Progress towards Copper Cable Identification M. Poole, J.L. Jonas, A.W.V. Poole, M.P. Roberts. Abstract— The Telecommunications industry in Southern Africa is faced with the problem of theft of the signal carrying copper wire, both from the ground and from telephone poles. In many cases, if the offenders are caught, the prosecuting party has no way of proving that the wire is the property of any one Telecommunication company, as any inked markings on the PVC insulating sheaths have been burned off along with the PVC itself. This paper • describes the problem • specifies the technical and preferred properties of a feasible solution • reports the preliminary investigations into the devising of an unambiguous “fingerprinting” of the ∼ =0.5 mm wires, including some impractical and impossible solutions • describes the development and implementation of an electrochemical marker which has been shown to work at 1 m/s. Keywords— Copper Copper wire.
fingerprinting,
Electrodeposition,
3 times the value of the wire, and that following the expiry of Telkom’s exclusivity rights in May 2002, service disruptions (for all reasons including theft) would translate into loss of custom - not just immediate loss of revenue. There is no off-the-shelf technology for marking the wire, and thus the technology may have commercial value to any competitors wishing to penetrate the South African market. When the problem was specified, it was made clear that direct theft prevention (i.e. “guarding”) systems were not to be of any concern to this research. Telkom felt that a successful prosecution of an offender should be the priority, since word would spread amongst the criminal fraternity that legal protection now existed. A. About the manufacturing process
I. Introduction
T
HE development of a comprehensive wire identification technique has become necessary in recent times. Disruptions in telecommunication service provision arising from copper theft have been reported in the press as having grown to around 1700 cables just in March of 2001[1]. A cable is a collection of between 10 and 2400 pairs of individually insulated (with coloured PVC) wires bound with wax paper and housed in a larger black PVC sheath (often with an Aluminium inner sheath for the larger cable. An example of a larger cable can be seen in figure 1).
Fig. 1. A photograph of a cross-section of an example of one of the larger cables. The outer black PVC is clearly visible, and the inner aluminium sheath is visible between the black PVC and the wax paper. Enclosed within the wax paper are the individually insulated and coloured conductors.
It is estimated that the cost of replacing cable is around Corresponding author address: c/o Department of Physics and Electronics, Rhodes University, Grahamstown, 6140. Email:
[email protected] Tel: +27 +83 225 5096 Fax: +27 +46 622 5049
Fig. 2. A schematic overview of the extrusion process, where the input 7 mm (or sometimes 3 mm) rod is drawn down to the final diameter and coated with insulator.
A cable manufacturer receives Cathodic Grade A Copper (the copper used to make copper wire as specified by the London Metal Exchange) in the form of a 7 mm or 3 mm rod (represented by A in figure 2) from the sole supplier in South Africa: Palaborwa Mining Company (PMC), and begins the process by extruding the rod through consecutive diamond dies (represented by C in figure 2) of decreasing diameter until the ultimate desired diameter (e.g. 0.5 mm) is achieved. As the rod becomes thinner, the wire moves faster with the latter stages moving at approximately 40 m/s. Following the extrusion, the wire is annealed at high current (no representation), cooled and passed through an aperture where a specially pigmented (red, green, white etc.) molten foam which cools and clings to the surface to act as insulator is introduced (represented by E in figure 2). The completed insulated wire is rolled onto a bobbin. The wires are then pulled off the bobbins and combined into twisted pairs; a collection of 10 twisted pairs (or 20 conductors) being a unit. Units are later combined to form a cable with an appropriate number of pairs; the final binding being the PVC sheath (with optional inner sheath of aluminium). The manufacturers do not only supply one Telecommu-
nications Company (Telco), and thus any addition to the manufacturing process would have to be configurable for easy enabling and disabling.
inhomogeneities were eliminated. This implies that chemical composition can not be employed to identify copper already produced or in the ground. The focus shifted to-
B. Describing the crime As reported in the media, “man-holes” are often the point of access when the wire is not cut directly from overhead lines. Once a section of copper has been cut (often at night), the cables are burned on a simple wood fire (with optional petrol catalyst). This serves 2 purposes, namely to remove any existing identification markings on the PVC sheath, and remove all the insulation and PVC too. This leaves bare, unidentifiable wire (see figure 3) which can be sold to a scrap metal dealer.
Fig. 3. A photograph of copper cable after burning on the wood fire. Note that there are no markings distinctive to any company (such as logo etc.) visible.
