Testing Isolation Resistances and Dielectric Withstand on Wires

I used to do this for a living at an electrical test lab. Why do you need to perform insulation resistance and dielectric withstand tests on individual wires. What is the actual application and how are you implementing it at the moment.

Jack thanks for your commect,

We need to do this to make sure that all wires in a wire bundle are good isolated. This is for the aircraft industry. My customers tell me how they what to have tested there harnesses, and we would like to improve this. But we must be able to convince them that the process that we would like to use is as good or even better than what they describe to us. These tests are set years and years ago. I think that it must be able to see that the test methods also improve and that old procedures can be reviewed. There are companies (test equipment companies) how tell me that making groups is as save as line by line testing. The only thing is that I do not have references that underscribe or prove this in a technical way.

So jack if you have experience on this please let me know or tell me where to find it.

Fesco

Bundle testing is a lot more complicated, and given the large number of wires in this case, the best option would be to automate the testing process somehow. A couple of ways include using a PLC or microcontroller to switch between wires and apply the test voltage. Another way is to immerse the bundle in a conductive fluid and apply the voltage to individual (or all of the) wires to test for any wire insulation faliure (insulation cuts, etc).

Not being in the aircraft industry myself, I would suggest you first get yourself a copy of the latest standard relating to aircraft wirining and cable loom testing that is applicable to your country and/or customer. There is bound to be one, and this should give you a better indication on what test procedures are to be imployed when testing aircraft looms. Actual test procedures and setups should be included (as they are in other industry standards).

Jack,

You are right that bundle testing is complicated. Therefore it is time consuming. There are many actual written procedures like in all industry standards. the only thing is that they don't Mather how much time it will take to test a complete bundle. In some cases it is allowed to make groups to speedup the test. These are normally customer made procedures. Still the procedures that tell me that it is necessary to use the "wire against all other wires" procedure is not clear for me. That's why I put this question on the net. What is technical the difference (and risc) between testing in groups or 1 wire against all other wires.

Test equipment vendors tell me that it is oke, but I still want prove, and until now I still do not have the prove that there is no risc that I do not miss one wire. If you look fro mass Hipot testing on the internet you get some information, but there is no technical part that proves that is the same as the traditional way of testing.

Thanks for your responses.

Fesco

Group testing of bundles should be ok, after all you are assuming the wires have already been tested/approved by the manufacturer and any actual insulation failure you observe will have been caused by mechanical damage during loom assembly (or possibly transport/handling) - nicks, cuts, etc. Testing using the immersion test may be your best option to speed things up.

Are these 100 wires in the same cable?

Normally a cable is organised, and a wire will see only the adjacant wires.

Make logical combinations: in twisted shielded systems you have the check the wires to the shield and to each other, and not to other pairs in the same cable.

Testing is time consuming and 90% boring, and most of all expensive. But it will take your work to the next level.

Hi Gwen,

These 100 wires are in the same bundle. What you stated is that if a twisted wire has no problem to its shield the internal wires do not need to be checked against other wires in the bundle. But unfortunately this is necessary because the internal wires in a twisted cable are finished in a connector. This is the reason that you need to check every wire against all other wires to make sure there is no isolation damage somewhere in the bundle.

You have a point that if we are able to make logic combinations this will decrease the number of tests.

Fesco

Never forget that testing 2 wires requires 1 test, but testing 3 wires requires 3 tests, 4 requires 7.

when the customer asks for a test between each possible combination you will see:

n : wire quantity

t : test quantity

t = 2^(n-1) -1

Grouping them can bring the tests back to a reasonable level. (99 on the ground and 1 on the HV side, and go through the bundle) When something fails you have to look for the exact combination that fails. (or throw the unit away, but fault location is better to avoid it in the future, or to discover a wrong or wearing out tool)

Isn't there a way to automate the work, test a bit more (resistance of the wire), log the results and ask some more money? (resistance check can show problems with connection points, that don't show up with HV tests.

When it is equipped with connectors it is fairly easy to work with a PLC that opens and closes contactor to combine the cables.

Logged automated systems always perform better than manual tests (a PLC never forgets a combination and will never try to work faster, skipping some items that are usually OK)

Gwen

Gwen,

In my cause I test wire 1 against ground (test 1). Then wire 2 against wire 1 and ground (test 2). Then wire 3 against wire 1 ,2 and ground(test 3) and so on. So I don't have the exponential number of tests. The way I test I'm sure I have tested each wire against all other wires.

By grouping them I can lower the amount of test. The way of thinking is to make groups on both sides the + and - side. If you change these groups you should be able to test all wires. Please have a look at "Mass Hipot testing on the internet. Still I do not have the prove that this way of testing makes sure that I do not Miss one combination.

I do not use a HV test to check the connection. This is done a different way. And I'm using automated test systems. The companies that sell these test sytems are not able (yet) to prove to me that it is save to use. So that is why I put this question on the net.

Thanks for your comment

Fesco

The companies that sell these test systems are not able (yet) to prove to me that it is save to use.

That is a typical reaction for your industry: they will not take the responsibility, they only sell you a system.

It's also why we don't sell our cables for aircraft use. I'm application designer at a major heat tracing wire manufacturer.

The method explained by Jack is also what I would do: put the mass and all wires except 1 at ground level and apply the test voltage at the one tested. With this you are sure that this wire has no insulation defects, no to the mass and not to another wire. The next step is to go through all other wires.

Gwen

if I were you I would found a metal pipe, put one wire stick its inner wall and link with its terminal and spread other, for example 10 or 20 wires or more over its outside wall, short one end of them then use a high rsisitance meter to test its insulation resistance. the voltage generated from teh meter depend on the endure voltage of the wire.
so that you can save some time and repeat labour.
hope this foolish method can do a trick.

you can also use a metal plate instead, one side lay 10, 20 or more wire with one end joint toghter and another side place several wire for test and link with plate. use the meter(Mohm moter) according to your demand insulation, for example, choose 600v, 800v or 1kv or more etc. you can tst many wire as you like. to see if they havea good insulation. if not you can use a coeficienc 0.618 saperate the wire into two parts and test, if the remain is good then you can test next part that you take away at first. so that you will have a good effective. of cause you can use 1/2 as well, but 0.618 is a best bet.

The wires will certainly be spark tested out of the factory. this leaves out the possibility that there is something wrong with the wire before it is used in the bundle. (At least we do this and we refuse the use of our cables in airplanes)

The problem is that the bundle, as a finished unit, must be checked on workmanship faults. (cutting to deep with the knive, ....)

Gwen

gwen you are right, bundle test is diffecult to identify if any two wire is btter insulation, becuase some two will be apart from long distance than you expect. say, three or more wire among these two wires. of cuase they have more insulatin you would get. although we can get test probebility but it cannt be used for plane and strict place. spark test will hurt outlayer of wire I thnk, its not used for this test. I still think the abuve method is a simple and better way for fast and simple testing. you can simply press the wire under a matel plate and linked. I choode matel plate because it can get a equal potential to each wire under it, so that you can test many wire a time. if you argue, how about their relation fo these wire under the plate, i can aswer, they can also bear simulaly voltage as they bear the volt to above them one. it seems not too clear what I describe. I try to draw a pic to show you. if yu suspect.

I came across this article recently. A portable Pulsed Arrested Spark Discharge (PASD) tester for testing aircraft wiring looms.

http://www.gizmag.com.au/go/5773/

May be of interest.