Core imbalance in RCDs

[Franc Zabkar]

How closely matched are the active and neutral windings in the toroid
of a typical RCD? Would it be possible for the imbalance, if any, to
be of such magnitude (eg 1mA/A ???) that high inrush currents (eg 30A)
could of themselves be enough to trip the RCD?


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[Franc Zabkar]

On 6 May 2005 17:16:07 -0700, bigcat@meeow.co.uk put finger to
keyboard and composed:

Franc Zabkar wrote:
How closely matched are the active and neutral windings in the toroid
of a typical RCD?

IIRC here theyre tested to trip between 5 and 15mA for 30mA rated
units. If those nrs arent exact, theyre about right.


Would it be possible for the imbalance, if any, to
be of such magnitude (eg 1mA/A ???) that high inrush currents (eg
30A)
could of themselves be enough to trip the RCD?

yes, hence they must all be tested and some are rejected.

But how are they tested? Are they tested by applying leakage in the
presence of full load current, or is the leakage applied in the
absence of load current, as is the case when tested by an electrician?
If the latter, then any inherent imbalance would not show up.

Just in case my question is unclear, allow me to rephrase it.

My scenario is one where there is zero earth leakage. Under such
circumstances the core in a perfect ELCB should be perfectly balanced,
ie the net flux should be zero. I'm asking whether it is possible for
there to be a non-zero flux in the core in the absence of earth
leakage. I could envisage this occurring if the active and neutral
windings were not geometrically identical, causing one to contribute
more flux than the other.

Let's assume that manufacturing tolerances produced a core with a
non-zero flux. For a load current of 1A, this flux may be equivalent
to that resulting from a 1mA imbalance, say. In this case, a 30A
inrush current would cause a 30mA imbalance which would be enough to
trip the ELCB.

Is this a feasible explanation for nuisance tripping at switch-on?

When I asked a local manufacturer my original question, their response
was inconclusive:

"There is a possibility of variance in the mA rating of an RCD.
Clipsal conducts the following tests 15mA 2 seconds and 30 mA 40 m/sec
or Less. In domestic applications this would be suitable but we do
offer what we call a G Type. This 'G Type' has a 10mS delay. Tests
have shown that most false tripping occurs within this 10mS period."


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[Tony Williams]

In article <nriq719rr16jpmujd13tbiu36mrbovqksj@4ax.com>,
Franc Zabkar <fzabkar@optussnet.com.au> wrote:

Just in case my question is unclear, allow me to rephrase it.

My scenario is one where there is zero earth leakage. Under such
circumstances the core in a perfect ELCB should be perfectly
balanced, ie the net flux should be zero. I'm asking whether it
is possible for there to be a non-zero flux in the core in the
absence of earth leakage. I could envisage this occurring if the
active and neutral windings were not geometrically identical,
causing one to contribute more flux than the other.

I've done tests on the effect of the geometry of a bar-primary
on the accuracy of a toroidal CT, at 400Hz and up to 500Arms.

The secondary current remained within 0.1% for all manner of
geometries.... off to one side, at an angle going through the
core, even when the bar did an abrupt right-angle as it left
the core. Quite surprising.

Let's assume that manufacturing tolerances produced a core with a
non-zero flux. For a load current of 1A, this flux may be
equivalent to that resulting from a 1mA imbalance, say. In this
case, a 30A inrush current would cause a 30mA imbalance which
would be enough to trip the ELCB.

Is this a feasible explanation for nuisance tripping at switch-on?

Filter capacitance in an appliance would be the first port
of call in most cases.

However, we had a switch-on nuisance tripping in this
house that took me days to find. It was random, but
eventually the pattern was that it happened when any high
current appliance was turned on, anywhere in the house.
All appliances were minutely checked and I started to
suspect the RCD also.... but the explanation turned out
to be stupidly simple.

We had acquired a Neutral-Earth short, caused by recent
work in the kitchen.

If the Neutral voltage is low, an N-E short will not
trip the RCD but what happens is that, every time a
high current load is switched on, some of it's return
current goes back via Earth, not through the Neutral
wire in the RCD. This is what unbalances the RCD.

--
Tony Williams.

