CONTENTS

 

 

pg

Overview 1

Origins of CPD 2

History of Metal Rectifiers 6

CPD at Oxford 6

CSO and CTB on Return Path 10

CPD on Forward Path 15

Monitoring CPD 19

The Day-Night Effect 20

Method of Identifying and

cleaning up CPD 24

Semiconductor Junctions 25

Oxidation of Metals 27

Types of Corrosion 28

Conclusions 30

Other Personal Observations 32

Bibliography 33

Acknowledgements 34

 

 

CHARACTERISATION OF COMMON PATH DISTORTIONS

 

OVERVIEW

For several years now the Return Path has been seen as a source of new income by majority of Cable operators. From being an add-on to the forward path it has now become an integral part of a CATV network. Health of Return Path is a major issue and though occupying only a fraction of bandwidth to that of the bigger brother, the Forward Path, it requires special treatment and more elaborate maintenance procedures. It has also proved itself to be prone to ingress, noise and the dreaded Common Path Distortions. This article discusses CPD, the least understood of the effects and very commonly observed on Return Paths. It normally presents itself as beats at 6/8 Mhz interval on majority of networks on both sides of the Atlantic. (6 Mhz in the NTSC and 8 Mhz on PAL B/G/I systems)

It has been observed on many networks in the UK. ItÆs exact origin has baffled many who has seen it for the first time and has been blamed on anything from stinger lengths to saturated ferrites in the splitters and taps. ItÆs unpredictable nature with no apparent correlation to anything has made matters worse. The fact that beat intervals correspond to the intervals of the Forward Path carriers, suggests a relationship, but this where the story ends for most. A study was undertaken to clear some doubts and to try to characterise the behaviour of CPD. Though it is an undesired effect on any network , which needs to be eliminated as quickly as possible, it was thought

useful to observe its behaviour. Result of which could throw some light on ways of alleviating the problem. Indeed, the study has brought to attention some useful observations, which are discussed in the forthcoming sections. This article highlights some of the characteristics of CPD: itÆs origins, composition, and some other behaviours which were monitored over several months. It is hope that this will increase awareness on dangers of having CPD on a network. Most of the observations and records were made on a network with IRC raster. However, these provided a unique opportunity to study CPD in details which would not have been possible on a network with an HRC raster. On the IRC network, the components of CPD, that is, CTB and CSO could easily be identified. This has shed some further light on the behaviour of the diode effect. Figure 1 is a Return Path spectrum showing the multiples 8 Mhz CPD products and an alarm carrier at 6.5 Mhz. Clusters of other distortion beats at 8 Mhz intervals are discussed later.

Influence of CPD on the Forward Path has been overlooked by many workers in this field. Because it is easily ævisibleÆ on the Return Path, it has been regarded almost exclusively as a Return Path Phenomenon. CPD plays a major part in Forward Path distortion. This has been brought to light in this study. The effect also explains some of the other curious day time-night time differences in picture quality that is often experienced in majority of networks across UK and elsewhere.

 

 

 

Figure 1

 

Origins of CPD

As the name suggests CPD occurs on the part of the network common to the Forward and Return Paths; more specifically at the connector interfaces in the coaxial plant. In a typical tree and branch network there can be hundreds of RF contacts on the rigid coaxial plant alone. Figure 2 is an example of a sequence of metal to metal contacts that are normally made between the centre conductor of the cable and input/output terminals of amplifiers. There are also other instances of such contacts, particularly at power passing taps and splitters where CPD can occur. These CPD hot-spots can also be found at input and output F-connectors of 8 and 16-way tap plates. It is a result of a combination of undesired processes that take place at such passive junctions and normally come into play after some incubation period during which a non-ohmic metal oxide cell is set-up. This is commonly referred to as the Diode or Bimetallic effect. Of these the Diode effect is probably the most appropriate generalised description. An ohmic device is one which has a linear I-V characteristic such as a resistor and a non-ohmic device has a non-linear or asymmetric response such as that of a diode curve show in Figure 3. Such non-linearities at contact surfaces give rise to CPD.

In microscopic terms connector surfaces have irregularities and numerous crevices. Figure 4 shows a typical male female interface. Such crevices are largely present on the threads of F-connectors where the peaks and troughs often have sharp edges. Any coating at such points is easily damaged by the mere action of threading on a connector. Unfortunately the F- connectors are seen as low cost items by the industry and hence are not precision made.

 

 

 

CABLE Cable - M terminator F - Stinger Amp.,

Tap,

Splitter





Figure 2

Figure 3

Threads are often left with rough edges where corrosion can start easily. Added to that the protective metal coatings on many connectors are insufficient and of the order of 2-3 microns only. Any superficial damage to this coating reveals the base metal which in presence of moisture and other corrosive materials accelerate the process. This is accelerated further by to changes in ambient temperature which promote æbreathingÆ through the minute gaps between the threads. Moisture in the air will tend to get trapped in the crevices from where evaporation is difficult. Corrosion cells thus set up accelerate the bimetallic effect. Threads on both the male and female F-connectors were in fact seen to be prone to sever oxidation.

