Buried SWA cable fault finding

[danielz000]

Hi all,

My first post here, thought I'd see if it was possible to get some advice or tips...

Basically, I had a large amount of block paving and large concrete base laid at the far end (25m) of my garden a few months ago. Under this I had 3 core 16mm SWA laid.

(I most definitely can't vouch for the quality of the work done by any of these guys)

More recently me and my brother built a large garage on the concrete base and had an electrician add a new CU in the garage and connect the other end of the SWA to an MCB in the CU in the house.

Today, after about 2 weeks of not having any problems, the RCD in the house tripped out and wouldn't switch back on unless I turned off the MCB to the garage. It's worth noting, it rained fairly heavily last night.

After trying a few different things, I've eventually disconnected the SWA from the CUs at both ends.

I've tested for continuity across all cable/ armour combinations and noticed there's continuity (beeping) when I touch the live and armour.

Is my assumption that someone has damaged the cable, breaking the live wire insulation and water has entered this and created a circuit between the live and the armour a reasonable one? Are there any other ways this continuity could occur given both ends are now fully disconnected?

Assuming this assumption is correct, I'm curious if there's a simple way for me to locate this fault without digging up 30m of block paving?

For example, I was wondering if, by using a multimeter and testing the resistance across the live and armour at both ends I might be able to work out how far down the wire the break is?

So maybe if I got a resistance 10 times higher at one end than the other I can assume the damage to the 25m long cable is 10% (2.5m) from the end of the cable with lower reading?

I guess I don't need to be super accurate and that there are industrial tools that could accurately locate the fault, all I really want to know is roughly were to start digging in the hope of keeping the damage to the paving to a minimum.

Thanks in advance for any help or suggestions.

Dan

 

[timhoward]

That leaves half of a pair of simultaneous equations and I gave up and opened the 1664 at that point. I’ll return to this one day…

To answer my own question in case anyone searches the forum for varley loop test in future…
The 2nd equation is derived from twice the resistance of a length of one conductor being equal R+X+y. Substituting that in leaves a ratio of resistance X to resistance of the total loop. I think.
I might try this and @pc1966 method out when I have time just for fun.

 

[Jim_e_Jib]

Deleted - misread post!

 

[marconi]

danielz000 - I am no more than 20 minutes drive from you. Do you want me to call in? If the method my colleagues have explained does not work on the damaged 25m length I will improvise a time domain reflectometer and use it during a second visit.

If he was my brother he would have been given an Essex smack for not waiting for some testing to be done ;-)

We will pass circa 50 Amps through the cable in order to produce a volt drop of the order 50 x rho = 50 x 0.00183 Ohm/A/m = 50 x 0.002 = 0.1 V per m or 25 x 0.1 = 2.5V over the length of the 25m cable. May even use an even larger current briefly.

Shall I buy the necessary items or do you want me to guide you through what you need and how to do it yourself? For safety's sake is the house end of the cable removed from the CU so that you would not be working near anything live?

 

[littlespark]

And was the fault actually under the blocks or concrete? Did you do a load of digging for nothing?

 

[DPG]

danielz000 - I am no more than 20 minutes drive from you. Do you want me to call in? If the method my colleagues have explained does not work on the damaged 25m length I will improvise a time domain reflectometer and use it during a second visit.

If he was my brother he would have been given an Essex smack for not waiting for some testing to be done ;-)

We will pass circa 50 Amps through the cable in order to produce a volt drop of the order 50 x rho = 50 x 0.00183 Ohm/A/m = 50 x 0.002 = 0.1 V per m or 25 x 0.1 = 2.5V over the length of the 25m cable. May even use an even larger current briefly.

Shall I buy the necessary items or do you want me to guide you through what you need and how to do it yourself? For safety's sake is the house end of the cable removed from the CU so that you would not be working near anything live?

The usual knowledge and helpfulness from Marconi, mixed in with a slight hint of violence ?

Be interested to see how this works out.