 

[Ross Herbert]

On Fri, 06 May 2005 08:13:54 +1000, Franc Zabkar
<fzabkar@optussnet.com.au> wrote:

How closely matched are the active and neutral windings in the toroid
of a typical RCD? Would it be possible for the imbalance, if any, to
be of such magnitude (eg 1mA/A ???) that high inrush currents (eg 30A)
could of themselves be enough to trip the RCD?


- Franc Zabkar

I don't have the A/NZ standards for RCD's but here is a typical RCD
tester which has pretty good specs and measures tripping current +/-
0.25mA and tripping time +/-0.1mS.

http://www.extron.com.au/MeasureSafe%2036A_brochure-v3-4.pdf

 

[Adrian Tuddenham]

Tony Williams <tonyw@ledelec.demon.co.uk> wrote:

I've done tests on the effect of the geometry of a bar-primary
on the accuracy of a toroidal CT, at 400Hz and up to 500Arms.

The secondary current remained within 0.1% for all manner of
geometries.... off to one side, at an angle going through the
core, even when the bar did an abrupt right-angle as it left
the core. Quite surprising.

If you have more than one primary turn, I would imagine that winding the
two primary conductors so that they magnetised different parts of the
core could cause imbalance due to slight variation in the core material.

....Just an aside, in case anyone was tempted not to wind them as a
bifilar pair.

--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

 

[Franc Zabkar]

On 8 May 2005 05:32:10 -0700, bigcat@meeow.co.uk put finger to
keyboard and composed:

The question had just been answered. Are you moron or troll?

Neither. I may be ignorant, though. In any case, you didn't understand
the question. Let me rephrase it in a way that even you can
understand. When an RCD is tested at the factory, is it tested using
active and neutral currents of 30mA and 0mA, or is it subjected to
full load currents of 10.030A and 10.000A? Is it possible for a
10.000A current in the active conductor to magnetise the core slightly
differently than a 10.000A current in the neutral conductor? Judging
from the other responses, the answer is no.

What prompted my question was a flurry of letters to Silicon Chip
magazine discussing the workings of a current transformer. There were
questions as to what constituted a turn or "half turn". This
challenged my understanding of the fundamentals.

Maybe someone can set me straight. The three diagrams depict a
conductor passing through a toroid, or maybe a current clamp meter. In
B and C, the conductor has a single loop, or "turn". In B, the ends of
the loop are outside the core, whereas in C they pass through the
centre. IIUC, cases A and B produce the same flux, but C produces
twice the flux. So a "turn" refers to the number of times the
conductor passes through the centre of the core, not to the number of
loops. Consequently it makes no sense to talk of partial turns. Have I
got this right?

__
/ \ toroid
I -> / \
A ------- | ------------ 1 "turn", 0 loops
\ /
\__/

___________
B __________/___ 1 turn, 1 loop
I -> / __ \
\/ \/
/\__/\
\ /
\__/ toroid

____
/ \ toroid
/ \
| ____|______
C _________\/___/ 2 turns, 1 loop
I -> /\__/\
\ /
\__/


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[Roger Johansson]

Franc Zabkar wrote:

Maybe someone can set me straight. The three diagrams depict a
conductor passing through a toroid, or maybe a current clamp meter. In
B and C, the conductor has a single loop, or "turn". In B, the ends of
the loop are outside the core, whereas in C they pass through the
centre. IIUC, cases A and B produce the same flux, but C produces
twice the flux. So a "turn" refers to the number of times the
conductor passes through the centre of the core, not to the number of
loops. Consequently it makes no sense to talk of partial turns. Have I
got this right?

It would be easier if you see a small magnetic field around any piece
of wire which carries AC current.

This magnetic field, this inductance, can be amplified, or multiplied,
in different ways.

One is to put magnetic material close to the wire, or around it.
Another is to arrange the wire in a loop, each turn will add to the
total magnetic field. Use both methods simultaneously and you get a
coil with a core.

Every part of a turn counts. For example a wire which is 3/4 wound
around a core creates a magnetic field which is 3/4 as strong as one
full turn.

