Finding damaged cables

From DIYWiki
Revision as of 16:47, 30 April 2022 by John Rumm (talk | contribs) (→‎The circuit: Added pics)
Jump to navigation Jump to search

Or, measuring stuff for nerds :-)

You know when you have a long non working cable that is built into the fabric of a building, you can see both ends, and they don't appear to be connected but you know they should be?

It might be an existing cable that used to work, but a rodent, or a careless picture hanger with hammer and nail, has got too close to it. Perhaps it is one you just spent hours pulling through conduit, under floors, into cable ducts, and plaster chases, but now you know something you did damaged it. The immediate question is then, how do I find the damage and fix it, without ripping it all out and starting again?

Here are a few ways

Cable resistance

If you have a sensitive low ohms meter, then sometimes you can use some careful measurements of cable resistance to find the location of a fault. This tends to work best when the fault is a short circuit, typically caused by something metal penetrating a cable like a screw or a nail.

Say you have and length of twin and earth cable, that is tripping the MCB as soon as you power it. With both ends disconnected, a quick resistance check between the cores might show that there is a Live to Earth fault on the cable. Now using a resistance table like this, you can quickly get a feeling for how far away from your end of the cable the short might be. If you then deliberately add a short at the far end between a couple of the other cores, then you can also get a feel for the resistance of the total length of cable. Repeating the test at the far end should give more confidence of the location.

Cable capacitance

The capacitance mode on this multimeter can measure down to 1nF. Connected to two wires at the end of over 150m of CAT5e network cable as shown here, the resolution is not really adequate to make anything other than a very general estimate.

A capacitor is created by having two conductors separated by a small insulated gap. It can store electrical charge in the electrical field created between the two conductors. This means that even a normal cable will show a (very small)

Connected to the same CAT5e cable as the multimeter, this LCR meter can show more resolution. One common estimate of the capacitance between wires in CAT5 cable is approximately 52 pF/m, That would estimate the cable length at 7.146 x 10-9 / 52 x 10-12 = 137m

capacitance between the cores. So if you can measure the capacitance between the cores of a cable, and know (or can measure) the capacitance between the cores of a known length of cable. You can estimate either the length of the cable, or the distance to the broken conductor.

The snag is that the capacitance of a length of cable is typically tiny (we are talking pico Farads per meter (where 1 pF = 0.000000000001 Farad!)). So to use this directly you will need a sensitive meter designed for this kind of measurement. Some multimeters can measure capacitance, but many can't read much below 1 nano Farad (nF). However this may give a rough idea on very long cable runs.

A dedicated LCR meter will give a much more accurate result.

Many network cable testers, are also able to measure cable lengths using this method. So for example you have a CAT 5E network cable with a broken pair, you may be able to use your cable tester to measure the capacitance on a couple of different sets of pairs. The one with the break will typically show a lower capacitance. The ratio of readings giving an indication of the location of the break

A network tester making a capacitive measurement of the length of patch lead. Note that the meter allows the user to set the capacitance per unit length value used. This allows various different types of cables to be tested - not just network cables. All you need is a value to use from a manufacturers data sheet, or a measurement made from a known length of the same cable, and a way to connect your cable to the RJ45 socket on the meter.

Time Domain Reflectometry

Another way to take advantage of the characteristics of a cable to make measurements, is using a Time Domain Reflectometer (TDR). If you inject a signal into a cable (in particular a very short or abrupt signal), there will be a brief period where the cable's capacitance will allow it to "charge up" in response to the injected signal. This charge then propagates along the cable. Should it reach the end (or an interruption), and find there is no attached "load" for it to flow into, that stored energy needs somewhere to go. So it will bounce back down the cable rather like a wave on the surface of water bouncing back from the edge of a pool.

At the injection end of the cable, there will be a brief period of time where you can see the effective load of the cable on the signal change while it charges initially, and then when when any reflections return.

A dedicated TDR is a sophisticated (and expensive) bit of test equipment, that can make measurements on cables by injecting a sharp electrical pulse or "edge" into a cable, and looking for the characteristic reflection that will tend to "bounce back" from with the end of the cable (or from the place that a step change occurs in the cable's characteristics occur). By measuring the time between the pulse and its echo or reflection, it can estimate how far away the reflection was generated based on the propagation speed of the cable (which will typically be a significant proportion of the speed of light)

TDRs can be bought on their own, but the facility will often be found in the more expensive network cable testers.

