AC mains current monitoring

Question

I'm designing an ATmega328 PCB with an optoisolated TRIAC controlling a 100W (~ 0.5A, 230V, PF = 0.98) load. I'd like to monitor the current with the processor and report an error if the current drops. The load is an electronic ballast driving 3 fluorescent lights and they are wired so that if one goes out, all of them go out. Therefore I only need to know if there is current present and not really measure it.

So far I've encountered 3 different solutions:

1. A current sensor IC like ACS712 that uses the Hall effect (I later found out that ACS712 is not suitable for mains voltage)
2. A shunt resistor with an opamp (galvanic contact problem due to Vcc on opamp)
3. A current transformer for galvanic isolation and shunt resistor like in the previous point

I have no idea which one is the easiest to implement (and cheapest) and which one is the safest. I'm not really sure about the exact circuits I should use for either of the solutions.

I'd really appreciate some help on this matter as well as an appropriate circuit. If there is more information needed let me know.

Answer 1

Allegro is the right Idea [it's one of the top names in my head when it comes to high current messurement]. They have chips which are suitable. A just very quick look on their product table: E.g. https://www.allegromicro.com/en/products/sense/current-sensor-ics/zero-to-fifty-amp-integrated-conductor-sensor-ics/acs730#PartTable should be better than the ACS712.

LEM is also an known company for isolated high current messurements, like:

https://www.lem.com/en/lzsr-series

https://www.lem.com/en/hmsr-miniature-current-sensors

---

Other Options are isolated current sense amplifiers (here TI and Analog Devices are the names on my list...) maybe

https://www.ti.com/product/AMC1300/datasheet

AD's current product range seems to offer isolated ADC's for that purpose:

https://www.analog.com/en/parametricsearch/11101#/

---

the solutions with hall like Allegro and LEM have the advantage of less looses. But I guess are more expensive.

Answer 2

I have used LEM, F. W. Bell, and Allegro parts, and they are great for solving a more complicated problem. But for a simple go / no-go sensor, I vote for the current transformer. Low cost, outstanding reliability, and easy to read.

Select a burden resistor value, plus a zener diode, and the clipped sine wave can be connected to any GPIO pin and detected as an incoming squarewave signal. Or add one small-signal diode and one capacitor, and now you have a DC level that represents the magnitude of the mains circuit current. Scale things such that the incoming signal is not large enough to be read as a logical 1 at the GPIO pin when the AC current is below your threshold value.

Answer 3

The current transformer could give you arbitrarily high galvanic isolation (>5kV) so that is your best bet if that is your top requirement. It's simple, tested, and true. This will likely be the largest (physically) solution.

The hall-effect IC's that others mentioned work great, are around similar price to a 1A current transformer, and provide isolation up to ~ 5kV. Should be a smaller solution footprint than the current transformer.

But if you're designing the PCB, it's multilayer, and the galvanic isolation requirements aren't too steep, then what I would do is etch the transformer directly in the PCB - you basically etch an Archimedes spiral of copper directly on the PCB and pass the current to be detected through the center of it. You now end up with a voltage proportional to the current on the ends of the spiral; the PCB becomes the current transformer and it's free with purchase of PCB. For additional galvanic isolation (creepage and clearance distances), consider placing terminal blocks (one on each side of the spiral center) as opposed to running the current-carrying tracks over the inner layers spiral. In fact, if your stuck with a 2-layer PCB, than you're going to need to use terminals or pins to pass current through the spiral trace.

I've done this in the past to measure harmonics on an AC line - the the output of the spiral ran through 2 or 3 op-amps for signal conditioning and the DAQ downstream processed the analog data.