Network Cables Online Blog    

First the Basics
I’ve lost track of the number of times I’ve been on a customers site and found that they have re-used the old Cat 3 and Cat 4 patch leads simply because they didn’t have any Cat 5 leads at the time. These leads almost never get changed because it hasn’t made any noticeable difference to the operation of the network. Years later of course, things start to go wrong and the old leads aren’t even suspected because “they’ve not been a problem in the past”. By all means use old leads on voice systems but never use them on data networks. It is like throwing a load of rocks into a smooth running stream, the data will probably still get through at first but when you increase ‘the flow’, they will start to impede the throughput. If you need to to buy some new patch leads this link will take you to the online store of Network Cables Online.

If the problem only affects one PC, take that PC to the patch cabinet and plug it directly into the hub. This may seem like an obvious thing to try but it has to be said as it proves whether the fault is actually a cabling problem or not. You should use a known good port on the hub and a known good patch lead, if the machine works then the problem is either the original hub port, the patch lead, the drop lead or the fixed cabling. From here it is easy to eliminate each part of the link, but it must be carried out methodically, one component at a time. Another point worth mentioning is that a patch lead that works fine in a Token Ring network may not work in an Ethernet network, this is because they use different pins on the RJ45 plug. Token Ring uses pins 3, 4, 5 & 6 whereas 10BaseT Ethernet uses 1, 2, 3 & 6, so any cable with a fault on pins 1 & 2 will work for Token Ring but not for Ethernet. Although a Cat 5/5E tester can be very expensive, you can buy a simple continuity tester for under $100 which will test for shorts, opens and crossed pairs, this will not prove your cabling is up to standard but they are quick and easy for finding faulty Cat 5 patch leads.

Another thing to watch out for is the wiring configuration, there are actually two different schemes allowed under the 568A standard. These are called 258A (or T568B), and 258B (or T568A). Pin for pin they are the same but with the orange and blue pairs swapped over, so as long as you have the same type of jack at each end, no problem. However, if you have 258A on one end, and 258B on the other then you have a crossed pair.

Check the time of day!
The time of day may indicate another cause of network problems. If the problem only occurs at a certain time, it maybe that the network is slowing due to an increase in traffic say at 9:00am or 5:00pm. If, for instance, the drawing office starts at 9:00am and twenty draughtsmen are all trying to pull large drawing files from a server which is on the general network, this will impact the rest of the company’s business. Likewise at 5:00pm when they are all saving their work back to the server the sudden increase in traffic may cause so many collisions that the network to grinds to a halt. A process of elimination is easy to implement and if this is the cause, it is time to put the drawing office and the CAD server on to its own hub or switch.

Electrically ‘noisy’ environments
Another common cause is electromagnetic interference from electric motors and sources of high frequency radio waves. The more obvious things to look for are cable routes that pass too close to lift motors, arc and spot welders, heavy plant machinery which use large electric motors, and fluorescent light fittings. All of these things, if situated close to the data cabling, could induce spikes into the network. Take a walk around the building and make a note of all possible causes, and then try to eliminate them one at a time.

What next?
Eliminating possible causes one at a time is the ideal, but sometimes the problem occurs so infrequently that it is almost impossible to track down, and specialist equipment and engineers have to be employed. ‘Network Sniffers’ and mains monitors can be hired from companies such as Livingston Hire, and if you are confident that you can correctly interpret the results, do it yourself. If not, well its the end of the line and time to call in the specialists.

To summarize, the following considerations should be taken into account:

1. Low grade patch leads and/or drop leads.

2. Time of day.

3. Increase in network traffic from other sources.

4. Electromagnetic interference.

Ethernet
Most of the points mentioned above should find Ethernet problems, however, there are a couple of things that should be taken into consideration when dealing with Gigabit Ethernet.

Although Gigabit Ethernet was designed to run on 100MHz cable, problems may arise with older Cat 5 systems. The more stringent Cat 5E standards take into consideration that Gigabit Ethernet uses a four pair transmission method, but this was not part of the test parameters with Cat 5. If you are trying to run Gigabit Ethernet over standard Cat 5 cabling, then the whole system should be tested to confirm that it meets the new Cat 5E standard.

It used to be said that multimode fibre was good for 2km, but recently it has been found that for Gigabit Ethernet applications the length limit is right down to around 220m over 62.5/125 fibre. The only way to prove if a fibre is good enough for Gigabit Ethernet is to use a certification tool. These are fairly expensive test instruments, but you should be able to hire one from a specialist hire company. The results give a clear ‘pass or fail’ for different applications, but bear in mind that one dirty connector can affect the results considerably.

