in doing so, be sure to use wide, low impedance traces between the input connector and the components before they tie to the plane.

It is important to remember once again that the filtering here works BOTH WAYS. It not only keeps power supply noise off the board, but it keeps board noise off the main power supply and keeps it from appearing on other boards in the system.

© 2000 Omni Graphics Ltd.