Female socket

Crevice

Male Pin Crevices

 

 

Figure 4

Under oxidising conditions these crevices are the initial points of attack. Given sufficient time, depending on external conditions, pockets of metal oxide cells will be formed at the connector interfaces. These oxides which may or may not be easily visible to the naked eye present potential barriers similar to those in semiconductor junctions. It is well known that oxides and sulphides of most common metals are semiconductors even in their polycrystalline or amorphous forms. Semiconductor junctions formed from such oxides are very crude form of diodes which exhibit extreme sensitivity to changes in mechanical pressures and temperature. They are akin to the original Point contact diodes behaviour of which was unpredictable and depended on mechanical conditions applied at their contact surfaces.

The author believes that Skin effect plays a major part at points of oxidation. Due to skin effect the Forward path RF currents are virtually at the surface of the conductors and are spread or distributed according to the surface irregularities at the contact. There will be points of high and points of low current densities set up at these surfaces which alter with time as metal oxides form in their paths. At some points these currents will be able to tunnel through some of these non-ohmic oxide barriers, giving rise to CPD. There is ineffect a flow of RF currents, some through the ohmic and some through the non-ohmic parts of the same connector.

Oxidised connectors of the type discussed above were not the only points where CPD was observed. CPD was also tracked to connectors contaminated by carbonaceous (organic) materials such as oils left behind during the manufacture of connectors. These apparently thin insulating materials between metal contacts can also form diode like junctions or cells. [1][3] Figure 5 shows an offending connector of this type where the centre male pin has lost its protective coating of tin to expose the unclean surface of carbonaceous material. Photo was taken under an electron microscope. Figure 6 shows another situation where the stinger terminal of an amplifier had collected contaminants, filling the space (crevice) between the round stinger surface and the flat terminal. Two tracks can be seen on either side of where the stinger was resting. On the same connector carbonaceous material can also be seen around the hole which normally supports the clamping screw. This material was thought to be a residue from the bath to clean the solder flux after the circuit board had been mounted. Both the above connectors were from the rigid coaxial plant; more specifically at the input/output of amplifiers.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Flacking of protective coat (tin) revealing uncleaned surface

Figure 5

 

Carbonaceous deposits on stinger terminal of an Amplifier

Figure 6

HISTORY OF METAL RECTIFIERS

That metal contacts can produce asymmetric conduction was observed more than a century ago. The earliest publication which can be traced relating to asymmetric conduction (diode effect) is that of Braun in 1874. [5][6] He noticed that the resistances of certain metallic sulphides appeared to vary with direction and magnitude of current passing through them. In the same year other workers (Shuster and Siemens) also observed that the irregularity was due to æsomeÆ property of contacts!

Though this phenomena of asymmetric contacts was observed well before the advent of semiconductor physics it however remained a mystery and without convincing explanation right into the 1930s when Schottky, Davidov and their colleagues came up with the space charge theories. The early observations were nevertheless put to practice and the rectifying diodes were manufactured using Selenium and later (1927) Copper oxide. Much work was carried out on oxides and sulphides up to then for use as semiconductors. However this was abandoned in preference to the new comers - Germanium and Silicon.

Metal oxides left the realms of semiconductors and was increasingly studied as the undesired corrosion product. It is curious to find that the numerous books and study of corrosion and oxidation of metals make use of junction and other semiconductor physics to explain oxidation rates and processes. It stands to reason then to assume that corrosion and the diode effect are related.

CPD at Oxford

Section of a typical Return Path is shown in Figure 7. Figures 8 to 11 are displays of return path spectrum of various nodes as seen at the Head End. Beats at multiple of 8 Mhz were prominent on all nodes. Because an IRC raster was used in the network, it gave distortion products on and around the multiples of 8 Mhz. This is discussed in the next section. The 6.5 Mhz carrier is the alarm carrier typically adjusted to read 30 dbmV.

All distortion beats on the return path were due to non linearities on the supposedly passive part of the network. That is, on the common path outside the confines of the diplexers. Thus the distortions can go forward as well as backward depending on the bandwidth open to them. (0 - 50 Mhz for return path and 70 - 860 Mhz for forward path)

Distortion products from the forward path amplifier should only be found in the forward path spectrum. It is important that the diplexer filtering is adequate so that those beats below 50 Mhz do not pass through the High pass filter and return via the Low pass filter. Figure 12 is a simplified electrical schematic of a part of a network showing forward and reverse amplifiers , H.P. and L.P. filters and the diodes (non-linearities) at connectors between amplifiers and at a splitter array. In the Oxford network CPD was found at points shown by diodes in the figure. It can be seen that any distortions produced at these non-linearities will go forward or reverse via appropriate diplexer branch. These distortions will æaddÆ to those produced by amplifiers in the forward path and as

such would be impossible to differentiate the two. On the return path however the CSO and CTB (see below) distortions are uniquely due to the common path non-linearities. These are easy to monitor and has been an effective maintenance check for keeping the network æcleanÆ.