AS400 cabling
AS400e series have three new features which are designed to speed up throughput to 5250 devices, unfortunately some Twinax hardware will not support them.

Split Mode
When operating in Split mode, the AS400 poles its ports two at a time, i.e. 0 & 4, 1 & 5, etc., this only causes a problem when eight port multiplexors are used as they will be trying to multiplex and de-multiplex both simultaneous signals. To overcome this, two four port mux’s should be used, or the Split mode feature turned off.

Optimized Mode
In optimized mode no overheads are added to the data frame which under normal conditions works fine, however some star hubs and mux’s need the additional information to work correctly. The solutions are to either buy new hubs and mux’s, or again turn off the Optimized mode feature on the AS400.

Express Mode
This is simply the AS400 running at 2Mbs instead of 1Mbs, and again some hubs and mux’s can’t handle this speed. The solution?, same as above really, buy new hubs or switch off the Express mode feature.

Length limits
The maximum distance for Twinax cable on each port of the Work Station Controller is 1800m, this limit is reduced to 1458m when running in Express mode. It is also a little known fact that there is a minimum length limit of 25 feet or about 8m on all types of cable used to transmit 5250 signals. This distance is the length of the signal itself, and if the cable is less than this, the signal can become corrupted by its own reflections before it has finished transmitting.

When using IBM Type 1 cabling, the limits are the same as Twinax operation (1800m) but this reduced to 1312m when running in Express mode.

For UTP installations the distances vary according to whether Twinax is used on part of the installation. The limits range from a maximum of 364m when only a few metres of Twinax has been used between the Work Station Controller and the hub, down to a minimum of only 36m where more than 1600m of Twinax has been used. For further information visit the IBM networking web site.

Where fibre is used between mux’s or converters, the maximum limit is 2400m, again this may be reduced if the AS400 is running in Express mode.

5250 Star hubs
Some active star hubs can introduce delays into the system, if these delays are close to the limits for 5250 operation, it can affect the reliability and performance of the network. Passive star hubs don’t cause delays, so if you are having intermittent problems with a particular line it might be worth trying a passive star in place of an active one. This is sometimes the case when a line of terminals has been working perfectly well on the old Twinax but problems start to arise when they are moved to the new structured cabling.

When running 5250 devices over a UTP structured cabling system problems can sometimes be due to mismatched baluns. Make sure all of the baluns
on each line are from the same manufacturer, and that they are wired for the same pin-out configuration, some star hubs use pins 1 & 2 as the active pins and some use 4 & 5.

RS232
With RS232, transmit and receive are connected via pins 2 and 3 of the D type plugs with pin 7 (on 25 way) being the ground point. The other pins are used to control the flow of data, usually referred to as ‘hard wired handshaking’, the pin-out chart below shows what each pin is used for.

A device can be either a Data Terminal Equipment (DTE) or a Data Communications Equipment (DCE), this determines whether pin 2 is transmit and pin 3 receive or vice versa. Usually you will find that terminals are DTE and printers are DCE and the ports on the server board will be configured to communicate with one or the other. This makes a big difference if you start moving things around and inadvertently connect terminals or printers into the wrong ports.

If you are confused about the different terms used in data communications this article written by Mark Barratt should help to clear things up.

Bandwidth is the difference between the highest and lowest frequencies which will propagate through an equipment or system. In many cases, the lower limit is DC, zero hertz, and so the bandwidth is the same as the upper frequency limit. The public telephone system constrains all signals to the range 300 Hz – 3 kHz. Its bandwidth is therefore 2.7kHz.

In the most obvious method of modulation (representing data electrically), two different voltages are used to represent a ‘1′ and a ‘0′. The receiver expects a data bit at a certain time, and samples the input voltage to determine the value of the bit. This is called “amplitude shift keying” (ASK). The maximum frequency of the signal will depend upon the slew rate (the time taken to change from 0 to 1, or vice versa). The maximum slew rate is the upper frequency limit, and the slew rate, in turn, limits the maximum data rate.

Plainly, the bandwidth of such a system directly limits the data rate, but in theory it need not. Consider a protocol which uses “frequency shift keying” instead. Here, two different frequencies (both of them within the legal bandwidth) are used to represent 1 and 0. The maximum data rate is now the maximum speed at which you can shift between the two frequencies. This is still limited by the bandwidth, but not so directly – the resulting maximum data rate is higher. And what happens if you use more than two frequencies? You can then transmit more than one bit of information per signal transition, upping the data rate again without increasing the maximum frequency of the signal.