When we look at readymade inductors, on toroid cores for example, we
see that the wires which leave the coil leave it a a right angle. That
is because it makes it easier to calculate the inductance. Wires at
right angles to the coil do not influence that field. You can use
parts of turns in your calculations if needed, if a wire leaves the
coil at another point that the entry point.

Your example pictures;

Number one shows a wire straight through a bead core.
That is not a full turn, it is rather 0.3 of a turn.

Your second examples shows a toroid with one loop, in total 1.3 turns.

The third picture, no comment, can't be sure what it shows.


--
Roger J.

 

[Franc Zabkar]

On 09 May 2005 09:14:59 GMT, "Roger Johansson" <no-email@no.invalid>
put finger to keyboard and composed:

Franc Zabkar wrote:

Maybe someone can set me straight. The three diagrams depict a
conductor passing through a toroid, or maybe a current clamp meter. In
B and C, the conductor has a single loop, or "turn". In B, the ends of
the loop are outside the core, whereas in C they pass through the
centre. IIUC, cases A and B produce the same flux, but C produces
twice the flux. So a "turn" refers to the number of times the
conductor passes through the centre of the core, not to the number of
loops. Consequently it makes no sense to talk of partial turns. Have I
got this right?

It would be easier if you see a small magnetic field around any piece
of wire which carries AC current.

This magnetic field, this inductance, can be amplified, or multiplied,
in different ways.

One is to put magnetic material close to the wire, or around it.
Another is to arrange the wire in a loop, each turn will add to the
total magnetic field. Use both methods simultaneously and you get a
coil with a core.

Every part of a turn counts. For example a wire which is 3/4 wound
around a core creates a magnetic field which is 3/4 as strong as one
full turn.

I would have thought so, too, hence my original question. However,
elsewhere in this thread, empirical evidence was presented that
suggests that it does matter at what angle the conductor enters and
exits the core.

When we look at readymade inductors, on toroid cores for example, we
see that the wires which leave the coil leave it a a right angle. That
is because it makes it easier to calculate the inductance. Wires at
right angles to the coil do not influence that field. You can use
parts of turns in your calculations if needed, if a wire leaves the
coil at another point that the entry point.

Your example pictures;

Number one shows a wire straight through a bead core.
That is not a full turn, it is rather 0.3 of a turn.

See "current transformers can be dangerous":
http://www.siliconchip.com.au/cms/A_104131/article.html

Editor says half turn, reader says full turn.

Also, "disagreement on current transformers":
http://www.siliconchip.com.au/cms/A_104162/article.html

A reader writes:

"With regard to the one conductor through the core being considered as
a full turn, many students question this but the lecturers were
adamant that it was correct and that it definitely could not be
treated as a 'half turn'".

Your second examples shows a toroid with one loop, in total 1.3 turns.

The third picture, no comment, can't be sure what it shows.

Perhaps if I redraw the conductor this way:

Z
A _____________
______________/___ I ->
I -> / \
B \ Y /
\__/

X

If I attach a current clamp meter between points A&B, X&Y, and Y&Z, I
expect to measure currents of I, I, and 2I. This would suggest that
the number of turns at these points are 1, 1, and 2, respectively.
Unfortunately I don't have access to a meter, so I am unable to test
my idea.


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[Franc Zabkar]

On Tue, 10 May 2005 07:51:39 +1000, Franc Zabkar
<fzabkar@optussnet.com.au> put finger to keyboard and composed:

I would have thought so, too, hence my original question. However,
elsewhere in this thread, empirical evidence was presented that
suggests that it does matter at what angle the conductor enters and
exits the core.

Oops, that should read "does not matter".


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[Roger Johansson]

Mark wrote:

There is no such thing as a fractional turn in a transformer.

I think the iron core makes it difficult to measure the fractional turns
effect, because the effect of another pass through the core means so
much more than shaping the wire around the outside of the transformer.

The effect of fractional turns are a lot easier to measure in a coil
without a core with a hole in it.

When there is a core which goes around the wire you get a strong
increase in inductance every time the wire passes through the hole.


--
Roger J.

 

[Tony Williams]

In article <m7mv711o7d1optojtlsk2del9d5qsnih9a@4ax.com>,
Franc Zabkar <fzabkar@optussnet.com.au> wrote:

Unfortunately I don't have access to a meter, so I am unable to
test my idea.

