Monitor for Integrity of Seams in a Shield Enclosure

This paper considers a new concept for measuring the performance of shield enclosures. It allows for the presence of aperture-like faults at unknown positions over the shield enclosure. The aperture penetration is measured in a manner appropriate to bounding the response of internal equipment. The technique leads to the location of the faults as well. Index Code-F4d/e.


Monitor for Integrity of Seams in a Shield Enclosure
Abstract-This paper considers a new concept for measuring the performance of shield enclosures. It allows for the presence of aperture-like faults at unknown positions over the shield enclosure. The aperture penetration is measured in a manner appropriate to bounding the response of internal equipment. The technique leads to the location of the faults as well.
Index Code-F4d/e. I. INTRODUCTION NE DESIGNS practical shield enclosures for the protec-0 tion of electronic equipment from various electromagnetic environments. One would also like to be able to measure the shield performance in such a way that the measured parameters were directly relatable to the electromagnetic coupling to the electronic equipment, and could be used to assure the satisfactory operation of that equipment. Furthermore, it would be helpful if this way to monitor the shield performance could be used regularly during the lifetime of the shield to assure its continued adequate performance.
For most purposes a sheet of metal is sufficiently conducting that diffusion of electromagnetic fields through it is not a significant problem [l]. Apertures (such as faulty seams) are in general more significant and will be considered here. In many situations conductor penetrations are the most important; other techniques apply to this case [2], [3].
We begin with a review of the relevant features of aperture penetration, especially for long thin apertures (slots), then the technique in [2], [3] for bounding aperture penetration is generalized to allow for multiple apertures at unknown locations over a shield enclosure. Both the existence and location of the apertures can be determined.

APERTURES I N SHIELDS
Generally for practical shield enclosures aperture penetration is dominant over diffusion penetration for coupling to interior equipment, especially when the equipment is located near such an aperture. One can model the penetrant fields by equivalent electric and magnetic dipoles. In this case the near fields (but not too near the aperture itself in units of aper- of the linear dimensions, so small apertures usually do very little while large apertures can be significant. Better than considering the electromagnetic fields per se penetrating apertures, it is more directly related to the interference problem to consider the resulting source terms for the internal equipment. As discussed in [2], [3] an appropriate canonical problem to consider is that of a wire passing by an aperture. As in Fig. 1 there are two equivalent sources, a series (longitudinal) v-oltage source V, related to the shortcircuit magnetic field H,,, and a parallel (transve'se) current source I, related to the short-circuit displacement D,, (or electric field). In the near field of the aperture (but not too near the aperture) these terms are proportional to the time derivative of the associated short-circuit fields. In spatial terms the dependence is r-* where r is the distance from the aperture to the wire. This can be seen by integrating an r -3 field over a distance comparable to r , or by direct appeal to canonical problems [ 11.
These sources are important in that they are part of the equations used to bound the propagation of signals through subshields as part of the general topological decomposition of complex systems as a product (or convolution in time domain) of matrices in the good-shielding approximation [2]- [5]. In the context of apertures, since the source terms depend on the distance of a conductor from the aperture, it is useful to define an exclusion volume as indicated in Fig. 1 [2], [3]. One can define this exclusion volume somewhat arbitrarily (including zero if conductors are allowed to come right up to the aperture plane), but its purpose is to provide a basis for defining the maximum values that the source terms may achieve for any allowed position of conductors near the aperture (on the transmission side). In terms of normalized source coefficients these are [I]   Here Hsc is taken in a direction to maximize /Vsl and ESc is, of course, the component normal to the surface. For wire-to-wire coupling through the aperture the coefficients are similarly defined 121, 131.
For present purposes let us consider the aperture of concern to be a thin slot, such as is appropriate for a fault in a seam of a shield enclosure. Such a slot might be modeled as an ellipse, for which the case of the short-circuit magnetic field parallel to the long dimension of the slot is most important. In this case, let the wire be transverse to the slot for maximum coupling. This is a key point for diagnosing the presence of such a fault in a shield; the sensing wire should run across the slot, not be parallel to it. If the slot is not resistively loaded, then for frequencies below the slot half-wave resonance the dominant coupling to the wire is proportional to the time derivative of the short-circuit magnetic field parallel to the slot (or the short-circuit surface current density transverse to the slot).
An This cross-association of electric and magnetic parameters seems fundamental to understanding shielding. One way to look at this is that the BLT equation [6] uses wave variables that are linear combinations of voltage and current. This allows the separation of waves on one side of topological boundaries from those on the other side via a scattering matrix. Inherent in this is the possibility of electric parameters on one side producing magnetic parameters on the other side, and conversely. Another way to look at this is through the concepts of transfer impedance and transfer admittance 111, although the latter parameter may sometimes be better cast in terms of charge than voltage as the exciting term on the first side of the boundary.

