Hi, all.

I've been plagued with intermittent nuisance trips on one of the 2 RCDs
in the new 'cheapest way to achieve 17th Edition'CU that our clueless
sparky installed during recent building work.

In due course, I plan to install a new CU which contains all RCBOs.

But in the meantime, I wanted to make some diagnostic measurements to
see what the actual residual leakage was, and where it was coming from.

Clamp meters are quite expensive, so I thought I'd D-I-Y...

I basically needed the current transformer with sense coil.
An old RCD was the donor for this.
I got an old Merlin Gerin 2-pole RCD, and carefully opened it.
I removed all the actual moving parts: toggle, contacts, springs etc.
I soldered the output braids directly to the output terminals.
It's now just a feed-thru.
All that was retained was the housing, current transformer, terminals
and a small PCB where the fragile sense coil wires terminated.
The PCB was held in place inside the housing by the actuation solenoid,
which was no longer required, so the tracks to it were cut.
The sense coil was in fact centre-tapped, and a pair of diodes provided
full-bave rectification to drive the solenoid. I did not use any of
this, and simply ran a pair of wires directly from the outer ends of the
coil, and ran these out to a pair of 4mm sockets.

During calibration, I found that the output voltage drifted about a bit
high when open-circuit, so I measured the original solenoid at 7k-Ohm
DC, and so I added a 10k load resistor across the coil. That settled
things right down.

Calibrarion consisted of using a handy 24v transformer to provide an AC
input, and a pair of Fluke meters. A series of resistors was chosen to
provide nominal calibration points at 0,1,2,3,4,5,6,7,8,9,10 mA, then
15,20,25,30,35,40 mA. The resistors were hooked up in the usual way to
provide an imballance ( 1 leg before the current transformer, one after.

The actual mA imballance was measured with one fluke, and the mV AC
output measured with another. The calibration was very linear, that's
why I went to bigger steps after 10mA. In my example, I got around 45 mV
/ mA.

After calibration at 24v ( keeps the power dissipation in the test
resistors down ), I did a couple of check points at 240v, and it was
spot on. It's easily accurate to within 0.5mA against the fluke, with
a resolution better than 0.1mA.

I'll post a photo.

OK, not the convenience of a clamp-on, but at least I can now make
meaningfull measurements and start fault-finding!

--
Ron

On Mar 3, 6:31*pm, Ron Lowe wrote:
Hi, all.

I've been plagued with intermittent nuisance trips on one of the 2 RCDs
in the new 'cheapest way to achieve 17th Edition'CU that our clueless
sparky installed during recent building work.

In due course, I plan to install a new CU which contains all RCBOs.

But in the meantime, I wanted to make some diagnostic measurements to
see what the actual residual leakage was, and where it was coming from.

Clamp meters are quite expensive, so I thought I'd D-I-Y...

I basically needed the current transformer with sense coil.
An old RCD was the donor for this.
I got an old Merlin Gerin 2-pole RCD, and carefully opened it.
I removed all the actual moving parts: toggle, contacts, springs etc.
I soldered the output braids directly to the output terminals.
It's now just a feed-thru.
All that was retained was the housing, current transformer, terminals
and a small PCB where the fragile sense coil wires terminated.
The PCB was held in place inside the housing by the actuation solenoid,
which was no longer required, so the tracks to it were cut.
The sense coil was in fact centre-tapped, and a pair of diodes provided
full-bave rectification to drive the solenoid. * I did not use any of
this, and simply ran a pair of wires directly from the outer ends of the
coil, and ran these out to a pair of 4mm sockets.

During calibration, I found that the output voltage drifted about a bit
high when open-circuit, so I measured the original solenoid at 7k-Ohm
DC, and *so I added a 10k load resistor across the coil. * That settled
things right down.

Calibrarion consisted of using a handy 24v transformer to provide an AC
input, and a pair of Fluke meters. * A series of resistors was chosen to
provide nominal calibration points at 0,1,2,3,4,5,6,7,8,9,10 mA, then
15,20,25,30,35,40 mA. * The resistors were hooked up in the usual way to
provide an imballance ( 1 leg before the current transformer, one after.

The actual mA imballance was measured with one fluke, and the mV AC
output measured with another. * The calibration was very linear, that's
why I went to bigger steps after 10mA. In my example, I got around 45 mV
/ mA.

After calibration at 24v ( keeps the power dissipation in the test
resistors down ), I did a couple of check points at 240v, and it was
spot on. * It's easily accurate to within 0.5mA against the fluke, with
a resolution better than 0.1mA.

I'll post a photo.

OK, not the convenience of a clamp-on, but at least I can now make
meaningfull measurements and start fault-finding!