It is techniques such as these which have allowed the development of 56k modems. Using a combination of multiple-level amplitude, frequency and phase modulation, they manage to extract up to 56,000 bits per second of performance from the aforementioned 2.7 kHz bandwidth. To achieve this using plain 2-level ASK would require a bandwidth of hundreds of kilohertz.

“Baud rate”, strictly, is a measure of “signal elements” per second, and is not a useful measure where the above signalling techniques are being used. Such systems are generally rated in “bits per second” bps. It is worth noting that manufacturers will claim the highest figure they can for this parameter, so that the figure will include bits which are part of the signalling protocol rather than the user’s data, and may even incorporate an assumption about the compressibility of the data. It is rarely (if ever) valid to divide bps by 8 to arrive at bytes of data transmitted/expected per second.

Balanced line operation is a transmission method which helps to eliminate the effects of noise on the cable. In the first diagram a coaxial cable is transmitting a 4V signal, this is unbalanced as all of the 4V signal is carried by the centre core of the coax with respect to the grounded screen. If 1V of noise is introduced, it adds to the signal being transmitted making 5V, this could interfere with our data.

With a balanced line transmission our 4V signal is split into +2V and -2V on one twisted pair, so we still have 4V between the two. Now when we introduce the 1V of noise, the +2V becomes +3V, and the -2V becomes -1V, but the potential difference between the two is still 4V. The devices we put on the ends of the cable to make the line balanced are called baluns, this name is derived from the function of the devices of converting between balanced and unbalanced transmission modes.

These days, more and more equipment is being designed to operate on balanced lines without the need for baluns, but there are still a lot of older systems out there that still use these converters.

So what causes the signal to attenuate?, and where does the crosstalk come from?
Below are of some of the terms used in high performance cable testing, and a description of what they mean.

Length
The length of a cable is one of the more obvious causes of attenuation because the longer it is, the more resistance it has, and therefore less of the signal will get through. To measure the length, a cable tester uses Time Domain Reflectometry (TDR). A pulse is sent down the cable and when it reaches the far end it reflects back, by measuring the time it takes to travel down the cable and back again, the tester can determine how long the cable is. To do this, the tester also needs to know how fast the pulsed signal is travelling, this is called the Nominal Velocity of Propagation (NVP) and is expressed as a percentage of the speed of light. The NVP is usually somewhere between 60% and 90% of the speed of light, with most Cat 5E cables being around 70%. Due to the twists in the cable, the measured length will be greater than the physical length, so if a run looks like it might be over 80m it would be wise to check it before it is tied up and terminated.

Wire Map
This test is to ensure that the two ends have been terminated pin for pin, i.e. that pin 1 at the patch panel goes to pin 1 at the outlet, pin 2 goes to pin 2 etc. etc. The wire map also checks for continuity, shorts, crossed pairs, reversed pairs and split pairs. A Split pair is probably the only thing that requires an explanation here, as they are undetectable with a simple continuity tester, this is because pin for pin they seem to be correct. As explained on the Cabling Basics page, balanced line operation requires that the signal is transmitted over a pair of wires that are twisted together, with a ’split pair’ the signal would be split between two different pairs.

Return Loss
When a cable is manufactured there are slight imperfections in the copper. These imperfections all contribute to the Structural Return Loss (SRL) measurement because each one causes an impedance mismatch which adds to the cables attenuation.DC loop resistance
This is simply the resistance between the two conductors of a twisted pair which is looped back at the far end. The primary purpose of this test is to make sure that there are no high resistance connections in the link.

Attenuation
This is the decrease in signal strength (expressed as negative dB) from one end of a cable to the other. The main causes of attenuation are impedance,
temperature, skin effect and dielectric loss. Impedance is the combination of resistance, inductance and capacitance in a cable, it is measured in Ohms and opposes the flow of current. Skin effect is phenomena which happens at high frequencies where the signal tries to escape from the confines of the copper and into the air. The signal travels along the outer ’skin’ of the copper which effectively reduces the cross sectional area of the cable and therefore increases its resistance.