A . Loops and Shields
We are here concerned with thin slots (seams) in shields for which the short-circuit magnetic field or surface current density is the appropriate excitation on the shield exterior. One excitation technique that has been proposed is a loop surrounding the shield [7]. This is most appropriate for low frequencies for which the loop is not resonant. In this case there is still the question of how large the loop should be compared to the shield dimensions to produce some sort of uniform magnetic field (incident or total) over the shield enclosure. Furthermore, it would be desirable to extend the frequency response to higher frequencies for which wavelengths are smaller than the shield dimensions.
On the inside of the shield the problem is somewhat similar. It has been proposed to place a loop inside the shield to enclose the maximum magnetic flux [SI, [9]. Since this is related to the line integral of the electric field around the loop, then at low frequencies this represents a maximum open-circuit voltage around some loop (neglecting multiple turns) that might be induced on equipment placed inside the shield. This includes not only aperture, but diffusion, contributions as well. Again at higher frequencies the loop can become resonant, destroying such a simple relationship to the internal electric field near the shield walls.
At this point we can note that transmitting from an external loop and receiving with an internal loop is only one way to look at this problem. One can also transmit from the inside and receive on the outside. These two procedures are equivalent under the usual requirements of reciprocity. Which technique is preferable often depends on other considerations such as the signal-to-noise ratio.

B. SCUTUM
Let us now extend the considerations of Section I1 in the direction of the aforementioned loops. Consider the external loop, but now deform it to be quite close to the exterior of the shield. Let this wire be maintained at a constant spacing from the shield walls so that it may be considered as a transmission line, of which the shield walls are an integral part. For this purpose let the spacing of the wire from the shield wall be small compared to both the radian wavelength and the radii of curvature of the shield walls near the wire.

Fig. 2. SCUTUM Concept.
As illustrated in Fig. 2 let us start the external transmission line at A and wrap it around the shield at a uniform spacing a from the shield, returning to a position A' near A . At A , drive the wire with respect to the shield with a voltage V,,, which might in general be CW or a pulse of some specified waveform. Let the other end of the wire A' be terminated to the shield in a constant resistance R,,, equal to the characteristic impedance of this transmission line. In this case we have established a wave propagating in one direction (from A to A ' ) around the shield exterior. The velocity of the wave is either the speed of light c if all the dielectrics are air, or somewhat less if other insulating dielectrics are used.
As the wave propagates around the shield there is a surface magnetic field, or equivalently a surface current density that extends to both sides of the wire a distance on the order of a . The actual distribution of this surface magnetic field (with apertures closed (shorted)) is readily calculated as in [lo]. SO a strip of width say 2a or somewhat more on the shield is effectively illuminated. Now turn our attention to the inside of the shield. Let US begin a similar wire at B (near A , except on the inside) and continue around the inside paralleling the external wire and ending at B' (near A ' , except on the inside). This interior wire is spaced a distance b from the shield walls and has its own characteristic impedance Ri, with respect to the shield walls as a transmission line. The propagation velocity is also c or somewhat less depending on the use of insulating dielectrics. Now perform a gedankenexperiment. Let a fast-rising pulse, say approximately a step function, be introduced on the external transmission line at a . It propagates along the external transmission line until it reaches some fault in the shield at some position, say X . At the corresponding position X on the internal transmission line a voltage source V, is induced sending a signal of voltage f Vs/2 propagating in both directions away from the fault with opposite polarities for the two waves. Assuming that the fault is a thin slot perpendicular to the two wires the coupling is maximized, and it is dominantly the external current (magnetic field) producing the internal voltage (electric field).
The two internal waves propagate away from X until they Going a step further, note that so far the technique covers four faces of a rectangular parallelpiped, but with current in only one of two possible directions on each face. So imagine three such pairs of transmission lines, each pair wrapping around in a manner orthogonal to the other two pairs. Then each face is tested in two orthogonal directions. Each pair tests four faces in one direction; three pairs cover six faces in two directions (a kind of octahedral symmetry).