I still have some CT test kit still around,
and it is a trivial experiment to do, as below.

The CT has a 1000 turn secondary and I still have
an old 100-way plug+socket arrangement, with 100
wires, arranged as 50+50 turns. So those 50+50
turns can be the primary, wired either as series-
-adding or series-opposing. A primary current
of 1A is equivalent to 100A in a bar-primary.

Here are the results.

1. Series-adding 50+50 turn pri, 1000 turn sec.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1a. Straight through, return bundle as far away as poss.
Ipri= 1.007A, Isec= 99mA.

1b. Return bundle touching the outer of the toroid.
Ipri= 1.016A, Isec= 99.9mA.

1c. Bundles wrapped to form a 1-turn close loop.
Ipri= 1.02A, Isec= 100mA.

2. Series-opposing 50-50 turn pri, 1000 turn sec.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2a. Straight through, return bundle as far away as poss.
Ipri= 1.020A, Isec= 7.6uA.... yes, microamps.

2b. Return bundle touching the outer of the toroid.
Ipri= 1.030A, Isec= 4.65uA.

2c. Bundles wrapped to form a 1-turn close loop.
Ipri= 1.030A, Isec= 0.75uA.

A wire through the middle of the toroidal CT constitutes
one full turn, what happens around the outside is second
order, and there is no such thing as fractional turns.

--
Tony Williams.

 

[Roger Johansson]

Tony Williams wrote:

and there is no such thing as fractional turns.

If we take a ferrit core shaped like a straight stick, a thousand turns
of wire on it. Like the ferrite antenna in a transistor radio.

I take another conductor, which carries an AC signal and strong
current, and put it close to the ferrite stick, tangentially.

It will induce a weak current in the fixed coil.

If I wrap the outer conductor closer to the core, and wrap it around
most of its circumference the induction will be increased. If I make a
full turn, or more turns, the influence/voltage will increase even more.
But even if the outer conductor only passes by in a straight line some
of the signal will be seen in the fixed coil.

That is an example of a fractional turn transformer.

Fractional turns are much more difficult to measure with a circular
core which goes around the wire, but it must be there, or the
fundamentals of physics wouldn't work.

By the way, I looked up "CT test kit" and got one single hit, it is a
test for chlamydia.
Of course you are talking something else, some transformer designer
kit, but I am not clear over what you have tested and how.



--
Roger J.

 

[Tony Williams]

In article <xn0e22z4s5thmkq005@news.sunsite.dk>,
Roger Johansson <no-email@no.invalid> wrote:

Tony Williams wrote:
and there is no such thing as fractional turns.

If you don't mind I'll drop the full paragraph back in,
because it was carefully worded.

<quote>
A wire through the middle of the toroidal CT constitutes
one full turn, what happens around the outside is second
order, and there is no such thing as fractional turns.
</quote>

Afaik this thread is only about toroidal CTs to which
my remarks were carefully addressed. I have no opinion
about fractional turns for other topologies.

[snip]

Fractional turns are much more difficult to measure with a
circular core which goes around the wire, but it must be there,
or the fundamentals of physics wouldn't work.

As already noted, there is no discernable difference
in the secondary current between a straight 100-turn
primary bundle, or when the bundle is twisted to try
and get a 1-turn tight loop.

In the bench setup here the CT has an internal dia
of about 28mm and the 100-wire bundle has an od of
about 10mm. So the straight bundle can go through
the CT at quite an acute angle. There is little
apparent difference in the secondary current between
the bundle going through the centre at a right angle,
or leaned over at an acute angle. Now that really
flies in the face of the tidy diagrams in textbooks.
I have no explanation for that.

By the way, I looked up "CT test kit" and got one single hit, it
is a test for chlamydia. Of course you are talking something
else, some transformer designer kit, but I am not clear over what
you have tested and how.

Careless wording, sorry. There are some bits
and pieces still around here from when I did precision
CT measurements and selections about 10 years ago.

There are a few actual CTs, 1000-turn secondary,
originally used in an aircraft with a bar-primary
to measure feeder currents of up to 500A, 400Hz.