Home made TDR

One can cobble together a basic TDR capability using off the shelf test gear like an oscilloscope. So long as you have a way of injecting a pulse into your cable under test, the scope should be able to measure the time between injection and echo.

Here are some experiments made using a simple home made pulse generator and an oscilloscope.

The circuit
Circuit for a battery powered oscillator designed to inject a "sharp edged" square wave into a cable under test. The LED and R1 just provide a visible indication that the circuit is on. C1 provides a little stabilisation for the power supply to the 74HC14E IC. R2 and C2 provide a delay in the feedback to the input of the first inverter. The Other 5 inverters are wired in parallel to the output of the first to boost the output drive. The trailing 220Ω resistors give some protection to the outputs from short circuit, and also (when added together) provide a nominal output impedance of approaching 50Ω since

The pulse generator circuit used here is very simple and you can find examples similar to this dotted about the internet - here is a good explanation of the principle of operation from what may be the original author of this particular design.

This circuit uses a common "hex inverter" IC that contains 6 simple "inverter" logic gates. (an inverter simply changes a "low" signal into an "high" signal, or a "high" signal into a "low").

If you connect the output of an inverter back to its own input, then it will switch back and fourth between high and low trying desperately to keep up with its own output! This turns it into an oscillator. By adding a resister and capacitor to this feedback loop, we can slow it down to get a moderate speed output. (the values used here give a square (ish) wave output at around 3.7kHz - the actual value does not really matter for this application).

To get a good pulse or "edge" for TDR style measurements, it ideally needs a nice fast "rise time". I.e. the change in voltage from off to on needs to happen very quickly. Common "74 series" logic chips like these come in a number of different versions. Choosing one with a fast rise time, and a "Schmitt trigger" will give good results in this application. This has the advantage of not only a fast rise time, but it also gives very clean high or low outputs, without any indeterminate value should the input not be clearly defined.

Building a unit to test
Circuit assembled on a small prototype board, and fixed to the case with hot snot. The output presented on a normal 50Ω BNC socket.
Pulse generator connected to an oscilloscope, showing the leading edge of section of the square wave output from the circuit.

To experiment a little I built up the circuit above in a small plastic box with a couple of AAA batteries for power. Connecting this up shows a neat square wave output:

Showing a square wave with a frequency around 3.7kHz

If we zoom right into one of those leading edges, we can see what it looks like:

A sharp rising edge with a rise time of under 3 nSec. There is a little bit of "ringing" shown at the top of the edge where it overshoots a little.
In Use
The test setup uses a BNC Tee connector to allow the pulse generator to feed a signal to the cable and to the oscilloscope. The cable connected to the Tee using a BNC to screw terminal adaptor. The scope will see the test signal, and the response from the cable superimposed onto it.

For a quick test I grabbed the nearest bit of spare cable, which was a length of 3 core flex.

If we hook up a length of cable (the far end of the cable is not connected to anything), then we can see the characteristic signal reflection arriving back from the end of the cable. So instead of the simple rising edge shown with no cable connected, you see the distorted signal shown above.

We can now measure the time from the start of the signal to the start of the reflection:

This shows the reflection is delayed by 25.80 nSec (shown as BX-AX in the info panel at the top left of the screen). This time will be the total return time for the signal to reach the end of the cable and come back. So the "one way" trip will have taken 12.9 nSec. The speed of light is around 3 x 108 m/Sec (usually shown simply as "c") in a vacuum, however the speed of a signal in a cable will be reduced somewhat. The amount it is reduced is known as the velocity factor, and that will depend on the type of cable (it could be anything from 0.4 to 0.9 of the speed of light. So if we multiply c by 12.9-9 Sec and by a velocity factor of 0.6, we get a distance of 2.32m. My trusty tape measure indicates 2.37m - so not bad going

We can repeat the test with did on the box of CAT5e in the capacitance measurement above, and we get:

Hear the round trip time is a much more substantial 1.48 uSec. Running the same calculation as above gives an estimated length of 133m
Other tests