M A I N T A I N THE C H A R A C T E R I S T I C IMPEDANCE OF BOTH TRANSMISSION L I N E S C O N S T A N T A S THEY PASS THROUGH THE CROSSING. USE ON B O T H O U T S I D E AND INSIDE OF SHIELD.
But why stop here? The external transmission line from A to A' paired with the internal transmission line from B to B' covers a strip or order 2a or 2b around the shield. Suppose now that the external wire is extended from A' to A" as a sort of helix around the shield with spacing from the wire to the previous (and subsequent) "windings" of order 2a, or perhaps more, so that coupling from one "turn" to the next is acceptably small. Continue this procedure until the appropriate shield walls (4 for a rectangular parallelpiped as in Fig. 2) are covered by this transmission line. Then terminate this transmission line at A" in its characteristic impedance.
Considering now the internal transmission line let it be similarly extended as a quasi helix, "paralleling" the external transmission line, to B" where it is now terminated. Compar-ing the signals at B and B" the previous concept of locating the fault is still applicable. It remains to consider details of the spacing of the "turns" from each other and from the shield to optimize the technique. For example, it may be desirable to have the internal and external wires not directly opposite to each other across the shield boundary. To optimize the uniformity of coupling from external to internal transmission lines their positions might be interspersed; the position of the internal transmission line might be spaced halfway between the positions of two adjacent turns of the external transmission line.
As the title of this subsection indicates, perhaps this concept needs a name. In this case we have shield closure using transitometer ultra monitor or SCUTUM (the Latin word for shield) for short. Note the concept of transitometer (related to the transmission of a signal through a fault), similar to the concept of a reflectometer.

C. Reduction of Mutual Coupling at Crossing of Transmission Line
When one of the transmission-line wires crosses another on the same side of a face of the shield enclosure there can be significant mutual coupling. Since the wires cross at right angles no magnetic flux from one wire links the other wire leading to no induced longitudinal voltage source. However, electric field from the charge on one wire does induce charge on the other wire, leading to an induced transverse current source.
In order to reduce this electric coupling at crossings the technique in Fig. 3 can be used. Here one wire is at least partly shielded from the other at the crossing. The wire closer to the shield at the crossing goes through some sort of tunnel (or even coaxial structure if desired) in which the wire dimensions andlor spacings are adjusted to preserve the character- istic impedance of the transmission line as it passes through the crossing. The other wire passes over this tunnel with perhaps additional conductors present, and the transmission-line impedance is also preserved through the junction in this case.
Note that this isolation structure is made of good conductors that are well bonded to the shield. Such a crossing should be made to occur at some position on a shield panel that is not too close to the seams where the shield panels are joined to each other.

IV. POSITIONING OF TRANSMISSION LINES
Consider now a pair of exterior and interior transmission lines (windings) as illustrated in Fig. 4. As previously discussed, the wires are spaced distances a and b from the shield wall. For simplicity now let a = b s shield wall thickness.
(4) Note the coordinate system with y = 0 as the shield wall and the wires (transmission lines) parallel to the z-axis. Now let x = nA,n = 0, +1, + 2 , . . .

( 5 )
be the x-coordinates of the wires with y = f a ( + for n even, -for n odd) as the y-coordinates. Now imagine that there is a current I in one of the wires (n = 0) and consider the fields and potentials. This has been worked out in [lo]. The normalized field on the shield wall (electric or magnetic) is (with value 1 at x = 0) Since we wish to minimize coupling to an adjacent wire on the same side, then we need In terms of potential one can obtain a more accurate estimate of the potential on one wire due to a unit potential on the first [lo], but the simple estimate in (8)  _ -2 s 9 s -
Then gl , g2 (for 6 > 2, and g3 represent the range of variation of g over q in the range of interest. The closeness of these numbers represents the uniformity of being able to detect a fault with respect to position on the shield wall.
As an example suppose we wished the coupling between wires on one side of the shield wall to be about 1 percent, giving fo = 0.01 This represents a factor of 2 variation in the detection of the signal through the fault with respect to the location of the fault on the shield wall. This illustrates the trade-off to be made in the design.

V. SUMMARY
This gives the general outline of the SCUTUM technique for characterizing and monitoring shield enclosures. It can be operated in CW or pulse mode, but pulse mode may be more convenient for fault location.
Our discussion has centered on aperture-like faults because of their practical importance. However, diffusion through metal walls is also picked up by this technique, which can be considered a generalization of the technique in [8].
The present discussion has considered locating a single aperture (say at X , X ) . The technique, however, is more general and can locate multiple apertures by considering successive pulses that arrive at B , the first pulse that arrives being from the first aperture, etc. If the propagation speeds on the outer and inner transmission lines are (nearly) the same, the pulses from all the apertures arrive at B" at (nearly) the same time. The time difference of pulse arrival between B and B" can then be used to locate all apertures (if the number is not too large) excited by the transmission-line pair.
There are still various detailed problems to be considered. For example, seams at the edges of the faces of the shield enclosure have different geometries for electromagnetic coupling from the case considered here. There are also the details of the construction of the transmission lines and spacing them from other external and internal structures.