 

[littlespark]

“Hi, brother…. This is the guy I met on the internet that says you’re a prat”

 

[pc1966]

To answer my own question in case anyone searches the forum for varley loop test in future…
The 2nd equation is derived from twice the resistance of a length of one conductor being equal R+X+y. Substituting that in leaves a ratio of resistance X to resistance of the total loop. I think.
I might try this and @pc1966 method out when I have time just for fun.

My concerns with the Varley Loop / Murry loop sort of tests here is the need for high current / low resistance adjustments. With ~25m of 16mm copper the total resistance of each conductor is of the order of 0.03ohms and you will struggle to find rheostats of that sort of value and ~10A rating just sitting around.

My approach is really a "Wheatstone Bridge" arrangement but instead of using the other arms to null (zero) a galvanometer, today we just use a decent DMM that will be > 1M impedance so can be ignored and if 1mV resolution for the meter we can get a meaningful reading / distance resolution with only a 100mV or so total volt drop. The balancing arm ratio then becomes maths on the two measured voltages.

So then test currents of the order of 5A are feasible and sustainable without worrying too much about how to get that much current (mid-size laboratory PSU, car battery with some current-limiting load like a 60W bulb, etc).

 

[pc1966]

3. LAP AC/DC Digital Multimeter 600V - https://www.NoLinkingToThis/p/lap-ac-dc-digital-multimeter-600v/161fg that's the multimeter I got but when I tried to measure resistance across the live and armour it just read "ERR". Nothing in the manual about how or why that can happen...

The Screwfix site lacks detailed specifications, such as the voltage ranges/resolutions supported. At least it is CAT-II rated, but that is not quite enough if measuring at the CU (instead of at the appliance after 13A fuse, etc) See:

Measurement category - Wikipedia

en.wikipedia.org

The usual reason for getting an error when attempting to measure resistance is it is seeing some residual voltage there. It could be the fault is intermittent so only manifesting at higher test voltages so it is picking up capacitively coupled voltages, or it could be there is a small AC voltage of the order of 0.1V or so induced from some other energised circuit or current flowing back via earthed metalwork in the garage, etc.

As general advice, always check for voltage before you measure for resistance! Multimeters are not considered safe for "proving dead" as there are so many ways you can get it wrong, possibly resulting in a fatal shock if you failed to detect high voltages, or an inadequately rated meter exploding if subject to a very high energy fault.

That is what the likes of GS38 health & safety advice call for the use of a dedicated voltage tester, along with some means of proving it is working before and after you checked for the presence of high voltages. This is the sort of thing you would be looking at for a low budget:

https://www.NoLinkingToThis/p/lap-ms8922b-ac-dc-2-pole-voltage-tester-with-rcd-400v/669hy

The proving units are significantly more expensive (in the £70++ range) but for occasional use you can check it works on a circuit you did not switch off before/after you test one you plan on working on.

4299 Proving Unit

Intelligent voltage ramp testing for voltage indicators and multimeters

www.cef.co.uk

 

[HappyHippyDad]

@danielz000 if marconi is offering to come and help, I would jump at the chance. His Knowledge is vast. You may not understand a 100% of what he is explaining to you though ?.

Ps.. I love @Lucien Nunes explanation of why a standard resistance test from each end may well not work! (i.e pin hole and filling channel tunnel). Really helps you see the problem, especially when you cannot understand all of the physics involved.

 

[pc1966]

If he was my brother he would have been given an Essex smack for not waiting for some testing to be done ;-)

THIS !!!

Though quite why you would give him a twin-foresail fishing boat is a bit of a mystery:

Betty CK145 History of the Essex-smack

Essex oyster smack Betty CK145 homepage

www.betty-ck145.com

 

[nicebutdim]

@danielz000 if marconi is offering to come and help, I would jump at the chance. His Knowledge is vast. You may not understand a 100% of what he is explaining to you though ?.

Ps.. I love @Lucien Nunes explanation of why a standard resistance test from each end may well not work! (i.e pin hole and filling channel tunnel). Really helps you see the problem, especially when you cannot understand all of the physics involved.

I might understand somewhere in the region of 5% on a good day (plenty of sleep and a decent meal the night before).