 

 

Figure 7

The Ideal Return Path Response

Figure 8

 

CPD on Return Paths of an IRC Network

 

 

Figure 9

 

Figure 10

 

RESPONSE FROM A SUB-HEAD END

 

Combined response from 25 nodes. H.P filtered to reintroduce the 6.5Mhz Alarm Carrier. CPD still above the noise floor.

Figure 11

 

Figure 12

 

 

 

 

 

CSO and CTB on Return Path

When carrier levels are sufficiently high the non-linearities on a network produce distortion products in the form of CSO and CTB. These distortions are generally spread throughout the spectrum of operation and beyond in 8 Mhz intervals. Number of beats and levels of these distortion products depend on the type of non-linearity and frequencies of the carriers. CPD hot-spots also produce such CTB and CSO beats.

In a network with HRC raster these beats fall at the same frequencies i.e. at exact multiples of 8 Mhz. In an IRC raster these beats are off-set around the multiples of 8 Mhz.

The IRC raster used in Oxford presented a unique opportunity to observe composition and behaviour of the distortion products. This was easier to observe on the return path spectrum. Equations for CSO and CTB products for IRC raster are given in table 1. Only those marked by asterisk plays a part in the return path (0-50 Mhz). Beats due to other combinations falling outside the band and on the forward path. Figure 13 is a return path spectrum around 24 Mhz showing all distortion products; including those due to aural carriers. Vision to Aural carrier level was 17 db. Corresponding ratio was also reflected in the levels of distortions. Clusters of beats similar to those in Figure 13 occurs at all multiples of 8 Mhz (Figures 8 to 11) Compare that to figure 14 from one of HRC raster where all distortions due to vision carriers fall at multiples of 8 Mhz. All CSO and CTB contributions due to vision carriers and mixture of vision and aural carriers can be identified in the former.

Some nodes however showed predominantly CSO beats and others CTB. Typical examples of which are shown in figure 15. This seems to indicate differing stages in the development of the diode effect. There was in fact strong evidence of predominantly CTB beats from ænewerÆ nodes as opposed to approximately equal levels both CTB and CSO. On some nodes there was total absence in CSO beats. Reasons for this variations is no doubt complex is outside the scope of this article.

At the time of this study no digital carries were present on the forward path. These are planned for on majority of networks in the near future. As these will be only 10 db below the analogue carriers it now becomes easy to predict the consequence launching Digital carriers on a network with CPD. Their distortion products will also be æonlyÆ 10 db below the beats due to the vision carriers. The return path spectrum could look like that shown in Figure 16. Digital carriers will in effect raise the ænoiseÆ floor to unacceptable level. It is however difficult to carry out any mathematical analysis as the behaviour of these diodes is so unpredictable. Elimination of CPD from the network will be a must for reliable return path communications.

 

 

 

 

 

 

 

 

 

 

 

 

Table 1

IRC channel raster at Oxford : Carrier frequencies at n8-0.75 Mhz

CSO

F1 + F2 : (n8-0.75) + (m8-0.75) = (k8-0.75) - 0.75

Beats at -0.75 Mhz of the carriers

* F1 - F2 : (n8-0.75) - (m8-0.75) = (k8-0.75) + 0.75

Beats at +0.75 Mhz of the carriers

CTB

F1 + F2 + F3 : (n8-0.75) + (m8-0.75) + (p8-0.75) = (k8-0.75) - 1.5

Beats at -1.5 Mhz of the carriers

F1 - F2 + F3 : (n8-0.75) - (m8-0.75) + (p8-0.75) = (k8-0.75)

Beats at the carrier frequencies

* F1 + F2 - F3 : (n8-0.75) + (m8-0.75) - (p8-0.75) = (k8-0.75)

Beats at the carrier frequencies

* F1 - F2 - F3 : (n8-0.75) -(m8-0.75) - (p8-0.75) = (k8-0.75) + 1.5

Beats at +1.5 Mhz of the carriers

 

 

 

 

 

 

 

CPD AT 24 Mhz ON IRC RASTER - OXFORD

CTB CSO CTB CTB CSO CTB CTB CSO CTB

F1 - (F2+6) - F3 (F1 - F2) (F1+6)+F2-F3

n8-1.25 n8 n8+1.25

(F1 + F2 - F3) (F1 - F2 - F3)

n8-0.75 n8+0.75

F1 - (F2+6) (F1+6) -F2

n8-2 n8+2

F1+F2-(F3+6) (F+6) = Audio carriers F1-F2-(F3+6)

n8-2.75 n8+2.75

Figure 13

 

CPD AT 24 Mhz ON HRC RASTER - WARWICK

 

F1 - F2

F1 - (F2+6) F1 + F2 - F3 (F1+6) - F2

F1 - (F2+6) - F3 F1 - F2 - F3 (F1+6) + F2 - F3

F1 + F2 - (F3+6) (F1+6) - (F2+6) F1 + F2 - (F3+6)

(F+6) = Audio carriers

All CTB + CSO beats at multiples of 8 Mhz

Figure 14

 

 

 

CSO < CTB

CSO > CTB

Figure 15

Figure 16

CPD on Forward Path

As mentioned earlier CPD is much easier to observe on the return path in the absence of other carriers. On the forward path however, distortion contributions due to CPD is not easily appreciated. That CPD plays a major part in the sum total of forward path distortions will be highlighted in this section. Significance and consequences of this are discussed later.

