Send comments to an expert

Screened cable connectors

You are here: Product design techniques > Cables > Screened cable connectors >

 

 

 

 

 

 

Screen termination

The overriding requirement for terminating a cable screen is a connection direct to the metal chassis or enclosure ground which exhibits the lowest possible impedance. This ensures that interference currents on the screen are routed to the enclosure without passing through or coupling to other circuits.

The best connection in this respect is one in which the shield is extended up to and makes a solid 360° connection with the ground plane or chassis. The effect of this geometry is to eliminate the inductance of the connection between cable screen and chassis. This is best achieved with a hard-wired cable termination using a conductive gland and ferrule which clamps over the cable screen. Any connector system will depart from the ideal, but a good construction can approach it.

Military-style connectors (MIL-DTL-38999) allow for this construction. Of the readily available commercial connectors, only those with a connector shell that is designed to make positive 360° contact with its mate are suitable. The cable screen must make 360° contact with a screened, conductive backshell which must itself be positively connected to the connector shell. An example of an approximation to this is the subminiature D range with dimpled tin-plated shells: both shells enclose the inner connectors over 360°, but the contact between them is only assured via the dimples on the long faces. A better construction is with a circular fully screwed shell mating interface.

Installing the mating connector

If a cable is to be screened, from the RF point of view the screen must be taken to the metal enclosure of the apparatus with the lowest inductance possible. The interference currents induced on the screen as the cable passes through the electromagnetic environment will flow into wherever the screen is connected. The whole purpose of the screen is to direct these currents away from the circuit, and conversely to prevent circuit currents from passing up the outside of the cable. Any inductance in series with the end of the cable screen will compromise this purpose. This depends on two things:

the way the cable screen is bonded to its connector shell, as above;

the way the mating connector shell is bonded to the chassis.

A direct metal-to-metal connection from the shell to the chassis, augmented if necessary by a conductive gasket, is best. But PCB-mounting connectors are often used with the connector shell taken into the PCB through mounting lugs. Although this isn't optimum, if the mounting lugs are bonded via a direct route down to the chassis, it may be adequate - the inductance of the bonding path is the determining factor.

If there is no chassis, then the connector shell can only be bonded directly to the interface ground plane. When there are other connections whose screens (or filters) are also taken to this plane, these connections from a preferential path for the interference currents, rather than having them pass into the circuit.

The pigtail question

The actual transfer impedance of a cable-plus-connector system is often dominated by that due to the connector.

A pigtail connection is one where the screen is brought down to a single wire and extended through a connector pin to the ground point. Because of its ease of assembly it is commonly used for connecting the screens of data cables. Unfortunately, it may be almost as bad as no connection at high frequencies because of the pigtail inductance, which appears in series with the cable screen and develops a voltage when interference currents flow down the screen to the ground connection. This voltage then couples readily from the end of the screen to the inner conductors. (This is a description of coupling into the system; the emissions coupling process is reciprocal.) The surface transfer impedance of such a connection rises rapidly with increasing frequency and effectively negates the value of a good HF screened cable.

If a pigtail connection is unavoidable then it must be as short as possible, and preferably doubled and taken through two pins on opposite ends of the connector so that its inductance is halved. Note that the effective length of the pigtail extends from the end of the cable screen through the connector and up to the point of the ground plane or chassis connection. Cable screens must always be taken to a point at which there is the minimum noise with respect to the system’s ground reference. If there is a conductive enclosure then the screen should be clamped to this.

This figure shows the effects of pigtail screen terminations, with different configurations of pigtail connections. The transfer impedance gives the voltage induced on inner conductors as a result of screen current, and it is dominated by the pigtail inductance. Therefore, the length of the pigtail becomes more critical as the frequency increases. The ideally terminated screen gives typically 50 dB of attenuation from 5 MHz up to 500 MHz, however with the pigtail termination this reduces to typically 10 dB at 500 MHz. This means that a high-quality screened cable is wasted if it is not terminated properly.

If only moderate screening is needed and only at low frequencies, say below 1MHz, then a pigtail termination can be acceptable if installation concerns over-ride EMC concerns, but pigtails should not be designed into a system build if they are unnecessary. For instance, a typical industrial cable termination uses terminal blocks for site wiring being brought into cabinets or module enclosures. These don't lend themselves to good screen termination, and indeed they encourage the pigtail, when a "screen" terminal is explicitly provided. A better approach is to provide a conductive saddle-clamp arrangement to a metal chassis plate just in front of the terminal blocks, so that the cable screens can be clamped down directly to this rather than being taken via pigtail wires to a terminal. This also has the benefit of giving a strain relief at the same point.


Click here for Questions