Hopefully the OP jumps at this chance and then reports back here on the experience.

 

[johntheo]

I might understand somewhere in the region of 5% on a good day (plenty of sleep and a decent meal the night before).

Hopefully the OP jumps at this chance and then reports back here on the experience.

If I'm interpreting it correctly the voltage drop (across R1+R2) is measured between the two ends of the faulty cable and then between one end of the faulty cable and the armour (just R1), one of the healthy cables is just being utilized as a 25 meter M.meter lead for the first reading. Not quite sure about the right calcs but assuming that the fault is say only 5 meters from the joined ends then the volt drop across RI might be V1x20/25. Would love to see actual test readings from this method.

 

[marconi]

If I'm interpreting it correctly the voltage drop (across R1+R2) is measured between the two ends of the faulty cable and then between one end of the faulty cable and the armour (just R1), one of the healthy cables is just being utilized as a 25 meter M.meter lead for the first reading. Not quite sure about the right calcs but assuming that the fault is say only 5 meters from the joined ends then the volt drop across RI might be V1x20/25. Would love to see actual test readings from this method.

I will do you a demo. All a bit academic now cos two joins underground not great future for this run of swa cable- one maybe tolerable but two …,…

 

[UNG]

All a bit academic now but was there any tests done before the cable was put into service / buried to check it's integrity

 

[freddo]

I must have a go at that Varley test. I have a fault on a 4 core swa , 2 cores low resistance to each other and to earth. Would be cool to find the location of the fault, though most of the cable is now under a building extension. I do have a 20A variable PSU and some big resistors and power variable resistors

 

[pc1966]

If I'm interpreting it correctly the voltage drop (across R1+R2) is measured between the two ends of the faulty cable and then between one end of the faulty cable and the armour (just R1), one of the healthy cables is just being utilized as a 25 meter M.meter lead for the first reading. Not quite sure about the right calcs but assuming that the fault is say only 5 meters from the joined ends then the volt drop across RI might be V1x20/25. Would love to see actual test readings from this method.

Most of the methods are doing this - establishing the resistance either side of the fault to armour. The different methods really come down to available test equipment, etc, and some minor variations in accuracy for a given assumption of test equipment, simil;arity of available 2nd conductors, etc.

One method mentioned that is not doing that is time-domain reflectometry (TDR) where a pulse is sent down the cable and where it encounters a change of characteristic some of it is reflected back. By measuring the time taken for this first reflection (you get another large one when the pulse reaches the far end) you can estimate the distance to the fault either directly (using the known velocity factor for the insulation) or by ration (from known length and far-end delay).

This is commonly used in RF systems but often indirectly (not a pulse, but maths performed on a swept-frequency reflection measurement), and in fibre optics using a laser pulse. Whereas in power systems they originally used the closing and then fault-triggered reopening of the supply breaker to generate the surge, though probably these days it is a safer low energy plust that is used!

 

[pc1966]

I must have a go at that Varley test. I have a fault on a 4 core swa , 2 cores low resistance to each other and to earth. Would be cool to find the location of the fault, though most of the cable is now under a building extension. I do have a 20A variable PSU and some big resistors and power variable resistors

Let us know how you get on with the different methods, would be good to have some hands-on experience of how easy each/any of them were.

 

[pc1966]

Overall though the underlying problem was most likely poor workmanship.

While you can bury SWA directly in the ground it should have sand or similar around it for several cm to make sure that sharp stones are not crushing the armour - there is always a limit to what it can take! Though it is just possible that very bad handling of the cable might have caused internal damage.

As others have also said, putting in duct is better for many reasons, though it also should have sand or very fine gravel surrounding it for the same physical protection reasons. The big advantages are: less likely cable damage, the ability (if sane length and route) to pull through a fault cable and bring its replacement in, and also to allow other cables if needed (say network for internet, etc).

Finally it should have been properly tested after initial burial and before final work over that area. And that means a proper high-voltage insulation check as well as the typical conductor / armour resistance chances you would do to verify any new circuit.