Presence of CPD on the forward path can only be observed if a comparison of beats is done before and after CPD is eliminated. This makes it a tedious and time consuming process. Presence of carriers as well as distortions due to the forward path amplifiers make it difficult to distinguish distortion due to diode effects. These are for all purposes similar in nature to those produced by any other non-linear devices such as amplifiers operating at its limits. æEfficiencyÆ of producing CTB and CSO will differ from device to device, however for a given channel plan, number of beats produced at any frequency will be the same.

To record distortion due to CPD a splitter output (subscriber drop) exhibiting diode effect was sought. This was done by monitoring the return path and using method of elimination starting at the node centre. An offending cabinet, and in particular the input F-connector on a 16 way splitter array was identified as the source. Due to their ætouchyÆ nature, care had to be taken so as not to break the diode cells. Level of carrier at 127.25 Mhz was recorded as 24dbmV. All FM carriers and the first 2 channels (127.25 & 135.25 Mhz) were then switched off to get a æcleanÆ part of the spectrum. A low pass filter with cut off at 168 Mhz was used so as not to saturate the analyser. Spectrum of the forward path from 70-130 Mhz was recorded without disturbing the connector. Record was also made of the spectrum at 127.25 Mhz to see the beats in more detail. Offending F-connector was then cleaned using a contact cleaner and the spectrum re-recorded. Figures 17 and 18 show these traces. Beats resulting from the diode effect is clearly evident in Figure 17. Due to IRC raster some of these beats are off-set from the carrier frequency. (See Table 1)

From Figure 17, (with CPD)

CTB was:

24 - (-28) = 52 dbc at 127.25 Mhz

and

24 - (-33) = 57 dbc at 128.75 Mhz.

CSO was:

24 - (-34) = 58 dbc at 128 Mhz

From Figure 18, (without CPD)

CTB was:

24 - (-38) = 62 dbc at 127.25 MHz

and

24 - (<-40) = >64dbc at 128.25 Mhz

CSO was:

24 - (-34) = 58 dbc at 128 Mhz

It can be seen that CPD was responsible for considerable distortions on the forward path. CTB had decreased by 10 db at 127.25 Mhz and by more than 7 db at 128.75 Mhz after cleaning the connector . CSO however had remained the same in both cases.




Forward Path with CPD: High-lighted area at 127.25 Mhz carrier

Figure 17




Forward Path without CPD : Beats only due to Amplifiers

Figure 18

This is the distortion due to amplifiers. Difference in distortions of this magnitude will certainly reflect on the picture quality, particularly in the case of IRC raster.

Beats due to vision plus aural carriers are also just visible at +/- 2 Mhz. in Figure 17. Vision to aural ratio was 17 db.

In the above test case, CPD had produced primarily CTB and non or very little CSO. This is in accordance with earlier observations on the return path which in many cases showed presence of CTB only. (pg. 10)

In an HRC network all distortion beats concentrate at the carrier frequency giving higher than expected jump in distortion.. Though this may not be ævisibleÆ with existing analogue carriers, with introduction of Digital carriers the consequences could be serious.

In networks with analogue only carriers, above mentioned distortion products are generally æhiddenÆ under the vision carriers. (Better in HRC raster and less effectively in IRC raster).

Distortions due to digital carriers will however produce considerable ænoiseÆ across the analogue bandwidths assuming 10 db back-off for Digital carriers. (This is much less than the 14 -17db back-off of aural carriers which still produced distortions.) There will thus be the narrow band CPD as observed above from analogue-analogue combinations and broad band from digital-digital and digital-analogue combinations. Depending on gravity of CPD, this can lead to considerable deterioration of both the analogue and digital carriers. Analogue signals may go ægrainierÆ,

but the digital carriers will succumb to the cliff effect much earlier.

Figure 19

Fig. 19 shows possible BER vs C/N curves for a clean and CPD infested network. The cliff effect will be even more pronounced on cold days as will be seen in the following sections. In such cases decreasing the digital and or analogue carrier levels from the normal by a few db could be an option. As this will also reduce CPD levels by two to three fold. More labour intensive but better option is to eliminate CPD.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Monitoring CPD

When monitored on the Return Path variations in levels of were observed to fall into three categories.

  1. Constant Dynamic changes.
  2. Random changes.
  3. Long term temperature

dependent changes.