And there is a little jig that allowed me to put
on 100-turns without actually winding them. It is
just a 100-way plug+socket with 100 wires, wired
so that the socket can be passed through the hole
in the CT and when they are mated the 100 wires
are connected in series. It was a lucky accident
that the arrangement was separate 50+50 wires, so
they can be connected as series-adding or -opposing.

For this experiment the 50+50 turn primary is just
being stimulated at 50Hz, off a variac, low voltage
transformer and series power resistor. Ammeters to
measure the pri and sec current. Bit rough and
ready, but enough to demo what the OP asked for.

--
Tony Williams.

 

[Roger Johansson]

Tony Williams wrote:

quote
A wire through the middle of the toroidal CT constitutes
one full turn, what happens around the outside is second
order, and there is no such thing as fractional turns.
/quote

Afaik this thread is only about toroidal CTs to which
my remarks were carefully addressed. I have no opinion
about fractional turns for other topologies.

Okay, let's concentrate on toroids.
I can imagine an experiment which could solve this problem.

If I had my electronics stuff around I would take a toroid transformer,
put an AC voltmeter on the coil with most turns, to make it easier to
measure the induced voltage. (maybe an amp before the voltmeter)

Then I would send an AC signal through a wire and hold it close to the
core, see its response, start shaping it around the core, 1/4 of a
turn, half a turn, 3/4 of a turn, and see if the voltage goes up
gradually as the wire is shaped better and better around the toroid
core, but without going through it.

I think we will find that the induced voltage goes up as we bring the
wire closer to the core, and as the wire is formed into an increasing
part of a full turn.

It would be interesting to see if a half turn gives exactly double the
induced voltage of a quarter turn.



--
Roger J.

 

[Tony Williams]

In article <xn0e232c35xu54d006@news.sunsite.dk>,
Roger Johansson <no-email@no.invalid> wrote:

Okay, let's concentrate on toroids.
I can imagine an experiment which could solve this problem.

If I had my electronics stuff around I would take a toroid
transformer, put an AC voltmeter on the coil with most turns, to
make it easier to measure the induced voltage. (maybe an amp
before the voltmeter)

Then I would send an AC signal through a wire and hold it close
to the core, see its response, start shaping it around the core,
1/4 of a turn, half a turn, 3/4 of a turn, and see if the voltage
goes up gradually as the wire is shaped better and better around
the toroid core, but without going through it.

1. |
| ____________
|| |---
1A || Toroid | 35mV
||____________|---
|
\|/

2.
____________
| |---
| Toroid | 0mV
|____________|---
1A----------------------->

/
3. /
/____________
|| |---
1A || Toroid | 50mV
||____________|---
\
\
\

4. __________________
| ____________
|| |---
1A || Toroid | 34mV
||____________|---
|__________________




Bit rough and ready. I'll do it more carefully
some time later today, to confirm (or not).

--
Tony Williams.

 

[Roger Johansson]

Tony Williams wrote:

If I had my electronics stuff around I would take a toroid
transformer, put an AC voltmeter on the coil with most turns, to
make it easier to measure the induced voltage. (maybe an amp
before the voltmeter)

Then I would send an AC signal through a wire and hold it close
to the core, see its response, start shaping it around the core,
1/4 of a turn, half a turn, 3/4 of a turn, and see if the voltage
goes up gradually as the wire is shaped better and better around
the toroid core, but without going through it.

1. |
| ____________
|| |---
1A || Toroid | 35mV
||____________|---
|
\|/

2.
____________
| |---
| Toroid | 0mV
|____________|---
1A-----------------------

/
3. /
/____________
|| |---
1A || Toroid | 50mV
||____________|---
\
\
\

4. __________________
| ____________
|| |---
1A || Toroid | 34mV
||____________|---
|__________________




Bit rough and ready. I'll do it more carefully
some time later today, to confirm (or not).

Nice pictures.

The first picture shows a 1/3 turn, as the wire goes straight
tangentially on the outside, result 35mV.

Picture 2, the effects on both sides cancel each other, zero result.

Picture 3, bending the wire shapes it better around the outside,
stronger signal, we could call it 1/3 turn plus a little more, =0.4
turns.