 

[marconi]

I must have a go at that Varley test. I have a fault on a 4 core swa , 2 cores low resistance to each other and to earth. Would be cool to find the location of the fault, though most of the cable is now under a building extension. I do have a 20A variable PSU and some big resistors and power variable resistors

I do not have a long length of swa with which to do a realistic demo using a high current. How about me sending you my test rig - the parts for which are on order and due today - and then you use it on your defective length of swa cable and see how it performs in practice? You can keep it afterwards.

What is the csa and length of your cable?

 

[freddo]

Yes I would be interested in playing with your bits! The CSA is 25mm2 length ~40M. Someone with some modern test gear was going to take a look last week but couldnt make it. I ended up retesting the 2 undamaged cores at 2500V today to prove they were still ok, got greater than 2500Mohm in all directions once I had cleaned the exposed insulation in the cabinet outside, as needed to get a reduced supply back on feeding half of the lighting. The fault is between yellow and red phases and armour today it measures around 100ohms on a continuity test. The 2 damaged cores are now extra cpcs. As most of the cable route is now built on it is likely a new route will have to be dug soon. A ground working company are looking in next week.

 

[marconi]

I will start to assemble the test rig tomorrow afternoon. I assume you have some jump leads to connect a 12 V battery to my board and a digital voltmeter able to read 12V dc. As the cable is 25mm2 I will arrange to produce some pulses of current of the order 70 to 80 Amps so that a volt drop of a few volts is produced along the red or yellow conductor. I am using twenty or so 12V 50W halogen lamps as my power resistors so you will need to wear some sunglasses when it pulses the current!

 

[freddo]

Yes I have all that, I have deep cycle 120 and 200Ah 12v leisure battery to hand.

 

[marconi]

Good. One last question - what is the settling time of your dvm? Ie, if you measure the terminal voltage of one of the 12V batteries how many seconds from making contact before the digital readout remains constant?

 

[freddo]

My old one here takes approximately 0.5s. Which is much faster than my newer meters...

 

[pc1966]

Can you video this testing out of interest?

Also it would be interesting to compare the measurement results of a couple of methods to see how closely they agree, and potentially to compare that to where the fault is finally located if the cable is dug up at some point.

 

[freddo]

Yes I should be able to do that it, would be more interesting than the usual junk i normally put on youtube.

 

[timhoward]

Semi-related interesting video.... I'm not suggesting this is the way forwards but it's VERY cool....
With external power source (up to 40v before resistors needed) this would actually do the Varley test as far as I can see.

 

[marconi]

Prototype 12V 25A load.

 

[marconi]

This test rig is a demonstrator not some finished item of test equipment so its construction is economic using many items just gathering dust in my man shed.

The interesting aspect of this project for me is not so much the method which is long established it is actually doing it practically. The hardest part of it is finding a cheap way to load up a 12V battery to cause a current of the order of 75 Amps to flow without everything bursting into flames or becoming a molten mess. In other words the power resistors. Hence my use of 12V 50W g6 halogens. The nice thing about these is that once lit their current is quickly constant during the test unlike using other resistors - and cheap.

Nevertheless, there will be 18 x 50W halogens lit producing mostly heat and some light - nearly a kW. You can see from the image I posted that I used cheap connector blocks - I could not find online a way of buying a score of g6 lampholder with leads without spending a small fortune. So I improvised. But by only having them on for a few seconds and then ensuring an interval of at least a minute before they next light I intend to ensure this does not happen.

I have some electronic components to mix together to produce a circuit which will pulse a current through the cable, alternate the dvm between the near and far end of the cable and have a limitation on the minimum interval between pulses of current - I have in mind 60 seconds and 5 seconds for freddo to take the two readings. He will press a button and then one pulse of current will occur to take the voltage measurements; then a wait of 60 seconds before another pulse can happen.

In an experiment it is normal practice to repeat measurements in order to deal with the inevitable errors and noise present. If it was me I would repeat the testing ten times to see if the results were consistent.

Our Jack Russell puppy is being sick so I must stop now.

 

[johntheo]

Is this a "continuation" of your suggestion in post #25 in that you require this pulsating system.?