1. As stated earlier, very dynamic level changes were observed on the return path. These were more pronounced on the HRC (Warwick) network than on the IRC (Oxford) network. All carriers on the HRC network had been synchronised. As a result, phase differences in beat frequencies would also be in a very narrow rang with greater chances of being either in phase or out of phase. It is in fact sum of beats after delays between each distortion point, either due to CPD or other active devices in the forward and return directions. In the IRC network where carriers were not synchronised, beats were distributed over wider frequencies giving less pronounced jumps in levels.

2. Random changes in levels including relative changes in CSO and CTB beats had no apparent trend and is thought to be a likely result of mechanical changes at the contacts. Non-linear region (Barrier layer) of a semiconductor junction is a very small region, only a few microns across. Minute dimensions involved means it can be easily upset by small changes at the contact surfaces. As most cabinets are installed on road sides, any mechanical stresses or vibrations caused by heavy traffic could alter these non-ohmic regions. This possibility was easily proved by lightly tapping on offending splitter arrays or even cabinet shells and observing the beat on the return path. These changed randomly with large swings; very often clearing out completely only to reappear later.

Daily (day and night) temperature changes which alter stresses on the cables will also have an effect on the connector surfaces. These changes can easily make or break diode cells which in reality are a crude form of diodes with behaviour akin to the original point contact diodes whose performance was unpredictable.

3. To observe temperature dependence of CPD, 10 nodes were monitored over several weeks at the Head End. Both CSO and CTB levels at 24 Mhz and 24.75 Mhz respectively were measured. Temperature readings as at Breeze Norton near Oxford were noted at the same time. Readings were only taken where there was at least 20oo0o C change in temperature.

Figures 20 to 29 show graphs of CSO and CTB vs. Temperature for the 10 nodes. Apart from some random jumps mentioned in 2 above, there was an overall trend of decreasing CPD with temperature. Other observations also come to light.

Most cabinets in the network housed amplifiers, which would have increased internal temperatures. Though the temperatures recorded were ambient, this would have reflected the relative temperatures inside the cabinets. Inverse temperature dependence of CPD indicates that it will be at its worse in the winter months than summer. Hence incidence of problems with the return path and cliff effect on digital carriers will be more during the colder spells! Any reliability and maintenance issues should take this into account.

Decreasing CPD with temperature follows the general temperature dependence of any crude diode I-V characteristics shown in Figure 30. [3-pg1802, 4-pg364, 7]. In such diodes as the temperature is increased more conduction takes place due to increased electron tunnelling and hence a reduction in non-linearity.

200 100 00 C

I

 

 


V

 

Figure 30

 

 

 

 

 

 

The Day-Night Effect

It is widely believed that the day time and night time difference in picture quality is due to sensitivity of the eyes which is more at night and so will see all minor artefacts on the picture. Various other reasons have also been put forward but remained far from being convincing. Most franchises have suffered from this phenomenon and Network Engineers and Technicians have often been cornered by subscribers for an explanation.

After studying behaviour of CPD with temperature changes the day-night differences in picture quality is to be expected as there is an increase in levels of CPD when temperatures fall during night time. This effect can now be explained with reference to CPD on the forward path.

Subscribers complaining of picture quality or picture break-ups on the Digital channels at night should be taken as a strong pointer to existence of CPD on the network.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 20

 

Figure 21

 

Figure 22

Figure 23

 

 

 

Figure 24

 

 

 

Figure 25

Figure 26

 

 

 

Figure 27

 

 

 

Figure 28

 

Figure 29

 

 

 

Method of Identifying and cleaning up CPD

Origins of CPD are always found at contacts. To change or clean such contacts involves down time. Thus a systematic approach is called for to carry out the work efficiently.

Where the offending contacts are few, trouble shooting should start from the node centre or from first amplifiers on each leg and work outwards.

On networks that has a high incidence of the diode effect, spread over all parts of the network, it is best to start trouble shooting from the outer most amplifier and work towards the centre. Cleaning out all contacts in the cabinet until the spectrum at the output of return path amp. is clean. It helps if effort has previously been made to identify CPD hot spots in the network. These hot spots are normally the first point of origin of CPD on the network. For example it was observed that diode effect was more common at 75 ohm F-terminals of an 8 way taps than at 16 way taps. As distortion is a function of carrier levels contacts with high carrier levels such as outputs of 8 way tap becomes a potential source of CPD. Added to that, in the reverse direction these beats also go through a lower tap/splitter losses. Similarly taps with highest input levels were also the first to show signs of CPD at the input F-connectors. On rigid coaxial connectors shorter run lengths showed higher levels of CPD due to higher carrier levels at end of these runs.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Semiconductor Junctions