Neat idea. Fancy posting this on the wiki?
http://wiki.diyfaq.org.uk/index.php?...ecial:Allpages
To create an article just type the name of it you want into the search
box

If you wanted to post a pic of the current transformer, it would make
it easier for people to copy it without pulling an rcd apart.


NT

In article ,
Ron Lowe writes:

During calibration, I found that the output voltage drifted about a bit
high when open-circuit, so I measured the original solenoid at 7k-Ohm
DC, and so I added a 10k load resistor across the coil. That settled
things right down.

The clue here is in the name - it's a _current_ transformer,
and so the output you want is the current, and not the voltage.
You can turn the current into a voltage by passing it through a
resistor, as you did.

Calibrarion consisted of using a handy 24v transformer to provide an AC
input, and a pair of Fluke meters. A series of resistors was chosen to
provide nominal calibration points at 0,1,2,3,4,5,6,7,8,9,10 mA, then
15,20,25,30,35,40 mA. The resistors were hooked up in the usual way to
provide an imballance ( 1 leg before the current transformer, one after.

The actual mA imballance was measured with one fluke, and the mV AC
output measured with another. The calibration was very linear, that's
why I went to bigger steps after 10mA. In my example, I got around 45 mV
/ mA.

After calibration at 24v ( keeps the power dissipation in the test
resistors down ), I did a couple of check points at 240v, and it was
spot on. It's easily accurate to within 0.5mA against the fluke, with
a resolution better than 0.1mA.

I'll post a photo.

OK, not the convenience of a clamp-on, but at least I can now make
meaningfull measurements and start fault-finding!

It's prudent to issue a warning about current transformers.
This goes back to the output being a current. If you leave
the output open circuit, then the current transformer will
need to develop thousands of volts in order to breakdown
the insulation so it passes the necessary output current.
In yours, the core will saturate limiting the output
current as soon as the insulation breaks down, but you
might still get a nasty belt off it. In larger ones
used to measure large load currents, you can get them
arcing between terminals if left open circuit, and they
are lethal. The moral here is never leave the output of
a current transformer open circuit. Always leave a
resistor across it (or short circuit it) to prevent the
output voltage heading off to infinity when the output
resistance is infinity. In your case, I would solder the
resistor into the old RCD carcuss, so it remains connected
if the external circuit is disconnected.

Also, bare in mind what happens if there is a fault, and
you get, say, a few hundred amps of inbalance for the time
it takes for the fuse/MCB to trip. Again, in your case the
core will saturate limiting the output power, but your
circuitry still needs to handle this. The core will have
been designed for only very short exposure to these
currents as it would have tripped the RCD, but you've
take the rest of the RCD away, so you're now relying on
something else to disconnect the supply, and it might not
act as quickly.

BTW, I did something similar about 15 years ago to measure
earth leakage in a computer room, but I was using a 5VA
160:1 current transformer, and that could have done some
real damage if I got it wrong. (That CT was used to measure
the current on our 500kVA supply, before it was upgraded
to 2MVA and needed bigger ones.)

--
Andrew Gabriel
[email address is not usable -- followup in the newsgroup]

On Mar 4, 9:46 am, (Andrew Gabriel) wrote:
In article ,

It's prudent to issue a warning about current transformers.
This goes back to the output being a current. If you leave
the output open circuit, then the current transformer will
need to develop thousands of volts in order to breakdown
the insulation so it passes the necessary output current.
In yours, the core will saturate limiting the output
current as soon as the insulation breaks down, but you
might still get a nasty belt off it. In larger ones
used to measure large load currents, you can get them
arcing between terminals if left open circuit, and they
are lethal. The moral here is never leave the output of
a current transformer open circuit. Always leave a
resistor across it (or short circuit it) to prevent the
output voltage heading off to infinity when the output
resistance is infinity. In your case, I would solder the
resistor into the old RCD carcuss, so it remains connected
if the external circuit is disconnected.

I was once asked to make a bunch of 5000:1 current transformers (i.e.
4999 turns).
To test them I had a large transformer with a one turn secondary
consisting of aluminium bars totalling 6x1 inches in cross section.
I never dared to have the CTs open circuit.

NT wrote:
On Mar 3, 6:31 pm, Ron Lowe wrote:
Hi, all.

I've been plagued with intermittent nuisance trips on one of the 2 RCDs

Clamp meters are quite expensive, so I thought I'd D-I-Y...



Neat idea. Fancy posting this on the wiki?
http://wiki.diyfaq.org.uk/index.php?...ecial:Allpages
To create an article just type the name of it you want into the search
box

If you wanted to post a pic of the current transformer, it would make
it easier for people to copy it without pulling an rcd apart.


NT

OK, I'll write it up with some pics.
Will take a few days.
I'm busy chasing faults with it now...

--
Ron