NEXT
This stands for Near End cross Talk, and it occurs because alternating current flow produces an electromagnetic field around the cable, this field then induces a current flow in adjacent cables. The strength of this field increases with the frequency of the signal, and because the speed of data transmissions is ever increasing, NEXT is a big problem.The name ‘Cross Talk’ comes from the telecommunications industry, you may have heard a faint conversation in the background while on the phone yourself, this is caused by the electromagnetic effect between adjacent telephone wires. In the transmission of data, cross talk is at its highest level in the RJ45 connection as it enters the cable, or at the ‘Near End’. The term ‘Near End’ is slightly confusing because data can travel in both directions, and the NEXT test is carried out in both directions automatically by the tester, so the NEXT result is relative to the end of the cable that it was carried out on.
The twists in a cable help to cancel out the effects of NEXT and the more twists there are, the better the cancellation, however, the twists also increase attenuation, so there is a trade off between NEXT cancellation and attenuation. The twist rates in data cables are optimised for the best overall performance, the twist rates are also varied for each pair within the cable to help combat crosstalk.

PSNEXT
This stands for Power Sum Near End Cross Talk and is actually just a calculation. When a tester carries out the NEXT test it measures the cross talk on each pair as affected by each of the other three pairs individually, PSNEXT is simply the addition of the three NEXT results for each pair. So this is the combined effect that a pair would be subject to when used in a network that supports a four pair transmissions method, e.g.. Gigabit Ethernet.

FEXT, ELFEXT and PSELFEXT
Basically, Far End Cross Talk (FEXT) is like NEXT but it is measured at the far end (well that seems logical!). However, on its own FEXT doesn’t mean much because the length of the cable determines how much the signal is attenuated before it can affect the pairs at the far end. To compensate for this, and to provide a more meaningful result, the attenuation is subtracted from the FEXT test and the result is then called Equal Level Far End Cross Talk (ELFEXT).
And of course, no test parameter these days would be complete without adding the results together for each pair and calling it a Power Sum measurement, so now we have Power Sum Equal Level Far End Cross Talk or PSELFEXT for short.

Delay
This is the propagation delay or the time it takes for the signal to travel from one end of the cable to the other, it is not very important on it own because it value is directly proportional to the length of the cable. What is important is the relationship between the delays on each of the four pairs. This brings us nicely on to …………………….

Delay Skew
Now this is important, Delay Skew is the difference between the fastest and slowest pairs. Some networks use a four pair transmission method, this means that the signal is split into four, sent down the four pairs in the cable and re-combined at the far end. It is essential that the signals reach the far end at near enough the same time, otherwise the signal will not be re-combined correctly.

The first thing to understand about testing data cables is the ACR, this stands for Attenuation to Crosstalk Ratio. The pink area in the graph is the attenuation, this can be caused by several things as will be explained below, and the blue area is the crosstalk. Attenuation is the reduction in signal strength over the length of the cable and frequency range, the crosstalk is the external noise that is introduced into the cable. So, if the two areas meet, the data signal will be lost because the crosstalk noise will be at the same level as the attenuated signal.

ACR is the most important result when testing a link because it represents the overall performance of the cable.

Q. I just found your web site and i thought that it was very informative, but i could not find any thing on the difference between stranded and solid cabling, could you tell me what the difference between them is? (I think stranded is used as patch leads and solid is used as a connector between the patch panel and the wall outlet).
A. You are correct, stranded cable is used for patch leads because it is more flexible than solid copper. The solid cable is used in the fixed part of the installation, ie. the cable between the patch cabinet and the wall outlets. Solid cable has better performance characteristics than stranded and it is cheaper to make.

Q. Thank you for your help, it was very useful. One other thing you could help me with, is the way that the solid and stranded cable are wired up different? because i know that you can get RJ45 plugs for solid and stranded cable (i know how to wire up stranded cable to a RJ45 plug), if so then how is the solid cable wired up to the RJ45 plug.
A. The colour codes are the same for solid and stranded cables, the difference is in the IDC (insulation displacement connector) in the RJ45 plug. Because the cores are different the contacts have to be slightly different to ensure a good contact is made.

Q. Our offices are moving into an existing building which was wired by the previous owners, or a contractor for them, and their wiring is one I have never seen before. Each wall outlet has two data jack which are “sharing” a single UTP cable, 2 pairs to the left jack and the other two pairs to the right jack. What problems are we likely to encounter with this setup? Our normal wiring method is one jack one cable…..

A. It sounds like costs were an issue when this building was cabled, but if it is configured for Ethernet (using pins 1, 2, 3 & 6) it should be OK for 10BaseT. I wouldn’t like to speculate on whether it will run at higher speeds because that is dependent on the quality of the installation and the amount of network traffic.