II. Properties of a solution Any viable solution would necessarily have to be N.1 fire-proof N.2 unambiguous - each fingerprint must be unique N.3 legally admissible. N.4 unobtrusive in the existing production systems so it can be disabled easily when not desired. A viable solution would preferably have the following desirable properties P.1 be detectable in the field (i.e. out of a laboratory) P.2 be easily and inexpensively implementable. III. Investigations and rejected solutions A. Mechanical Deformation An obvious first thought is to “scratch” a fingerprint (mechanically deform) on the surface of the wires. However, since the wires are extruded at approximately 40 m/s, any extra stress or impulse applied to the wire could very easily cause the wire to “snap” - thus scrapping an entire run. Mechanical deformation would have to occur at position D in figure 2. B. Chemical Inhomogeneities (intrinsic and artificial) One of the first branches of the investigation was to see whether there is any inhomogeneity within and between “batches” of raw material which would intrinsically fingerprint the wire. Preliminary electron microscopy (see figure 4) showed that the purity of the copper is extreme, and that non copper elements, when detectable, were in sufficiently low concentrations and that no field device would have detection limits suitable for this task; thus intrinsic
Fig. 4. Composition of 3 mm copper rod. Note that the concentrations will not add to 100% since the elements were not all detected every sample, and the averages are only of the times the elements were actually detected with 99% confidence. Thus the copper seems to have a lower concentration than would be expected (actual average composition is always in excess of 99.95%) and the other elements seem to have higher than expected composition. Averages over all the samples, including the times when the other elements were present below the detection limits of the microprobe would reflect this.
wards trying to artificially enhance the concentrations of the non-copper elements (henceforth referred to as impurities). It was considered whether it would be possible to deliberately and methodically raise the concentrations of the impurities so that they could be later detected. Assays of PMC’s copper were analysed and it became clear that any contamination would have to be an non-metal element because the trace metals present had concentrations near to the maximum specified in the London Metal Exchange specification for Grade A cathodic copper. Despite the concentrations of the non-metal impurities being more than one order of magnitude lower than the maximum specified, any attempt to augment these impurities could result in • reduction of the company’s copper value • serious modification to the existing industrial system (in violation of property N.4) • difficulty with – controlling homogeneity of impurities (avoiding “clumping”) – recording and tracing the correspondence between chemical impurity and product (preference P.2) – Copper having poorer electrical characteristics – toxicity of candidate impurities, since the elements having super-low (ppb) concentrations (e.g. As) are toxic. – designing field test equipment (preference P.1) C. Modifying the insulation The insulating layer encasing each individual wire is made of PVC mixed with a colourant called Masterbatch and some curing agents. It is introduced to the wire at the point E in figure 2. The PVC is molten, and expands as it cools when exiting the nozzle represented by the chevron (at point E) in the figure. A short while after beginning
to cool it solidifies and shrinks resulting in strong cohesion within the insulator, but negligible adhesion to the surface of the wire. PVC combusts exothermically at around 137C, and thus burning off the plastic is simply achieved with an ordinary wood fire. It is plausible that some sentinel contaminant (metals, salts etc.) introduced to the molten PVC mix, would alloy with the surface of the copper as the insulation is burned off. This possibility is being investigated in a related study. D. Active surface treatment More recently, the investigation has been focused on electrodeposition of another element (nickel in this case) onto the surface of the copper wire. Electrodeposition has wide application in technology, particularly