Every contact involves at least two materials; three or more if interface barriers are present as in the case of contacts under consideration. In semiconductor language contacts are said to occur when two substances are brought together, while junctions occur when structural changes take place as a result. Region of these structural change is very close to the contacts and is known as the barrier layer. Distribution of charges in this layer sets up a potential barrier through which electrons cannot cross easily. These barriers are always present at interfaces between material of differing conductivity. Hence in theory it also exists at metal to metal interfaces. However in such cases the barrier height is negligible and play no part in the total conductivity. At metal-semiconductors such as metal oxides, (CuO û p-type, NiO û n-type, SnO û n-type) a potential barrier can be formed such that it will make the junction non ûsymmetrical to current flow. The junction is thus said to be non-ohmic. Semiconductor barriers can be formed by many metal oxides, sulphides, halides and also very thin layers of what is normally classified as insulators. Fig.31 shows a resistivity scale on which are indicated a number of familiar substances according to their resistivity values. Generally speaking those substances whose resistivities lie between 10 - 3 ohm-cm and 1010 ohm-cm are regarded as semiconductors. Many metallic oxides, sulphides, carbides fall in this category of semiconductors and will form barrier junctions when in contact with metals. Very thin (10 to 50 Angstroms) insulating materials between metal contacts can also form diode like junctions or cells. [1][3]

Barriers formed from such polycrystalline or intergranular structures are pressure dependent as this leads to deformation of surface layers. Also because oxide films can vary in thickness and structure a varying degree of diode effect may be exhibited.

Analysing junctions or barriers on a network is almost impossible and some degree of extrapolation has to be involved. Junctions created in laboratories will be easier to analyse than those created by freak as on any network. One aspect is totally clear: though contact effect may be confined to regions which are thin compared with total dimensions, they cannot be ignored. The extent to which contacts make their presence felt will depend on the material associated, temperature, contact area, RF current flow etc.

Practically all important phenomenon that takes place in solid-state devices like diodes, transistors, switching devices, etc. are at junctions or at transition region between the two different materials. To discuss the detailed physics involved at barriers is beyond the scope of this article. Reader is referred to many books and chapters on semiconductors that explain barrier and junction theories. Though most of the present day studies and observations concentrate on semiconductors such as silicon, germanium, Gallium Arsenide and others, the principles could be extrapolated to explain the diode effect observed on CATV networks.

A wide variety of diode characteristics can be achieved by æcontrollingÆ a microscopic area û the barrier or junction between materials. This point cannot be stresses enough as it forms the basis of understanding why it is important to maintain clean contacts on the network.

 

 

 

METALS SEMICONDUCTORS INSULATORS

 

 

 





10 -6 10-3 100 103 106 109 1012 1015 1018 1021

RESISTIVITY IN OHM-CENTIMETRES

Figure 31

 

 

 

 

 

Oxidation of Metals

This is said to occur when a metal is transformed to one or more of its oxide states. A number of publications concerned with oxidation and corrosion processes deal with the subject from both theoretical and practical point of view. A list is included at the end of the section. Most of the information in this section is derived from these and other books. The subject is vast and is taken seriously in a number of industries ranging from manufacture of semiconductor products to ship building, as it is often the failure of quite small components which can make a large capital installation useless!

Metals are derived from their natural state as oxides, hydroxides, carbonates, etc. by various metallurgical processes. These are in nearly all cases endothermic processes where energy or heat is given to form the metal. There is hence an intrinsic tendency for the metals to revert to their compounds and give off energy in the process. Fortunately the process of oxidation or conversion of metal back into its oxide is a surface chemical reaction only and therefor there are a number of ways of slowing down or stopping this reaction.

Processes involves in the oxidation will differ with external conditions. Localised oxidation is normally a result of setting up of corrosion cells on the metal. There are various ways this can occur and are discusses below.

Figure 32 shows a simple corrosion cell were an anode and a cathode has been formed due to one of the many physical and chemical conditions. Same as in the electrolytic cells the anode is the electrode to which the anions or negatively charged ions travel and the cathode is the electrode to which the cations or positively charged ions travel. Under this condition the electrons flow away from the anode to the cathode. The driving force for this current being the difference in electric potential between the anode and the cathode.

 

O2 from air

(considering air as an electrolyte)

 

 




oxide

Anode Cathode

e.g. iron flow of electrons e.g. copper


Figure 32

To determine which of the two metals is likely to become the anode and hence corrode (corrosion /oxidation is always at the anode) and which is likely to remain the cathode, reference is normally made to the standard electrochemical series. [9 - page 14] The series is however only a guide because there are other effects such as formation of passive films which can reverse the process and make an anodic metal to behave as cathode and vice versa. Thus such series can be misleading and a more practical solutions published by various establishments such as BSI or MOD should be consulted. [ BSI publication æCommentary on corrosion at bimetallic contacts and its alleviationÆ - PD 6484 :1979 is useful.

Types of Corrosion

There are numerous types of corrosions, each relevant to specific application. However when considering electrical contacts the following types are perhaps the most relevant.

1. Bimetallic corrosion

2. Single metal corrosion

3. Crevice corrosion

4. Stress Corrosion

Bimetallic Corrosion

Bimetallic corrosion is said to occur when two metals are in direct or in some cases indirect electrical contact and are also connected by an electrolyte, say air or water. Under such conditions and depending on metals, a corrosion cell will be set up. The net effect is to corrode the anodic member while the nobler metal, cathode, is protected.

