Archived Forum

I am going to say that a "snug fit" really isn't necessary, although it would help to maintain alignment with the wire it's reading. What you really don't want is a CT that flops around loosely on the wire, changing it's position each time it's disturbed.

What's probably a lot more important is that you maintain as perpendicular an orientation to the wire as possible. Making a form of some sort so that the wire is as much in the center of the CT as possible, and as perpendicular as possible will mean that the CT doesn't change it's physical relationship (and therefore the strength and quality of the magnetic flux it's reading from the wire) as possible.

If you can't get the wire completely in the center of a large CT, or can't clamp it on the wire so it sticks out at a right angle, it will still work, but changing either of those properties will affect your readings. (center-of-CT is probably less important than angle, overall)

Once you have your equipment calibrated, you really don't want to change things between the wire and the CT, as this will change the coupling between the wire and the CT, negating your calibration settings. A snug fit CT will have less room to change, so would be less affected by movement, being disturbed, etc.

Not disturbing the core of the c.t. once it is fitted and the phase angle correction performed is much more important than any of the points mentioned by 09540. In tests, I have not been able to discern any effect from either the current-carrying conductor not being centred, nor from it not being perpendicular to the plane of the core, but nudging the core and affecting the mating and alignment certainly has a dramatic effect on the phase error, but a very much smaller effect on the amplitude.  I'd be interested to know what is the theoretical basis for thinking that not being centred nor perpendicular might affect the output.

I just wrapped some electrical tape around the conductors, and gently pressed one side of the CT onto the tape. The conductors I'm measuring are close to vertical, so that and gravity work well to keep them stationary and reasonably aligned. You could also clip the CT on top of a wrap of tape, but then you'll have to take some care to make sure there's not so much tape that the two halves of the ferrite get separated.

While there are definite differences in transformer outpt due to orientation within the magnetic field around the conductor under observation, I doubt they really come into play for this application. There are likely larger error sources elsewhere in the system, as long as the relative CT angle isn't severe. The CTs specified aren't exactly precision wound either, so even if you have the CT housing perpendicular to the conductor, not all the transformer wires will be. I agree that keeping it stationary once calibrated is the more important aspect here.

Ferrite is a brittle material, and I would be very wary of recommending that anyone should force the c.t. over a wrapping of tape, as it could very easily shatter the core.

And I'd still like to know the basis for thinking that the orientation of the core affects the output in any significant way. I think it's much more likely that you observed the effect of misalignment of the core halves. I'll try to come back tomorrow with some measurements, as I've got my c.t. test rig set up having just done some phase error measurements.

I have now made some measurements. The test rig was the one used for the c.t. tests, the primary a single strand of uninsulated copper wire 1mm in diameter, the test current 5 A approx.

With arbitrary units for primary and secondary currents scaled so that the ratio of I_s/I_p was approximately 1, with the conductor centred in the aperture and perpendicular to the face of the c.t., the average of 4 readings was I_s/I_p = 1.00518. However, the spread of readings was 1.00279 to 1.00679,  i.e. ±0.19%.

With the conductor touching the core at the mating surface next to the hinge, the average was 1.00599, i.e. +0.08%

With the conductor slanted at 30° in the plane of the hinge, the average was 1.00518 i.e. ±0.0%

From that I conclude that the assertion that the orientation of the c.t. on the cable will affect the reading is groundless.

I don't at all understand the theoretical basis, but I think it requires Maxwell's equations since Amp's law doesn't deal with asymmetry. I'm not even totally sure that CT orientation variance within the field necessarily qualifies as asymmetry for purposes of Amp's law.

Thanks for taking the time to run some tests. I don't at all doubt that, for these purposes, any difference in orientation doesn't affect the measured current in any substantial way. Your tests bear that out.

There could be a change in phase shift or other aspects of the wave that aren't apparent by taking the measured current on its own. I would be interested to see what a scope tells us. Ideally, simultaneous measurement of primary voltage and CT ouput would be good. It would also be good to get an idea of how things look under changing current conditions. This would tell us if there is any nonlinearity induced by a difference in orientation. Ideally the current would vary from the lower to upper bounds of the CT, 1A to 100A or so, but not sure the high end of that is attainable with readily available equipment.

Barring any movement of the CT during a measurement, I'd think that the maximum chance of any difference would be a rotation of the primary along both the plane of the hinge and the plane that passes perpendicular through the hinge. In other words, oriented such that the wire is touching the coil at the corner of the CT housing on one side, but touching the corner of the CT housing furthest away.

If you want that (just one measurement from each set):

Centred:

Phase error: 6.65°

Perpendicular - offset:

Phase error: 6.57°

Slant:

Phase error: 6.69°

The phase error of one from the recent batch of YHDC 100 A c.t's varies from around 1° lag at 100 mA to a maximum lead of 4° around 2 A, falling to 2.5° around 50 A only to rise again to 2.68° at 100 A with an 18 Ω burden, and 6.4° at 40 mA rising to 7.5° and falling again to 5.5° at 16 A for a 120 Ω burden (which is what I used for these tests). The phase shifts are for the 50 Hz component.

Now will you believe it makes no material difference?

For the standard type of YHDC CT, with its plastic "closure springs", there is very little force holding the two parts of the ferrite core together.  It is important to make make sure that nothing prevents these feeble plastic components from doing their job.  A good joint between the two parts of the core is vital.

Ideally, the wiring should be arranged so the CT sits perpendicular to the cable that it is clipped around.  Increasing the diameter of the cable with tape so that the core is more central within the CT's aperture can only be a good thing.  But if the CT naturally wants to sit skew, then it may be best to leave it alone.  Wrapping tape around the cable in order to "square up the CT" could put more strain on the "plasticery", as well as the ferrite core.

Robin - It makes no significant difference. 0.1° or so of phase error disappears into obscurity when compared to the variation with current. Read the numbers. I've spent 2 days measuring phase errors on c.t's and I'll publish the results in due course. I can assure you from bitter experience that it is really, really tricky to get consistent results on either side of opening the c.t., quite often ½° difference pops up no matter how much jiggling and tapping I do to get the cores to seat properly. So as I said a long way above and as you've repeated, tape could well be a very bad idea.

Robert, thanks for running all those tests. Definitely nothing to worry about here.

Definitely have to be careful with using tape. I mostly used tape as a way to keep the CT from sliding down the cable. No pressure is exerted on the ferrite.