I’m starting with the selection of the current transformer (CT) sensor as selecting and sourcing the right component is probably going to be the most challenging part of this project.
Clip-on, split core current transformers such as the YHDC SCT-013-000, that is recommended for the OpenEnergyMonitor, are designed for the non-invasive monitoring of existing cabling. This type of sensor is physically too large to be used being a wall socket (YHDC SCT-013-000: 57 x 32 x 22mm).
The ring-type, solid core current transformers are physically smaller. Whilst they are non-invasive, e.g. not connected in-line, their design requires that the cabling be altered to pass through the CT. That’s OK, I need to rewire the sockets anyway.
Trawling through a virtual mountain of data sheets: there are an (almost) bewildering number of PCB mount CTs that are both small enough to fit behind a socket and reasonably priced; and only a few fully encapsulated CTs that meet those same requirements.
The screw terminals on the back of the sockets are only partially shrouded and CT will be in very close proximity to those terminals. Once the CT is fitted behind the socket and the socket is secured in the box, it will be difficult to visually inspect the assembly. So my preference is for a sensor design that includes fly leads, rather than having to create one by repurposing a PCB mounted sensor.
Based on size and cost, I’ve identified two potential sensors: the Pulse PA3208NL which comes with fly leads; and the slightly smaller, PCB mounted Talema AC1020. They both cost under €7 each.
The AC1020 is 23.8 x 23.8 x 11.1mm and is of a typical iron core type design. It is rated for a primary current of up to 20A RMS. It has a 1000:1 turn ratio and a 1A primary current should result in a 1mA secondary output current.
The PA3208NL is approx. 30% larger than the AC1020, 30.6 x 30.2 x 14.7mm, but has the benefit of coming with fly leads. The sensor uses an alternative design, the Rogowski coil principle, which doesn’t require a metal core. Instead of generating a secondary current that is proportional to the primary current, a Rogowski coil generates a secondary voltage that is proportional to the rate of change of the primary current. The PA3208NL generates a secondary voltage of 383μV/A at 50Hz.
This means that a lossy integrator circuit is required to obtain a voltage that is proportional to the primary current. Which makes this type of sensor harder to use than an iron core type sensor. Great news for me, I like a challenge and it means I get to play with one of my favourite childhood toys – op-amps.
The Selection Process
Assuming it physically fits, the PA3208NL is my CT sensor of choice. So, does it fit?
The answer is yes, just. I created a test harness using a 50mm conduit box (ABB DZ50) and a length of stranded, 3-core, 10A extension cable. With some manipulation I was able to fit the sensor behind the socket and screw on the metal plate holding the socket.
With the thinner (1.0mm²), flexible cable I’m using, it was easier to pass all 3 cores through the sensor but this won’t be possible with the thicker (1.5mm² and 2.5mm²), rigid cable used in my wall sockets.
With the thinner (1.0mm²), flexible cable I’m using, it was easier to pass all 3 cores through the sensor but this won’t be possible with the thicker (1.5mm² and 2.5mm²), rigid cable used to cable my wall sockets.
One area of concern is the potential for damage to the insulation. It took 5 attempts to correctly place the sensor. Doing so resulted in visible but superficial damage to the insulation on the cores.
Another area of concern is that the CT sensor covers up part of the strip (36mm x 10mm) of narrow slots at the back of the socket that provide ventilation for the USB electronics. The sensor only covers up approx. 15% of the total area but nevertheless I’ll be curious to see if this results in a measurable increase in temperature in the box.
Lastly, to make prototyping the sensor electronics easier, I fixed a small breadboard to the conduit box using cable ties.
The sensor electronics.