Picture 4, the wire influences the opposite side of the toroid so it
cancels some of the influence on the other side.

We could also put the wire through the core once, to see how much
stronger the influence gets, to compare the result with the fractional
turns experiments.


--
Roger J.

 

[John Perry]

Roger Johansson wrote:

Tony Williams wrote:


Nice pictures.

....but very wrong (electrically speaking).

The first picture shows a 1/3 turn, as the wire goes straight
tangentially on the outside, result 35mV.

There's no such thing as a partial turn. Think about it. If you have
current flow, you have a circuit, which means you have a complete loop
(i. e., turn). It doesn't matter much whether the return path is 1/16"
away, or 12" away, or 100ft away. It's still a complete loop.

The larger the loop is, the more opportunity there is for parasitics to
siphon off some of the field; but the core is generally very high mu,
and offsets the parasitics almost completely.

That's why standard current transformers (remember the discussion of a
few days ago?) work quite well with just a single cable threaded through
the core.

Picture 1 gave a tiny voltage because a tiny fraction of the field
passed through the core hole, inducing voltages into the near wires
without being cancelled by corresponding induction into the opposite
wires on the core.

Picture 2 had roughly equal and cancelling fields inside the hole.

Picture 3 had a slightly higher voltage than 1, probably because the
wires were not bent exactly at the same angles, so got a trifle more of
the uncancelled field in the hole.

Picture 4 is picture 1 plus 2 copies of picture 2. Same voltage out as 1.

I don't know how many turns you had on the secondary, butput the primary
wire inside the hole and the output voltage will be very high if you've
wound it at all like a CT (which is what a GFCI or RCD is, after all).

Put the primary inside the hole, loop it back inside again without going
around the core, and you'll have zero out. That's a GFCI/RCD.

jp

 

[Roger Johansson]

John Perry wrote:

There's no such thing as a partial turn. Think about it.

Think about coils without a core which goes around the wires. Air cores
and cores which are like a straight stick.

In all such cases partial turns work exactly as expected. You can use
it to fine-adjust the transformed voltage relation to the input voltage.

Toroids are just a special case, which cannot change the basic laws of
physics, and cannot change the fact that partial turns give different
voltages,

but the partial turns have much less effect in a toroid transformer,
because the effect is masked by the strong effect of going through the
core or not.


--
Roger J.

 

[Franc Zabkar]

On Tue, 10 May 2005 09:25:05 +0100, Tony Williams
<tonyw@ledelec.demon.co.uk> put finger to keyboard and composed:

In article <m7mv711o7d1optojtlsk2del9d5qsnih9a@4ax.com>,
Franc Zabkar <fzabkar@optussnet.com.au> wrote:

Unfortunately I don't have access to a meter, so I am unable to
test my idea.

I still have some CT test kit still around,
and it is a trivial experiment to do, as below.

Here are the results.

<snip>

Thanks very much for that.

A wire through the middle of the toroidal CT constitutes
one full turn, what happens around the outside is second
order, and there is no such thing as fractional turns.

I guess another way of looking at it is to say that a "turn" refers to
the number of times a conductor passes *through* the core rather than
*around* the core.


- Franc Zabkar
--
Please remove one 's' from my address when replying by email.

 

[John Perry]

Roger Johansson wrote:

John Perry wrote:


There's no such thing as a partial turn. Think about it.


Think about coils without a core which goes around the wires. Air cores
and cores which are like a straight stick.

Of course. Here there's no closed magnetic circuit to confine the
field. There's plenty more leeway for field from one segment of
conductor to avoid all other conductors, or cancel from conductor to
conductor. There is no such leeway in a toroidal or other closed
magnetic circuit.

In all such cases partial turns work exactly as expected. You can use
it to fine-adjust the transformed voltage relation to the input voltage.

Toroids are just a special case, which cannot change the basic laws of
physics, and cannot change the fact that partial turns give different
voltages, but the partial turns have much less effect in a toroid transformer,
because the effect is masked by the strong effect of going through the
core or not.

Actually, no. Toroids enforce the basic laws of physics by not allowing
parts of fields to go astray.

Depends upon your point of view .

jp