in microelectro-mechanical systems (MEMS)[2]. Nickel and copper have similar electrical and physical properties (thus satisfying property N.1) with the one important distinction that nickel is ferromagnetic whereas copper is not. In accordance with properties N.2 and N.3, the nickel is to be deposited (with a thickness t) in a binary manner with the presence of nickel representing a logical 1 and no nickel a 0. For this to occur, an inert anode, together with the (earthed) copper wire acting as a cathode, are present in a nickel-rich electrolytic solution (e.g. NiC2 (aq) ) as an electric current passes from anode to cathode through the solution. The plating will only occur when the current is present, thus “bits” can be set by either applying the current or withholding it. Each wire (of diameter d) is plated with an n-bit code (allowing for 2n distinct “fingerprints”), repeated over the entire length of the wire every metres. The method is proposed to be introduced at point D in the extrusion process (figure 2). Figure 5 shows one complete 10 bit pattern (not drawn to scale). These assumptions are
current For a given thickness t of nickel, the volume V of a bit (of width w = /n) is V = π[(d/2 + t)2 − (d/2)2 ]w = π[t2 + dt]w
(1)
The number of nickel atoms per m3 of metal is NN i =
ρ NA = 9.125 × 1028 m−3 MN i
(2)
where ρ is the density of Ni metal, taken to be 8906 kg/m3 , MN i is the molar mass of Ni, taken to be 58.693×10−3 kg/mol NA is Avogadro’s number, 6.023×1023 mol−1 1 electron = 1.602×10−19 Coulomb. Each atom requires 2 electrons, so the charge required per m3 is Q = 2 × NN i × 1.602 × 10−19 = 2.924 × 1010 C/m3
(3)
Each bit must be plated over a time τ = w/v seconds. This means the required average plating current, assuming 95% cathodic efficiency [4] is given by I=
Q × V. τ × 0.95
(4)
Substituting V in from (1) I=
Q × v.π[t2 + dt]w = κv[t2 + dt] w × 0.95
(5)
where κ has a constant value of 9.7×1010 C/m3 . Furthermore, since dt t2 (for a 0.6 mm wire, dt ≈ 106 × t2 ) we can ignore the t2 term and say I =κ×v×d×t
(6)
which for The period τ of each bit is given by w/v = nv n = 64 bits, = 1 m and v = 40 m/s is approximately 260 µs. This means the current will have to be switched at a frequency (“bit-rate”) of 3.846 kHz.
D.1 Proposed plating technique
Fig. 5. A representation of an arbitrary 10 bit sequence. The darker gray represents the copper wire of diameter d, the lighter grey represents the nickel plating of thickness t. In practice there will probably have to be some “start” and “stop” bits to enable reading of the code - this figure is by way of example only
made in what follows: • The “bits” are cylindrically symmetrical • The electoplating is uniform • The electrical characteristics of the wire will be negligibly affected 1 • There will be no electroless plating when there is no 1 “Electroless plating uses a redox reaction to deposit metal on an object without the passage of an electric current.”[3]
Consider equation 6. For v = 40 m/s, t = 1 nm, and d = 0.6 mm, the required current is approximately 2.2 A. The most convenient and efficient way to achieve this is by employing a current source2 . For the following discussion, figure 6 refers. The current produced by the source is passed through the load to cause plating (logical 1), and shunted to ground via a power MOSFET for a logical 0.3 This has the advantage of eliminating thermal cycling in the sense resistor and the “settling time” of the source. Many of the circuit ideas were borrowed from The Art of Electronics[8]. 2 A current source supplies a fixed current to any impedance within its compliance. The benefits, design and refinements of the current source took some time and research, and are discussed at length elsewhere. 3 Thanks to Spehro Pefhany [5], Winfield Hill [6] and Tony Williams [7] as well as the sci.electronics.design Newsgroup on Usenet for help and guidance here.
Fig. 6. A representation of how the prototype plating takes place. The microcontroller switches the current flow by controlling the MOSFET. When the MOSFET is ON the current is shunted to earth, and when OFF the current passes through the NiC2 causing plating on the wire. The wire is earthed inherently. The microcontroller (Atmel At90S8535) monitors the current source’s output current via its analogue to digital converter (A2D).