In atmospheric conditions such as in a typical network the bimetallic effect is restricted to an area close to the contact (2-3mm). This is because air is a poor conductor and the ionic currents are restricted to an area close to the contact.

It has been suggested [10- pg 52] that metals can be divided into 4 main groups which in general gives a measure of bimetallic corrosion.

-- Magnesium and its alloys

-- Cadmium, Zinc, Aluminium and

its alloys

-- Iron, Lead, Tin, and their alloys

(except stainless steel)

-- Copper, Chromium, Nickel,

Silver, Gold, Platinum,

Titanium, Cobalt, Stainless

Steel, and Graphite.

No bimetallic effect is likely between metals in the same group, some effect is to be expected between metals in adjacent groups, and serious effect is possible between metals from non-adjacent groups.

Normally protection from bimetallic corrosion is obtained by metal coating one or both metals. This protection is however specific to the metals concerned and not universal solution, e.g. electroplating brass with say cadmium will give good protection in one situation but be corrosive in other. Hence what seems to be an ideal protection to one manufacturer could be rendered useless when coupled with ideal solution of another.

Single Metal Corrosion

Oxidation of single metal can be caused by two effects. One is by internal changes of impurities which can form an anodic and cathodic regions. The second however is more relevant to the situations encountered in CATV networks; when different sections of a single metal is exposed to different oxidising conditions, as for example when a stinger is clamped by a screw to the terminal post. Here a part of the same metal is exposed to the air while other part is covered and hence starved of oxygen. The differential aeration condition thus set up at the surface give rise to an anodic and cathodic sections in the same metal. [9 - pg21]

In such a situation the metal surface exposed to oxygen becomes the cathode of the corrosion cell, while the section starved of oxygen becomes the anode. The actual process of oxidation is complex. In short the excess electron at the cathode is picked up by the oxygen to become an negative ion and in turn is attracted to the anode. A process known as chemisorbtion describes the process in more details.[11 ]

Above process can be aggravated by existence of crevices at contact points. Non conducting substances pressed against metals are also known to form corrosion in single metals. [13 - pg61]

Crevice Corrosion

Crevices are deep laying sections which are starved of oxygen as compared to the more exposed parts. Thus a differential aeration condition also apply here. This type of oxidation is made worse if the crevice get filled with moisture that cannot evaporate. [9 - pg33]

Almost all types of contacts and interfaces encountered at input and output of amplifiers, taps and splitter arrays are potential sources of crevices as shown in figure 33 and 34. Apart from promoting oxidation they also act as traps for contaminants from the atmosphere.

Corrosion due to Stress

It is known that most highly stressed or deformed part of a metal is usually anodic to the remainder. [12 - pg52]. This form of corrosion could take place at mechanical situations involved in clamping stingers as mentioned previously. The internal stresses created by clamping the stingers could start the corrosion process. This kind of corrosion may also appear at cut edges where internal stresses have been produced by shearing or machining during manufacture. [10 - pg53, 13 - pg45]. On abraded items the point of attack may be arranged along the deeper grinding groves. On bent items metals attack may start at the bends.

One or all of the above corrosive or oxidising conditions may be present at connection interfaces in CATV networks. As mentioned the stingers and the F-connectors are particularly prone to oxidising condition, as all the above conditions; bimetallic, differential aeration, crevice and stress apply. Hence bimetallic corrosion is not the only type of corrosion effecting the connectors. More research and observation needs to be made to alleviate the phenomenon from all CATV networks.

 

 

Figure 33

 

 

 

 

 

Figure 34

 

 

Conclusions

It has been established that origins of CPD is at contacts. Observations and trials show that the effect can be severe on both the return and forward paths. On return path the effect may be ædrownedÆ in noise by the time all upstream paths are combined depending on severity of CPD. See Fig. 10. On forward path however the effect will be visible on all services with the introduction of Digital carriers. There will be no inherent protection as presently enjoyed in HRC networks. Many networks already show excessive levels of distortions with no apparent effect on picture quality and hence there is a tendency to overdrive the amplifiers. These networks might in fact suffer more severely as pointed out earlier due to wide band beats æconcentratingæ at carrier frequencies. CPD was seen to occur at much lower carrier levels than those required to produce similar levels of distortions from amplifiers. CPD will thus become the new limiting factor but with very low level of predictability.

Types of connectors which has given rise to CPD in Oxford has been identified. Other franchises will have to carry out work independently to find the hot spots in their networks.

Often connectors are one of the least understood and neglected part of a network. It is perhaps time, though late, to control the quality and in particular their behaviour when in contact with other metals. Connectors have been imported from all parts of the world and often their qualities have been driven by price competitions. It is hoped that with inputs from other franchises a pool of information of is gathered and made available to all franchises. Manufacturers should be expected to give recommendations and in general participate in alleviating the problem.

Independent of this a two pronged attack is called for if CPD is to be contained:

1. Keep connectors, both internally and externally, clean and free of contaminants. Methods of æCPD proofingÆ connectors should be introduced. This subject requires further study.