If the installation was originally wired for Token Ring then it will use pins 3, 4, 5 & 6 which will not work with Ethernet.
Crosstalk could be an issue if the pairs have been split between outlets and/or between pins on the jack, (pins 1 & 2 should be a pair and 3 & 6 should be a pair).

I hope this helps, and although it is possible that you will have no problems, I would strongly advise a rewire. This is because it will be easier at this time to carry out the work and, as you move to higher speeds in the future this wiring configuration will undoubtedly start to cause problems.

Cat 5 and Cat 5E
The basic Cat 5 system used to be the only real choice, but developments in Ethernet technology led to the introduction of ‘Enhanced Category 5′ or Cat 5E. Both systems are capable of transmission rates up to 100MHz, but the test parameters for Cat 5 assumed that data signals would only use two of the four pairs (one pair for transmitting and one pair for receiving) and crosstalk measurements were only taken between each pair combination. With Gigabit Ethernet however, all four pairs can be used to transmit simultaneously, and so the cross talk on each pair has to be measured for the combined effects of the other three pairs.

Cat 6
At last! the standard for Cat 6 has been approved for publication by the EIA (TIA/EIA-568-B.2-1). Category 6 is capable of transmission frequencies up to 250Mhz and has a positive power sum attenuation to crosstalk ratio upto 200MHz using improved cables and RJ45 connectors. The problem that manufacturers have, is that to meet the Cat 6 specification, requires the use of cables and connectors which are designed to work together as a ‘tuned’ system. This means that if you install a Cat 6 system the manufacturer will only guarantee performance if all of the components including the patch leads are from their Cat 6 product range. In fact, by mixing Cat 6 components from different manufacturers you could end up with a system with worse performance characteristics than a conventional Cat 5e system. That said, it is worth noting that Cat 6 systems are backwards compatible with Cat 5/5e cabling and when mixed with these lower bandwidth systems the performance criteria of the lower specification will still be met.

Testing Cat 6 cables can be a frustrating process, apart from taking longer because the tester has to scan frequency steps up to 200MHz instead of 100MHz, the fine line between pass and fail is accentuated it seems by the slightest kink and twist. The most significant factor when testing a Cat 6 system can be return loss failures due to the test leads themselves. All connectors have a life cycle and with the average RJ45 connector this is around one or two thousand insertions, so test leads should be replaced after every 1000 tests or so. OK, not a problem but at around $200 per set this cost will have to be considered when pricing jobs.

Fluke seem to have a solution to this problem with their DSP-LIA101S Permanent Link Adapters. The connector at the end of the leads are interchangeable and replaceable with connectors from different manufacturers to ensure compatibility with the system under test. Although a good idea, the adapters are over $500 and a new pair of “Personality Modules” cost over $100. Surely the test plugs should now be considered as ‘consumables’ and the price lowered to reflect this.

Cat 7
This is proposed to be a 600MHz system using a shielded cable with individually screened pairs and a new type of connector. The cable and connectors are slightly bigger than Cat 5e and installation time can be increased because of the complexity of the termination. There are two main draw backs with installing this type of cabling, the first is the additional cost involved, and the second is that almost all networking hardware uses RJ45 jacks. To connect to the cabling system, you have to use Cat 7 to Cat 5e patch leads, and because any system is only as good as its weakest link, your speed is back down to 100MHz. Ratification of the Cat 7 standard could be two years away by which time fibre might be a cheaper alternative.

Shielded or Unshielded
This is a subject that has been debated and argued over for a long time, and as yet, there are still no definite answers. Most countries in Europe, and in particular Germany, argue that apart from protecting data signals against high frequency noise from outside sources, shielded cable also protects the humans against the possibility of having their brains fried due to the effects of high frequency emissions from the cable itself. Other countries, such as the UK, US and Canada, aren’t particularly bothered by this because nothing has been proved, and after all, millions of people wander around with mobile phones pressed against the side of their heads with no apparent side effects, er… yet. My advice would be to install unshielded cable unless the customer insists on a shielded system.

Shielded cables and components are more expensive and are more time consuming to terminate, you should also bear in mind that a shielded cable that isn’t properly grounded has worse performance characteristics than an unshielded cable. If a shielded cable isn’t grounded at all, the screen can act like an antenna and induce all manner of noise on to the data signal.

Low Smoke Zero Halogen
In public buildings, such as airports, shops and hospitals, then the cable should be Low Smoke Zero Halogen (LS0H or LSZH).