Preliminary tests with I = 1 A, v = 1 m/s, d = 0.4 mm are presently being analysed with an electron microprobe. Presently research is focused on simulating the industrial production conditions facilitating tests at 40 m/s. Once refinements have been made in the laboratory, the equipment will be tested under real industrial conditions. D.2 Selection of the electrolyte NiC2 .6H2 0 dissolved to the highest concentration in water (approximately 254 g/L at 20deg[9]) was selected as the electrolyte. Simple conductivity experiments revealed that Nickel Chloride is more conductive than Nickel Sulphate, in other words a lower voltage required for the same current and geometry, at the same concentration, (corroborated by [10]) and can furthermore be mixed to higher concentrations. pH is not deliberately manipulated in any investigations so far since introduction of acid into the industrial environment may violate N.2 or prove difficult to contain, and may not be necessary. No brighteners, anti-pitting agents or manipulation of bath temperature are employed. Experiments with optimal bath ingredients will commence if deemed necessary. D.3 Proposed field equipment In accordance with desirable property P.2, a field device for detecting the presence of Ni is to be devised. A naked wire, plated with its binary code will be subjected to a rapidly alternating (~100kHz) magnetic field produced by applying an alternating voltage of the same frequency across the wire which forms the solenoid in figure 7. Since the nickel is ferromagnetic, it will be magnetised, whereas the copper will not. The same equipment used to magnetise the wire will be used to read back the code, but the system will work in re-
Fig. 7. A schematic of the proposed head for reading from and writing to the nickel-plated copper. The fringes (illustrated by blue arcs) of the magnetic field present in the gap in the magnet magnetise the nickel (represented by the black bands) . The copper will not be affected. There will obviously have to be some auxiliary electronics and interface to the microcontroller
verse. The alternating voltage source will be disconnected and a voltmeter read into the microcontroller A2D will be connected instead. The magnetised nickel bits will cause an alternating voltage as they pass the aperture, where the copper will not. Thus the code can be read back since when the magnetised nickel passes the aperture, the alternating magnetic field will induce an alternating voltage in the coil. There will be no voltage when the copper passes the aperture since it will not have been magnetised in the preceding process. Investigations here are focused on determining the minimum “bit-width” (w), or alternatively finding the maximum bit-density which the extrusion process can support. Factors likely to influence the resolution are the activation time of the solution, the dimensions of the anode and the geometry of the electric field between the anode and the cathode. The minimum magnetizable thickness, tmin will also have to be determined. IV. Conclusions and Discussion The problem of theft of copper wire in the telecommunications industry in South Africa is undeniable. A technique of fingerprinting is described as well as the set of properties, both necessary and desirable which such a solution would possess, is described here. Possible solutions that have been eliminated or deferred include • Mechanical Deformation due to the risk of compromising the integrity of the wire (breakage) • Artificial doping of the copper with contaminant sub-
stances for logistical and financial reasons Present investigations include • • •
Determining the minimum “bit-width” Developing a magnetic read/write head Industrial implementation and testing
We have shown that a microcontroller can be used to control the electrodeposition nickel in a desired binary pattern onto copper wire from a nickel chloride solution in a laboratory; at least at 1 m/s. No attempts to optimise the composition, pH and temperature of the solution have been made. The code is to be read back using an electromagnetic technique, but the possibility of chemical treatment or developing has not been eliminated.
Acknowledgments This work was undertaken in the Distributed Multimedia CoE at Rhodes University, with financial support from Telkom, Lucent Technologies, Dimension Data, and THRIP. This work would not have been possible without the guidance and suggestions of Prof. J.L. Jonas, Prof. A.W.V Poole, and Dr. M.P. Roberts. Thanks to Baron Peterssen and David Browne of Telkom S.A. Limited for their guidance, and who presented the problem in the first place. Aberdare Cables of Port Elizabeth deserve a special mention for their help in sharing crucial industrial information, notably Mr. Chris Huntly. A special thanks to Prof. Winfield Hill, Spehro Pefhany, Tony Williams, Richard Grant and Anthony Sullivan whose technical insight and electronic suggestions have made the plating possible.
References [1]
M. Soggot, S. Brummer, and D. Shapshak, Right Royal Scandal, Mail&Guardian, July 20th to 26th edition, 2001. [2] T. Fritz, W. Mokwa, and U. Schnakenberg, “Material characterisation of electroplated nickel structures for microsystem technology,” Electrochimica Acta, vol. 47, no. 0013-4686, pp. 55 to 60, March 2001. [3] Fred Senese, What is electroless plating?, http://antoine.fsu.umd.educhem/senese/101/redox/faq /electrolessplating.shtml,
[email protected], 28 May 2002. [4] Payal, “Re: Problems with rate of electroplating,” Newsgroup, http://www.finishing.com, Charlotte, NC, USA, June 2001. [5] Spehro Pefhany, “Personal communications,”
[email protected], 2001. [6] Winfield Hill, “Personal communications,”
[email protected], 2001. [7] Tony Williams, “Personal communications,”
[email protected], 2001. [8] Paul Horowitz and Winfield Hill, The Art of Electronics, Cambridge University Press, The Pitt Building, Trumpington Street, Cambridge CB2 1RP, United Kingdom, 2nd edition, 1998. [9] http://avogadro.chem.iastate.edu/MSDS/NiCl2.html, “Nickel (ii) chloride hexahydrate,” 28 May 2002. [10] George Di Bari, ASM Handbook, vol. 5, ASM International, Materials Park, OH 44073, 1994.
M.Poole is researching for an MSc. in Physics with Electronics in the faculty of science at Rhodes University, Grahamstown. He completed his honours in 2000, and has been conducting research for Telkom since then.