2. Avoid metal-metal contacts that promote bimetallic corrosion.

Cleaning materials for connectors of different shapes and sizes are virtually non existent. These should become part of technicians tool kit together with cleaning kit for fibre connectors. Essential cleaning of fibre connectors is widely recognised. Similar school of thought is also required for the metal connectors.

Though connectors in all U.K. franchises has some protection by virtue of being enclosed by cabinets it is forgotten that they have very little protection from contaminated air which is free to circulate within, creating its own micro-climate. Very often exhaust fumes, oils, general pollution and moisture from the ducts settle on the connectors and crevices. There was some evidence during the trials that incidence of diode effect was more in cabinets on roads with heavy traffic. Setting up of corrosion cells is easily promoted under these conditions.

Introduction of protective covers or flooding compounds for connectors on the rigid plant as well as F-connectors is called for. Protecting the æpermanentÆ rigid coax connectors with heat shrinks or amalgamating tape could be effective. During installation of main network, connector cleanliness should be a priority particularly avoiding sweaty/oily hands on critical parts of the connection chain. On same token manufacturers should introduce plastic caps on both the male and female connectors. This will give protection during storage and installation. Amplifier manufacturers could do well by ensuring that stinger terminals are free of oils.

F-connectors for drop as well as 75 ohm terminals are particularly prone to moisture ingress. Diode effect on threaded part of the 75 ohm terminal was responsible for CPD in a large no of cases. Diode effect here is only separated by 75 ohm resistor; hence distortions produced on these threads will easily appear on the return path.

High incidence of corrosion at splitter arrays calls for a closer study of electroplating on the threads of both the splitters and the 75 ohm F-terminals. Onset of corrosion on threads has been discussed earlier.

Effect of CPD with introduction of Digital carriers was highlighted. The study revealed in more details composition and behaviour of the distortions. Both CTB and CSO were identified as products of CPD.

Importantly its variation with temperature was recorded. Observations indicate that CPD is more likely to be present in areas of high humidity in association with large fluctuation in temperatures. These conditions are met in all networks in the UK. Cabinet enclosures trap moisture from underground ducts and temperature swing is fairly large during the summer months, made worse by combination of dark cabinet colour and heat from the amplifiers.

Stringent preventative measures, method of identifying CPD hot-spots on the network and keeping close check on supplier quality, outside the present practices, is needed to eliminate CPD.

 

 

 

 

 

 

 

 

 

 

 

Other personal observations

I have the benefit of being on the Oxford network and so am able to watch the pictures fairly regularly.

On several occasions during reasonably heavy down pours the picture quality has been seen to be much better. I have often wondered if this could be due to changes in conductivity of soil around the cabinets and tests have been conducted by earthing all cabinet backboards on the part of the network concerned with no avail. Now having played with CPD and studied its behaviour I am convinced the picture improvement is due to rain drops falling on the cabinets which in effect simulate sustained vibrations. This will tend to disturb and break up the diode cells at the connectors and hence reduce CPD.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Bibliography

 

 

1. Semiconductor Contacts - an approach to ideas and models. By Heinz K. Henisch. Clarendon Press. 1989

2. The Properties, Physics and Design of Semiconductor Devices. By John N. Shive. D. van Nostrand co.inc. 1959.

3. Generalised Formula for the Electric Tunnel Effect between Similar Electrodes separated by a Thin Insulating Film. By John. G. Simmons. Journal of App. Phy. V34 -6. June 1963. pg 1793.

4. Blocking Layer Rectifiers. - W. Ch van Geel - Philips Tec. Rev. V4 - pg100.

5. Metal Rectifiers - A. L. Williams and L. E. Thompson.

6. F. Braun - Ann. Phy. Pogg. 1874, 153, pg556

7. CATV Return Path Characterisation for Reliable Communications. - Charles A. Eldering et. al. IEEE. Comms. Mag. Aug. 1995. pg62.

8. Semiconductor Devices and Applications. - R. A. Greiner. pg140 McGraw Hill 1961.

9. The Prevention of Corrosion. - R. M. E. Diamant. Business Books Ltd., 1971.

10. Protection of Metals from Corrosion in Storage and Transit. - P. D. Donovan. Ellis Horwood Ltd.

11. Oxidation of Metals. - Karl Hauffe. Plenum Press. 1965.

12. Corrosion and Protection of Metals. - G. T. Bachvalov and A. V. Turkovskaya. Pergamon Press. 1965.

13. An Introduction to Metallic Corrosion. - Ulick R. Evans. Edward Arnold. 1981.

 

 

 

 

 

 

 

 

 

Acknowledgements

 

I would like to thank Ian Banner and Mike Thornton for supporting the project and special thanks to Kevin Horswell Oxford) and Graham White (Warwick) who provided facilities and help to carry out the work. Also thanks to John Pozzi, Trevor Belcher, Chris Smithson and Brian Lewis who helped out in the field and for showing patience in the, at times, painstaking work.