I know I shouldn't. But I just wanted to ask due to a specific requirement. I also searched the forum but couldn't find any similar question. Please give me a link if there's any.

I'm asked to modify one of our own design commercial three-phase electric meters according to a big customer's requirement: The e-meter should never affected by outer magnetic flux densities up to 300mT, and it should be as cheap as other cheaper brands' e-meters.

I decided to modify and "cheapen" the power supply first. In the original design the power supply consists of a flyback regulator following a three-phase full-wave rectifier (simplified):


simulate this circuit – Schematic created using CircuitLab

VDD is the supply for digital blocks (MCU, ADCs, communication interfaces etc.) and, as can be seen from the simplified schematic above, its ground (shown as SEC-GND) is tied to the NEUTRAL through a ferrite bead to guarantee accurate phase voltage measurements.

So I decided to make a transformerless power supply since it is relatively cheaper and is relatively resistant to magnetic fields due to the lack of magnetic components (i.e. inductor/transformer):


simulate this circuit

The problem is that I don't know what should I do to guarantee accurate phase voltage measurements. I know I shouldn't connect the rectifier's GND to the NEUTRAL even if the system is ideally balanced (The system wil never be balanced because the phase voltage differs in tougher environments here in Turkey).

Any suggestions?

  • \$\begingroup\$ You must isolate on the measurement side now. \$\endgroup\$
    – Jeroen3
    Commented Jan 24, 2020 at 14:14
  • \$\begingroup\$ Does your system need to be isolated as well? The old design was. \$\endgroup\$
    – Aaron
    Commented Jan 24, 2020 at 17:02
  • \$\begingroup\$ @Aaron isolation is not needed. Our first concern is the cost since a flyback regulator (isolated or not) is expensive. \$\endgroup\$ Commented Jan 24, 2020 at 17:54

2 Answers 2


Directly connecting neutral should not be a problem, many single-phase transformerless supplies do that.

enter image description here

This image from atmel app note AVR465 "single phase energy meter", you may find the remainder of that document interesting too,

A three phase supply made by replicating the capacitor and rectifier part will work in a similar way but with less ripple at the regulator input.

  • 1
    \$\begingroup\$ Maybe you could elaborate? \$\endgroup\$
    – Voltage Spike
    Commented Jan 25, 2020 at 3:20

A power supply that uses transformers is okay, provided that the magnetic field will not cause the transformer to malfunction. You don’t need the transformer to be involved in the measurements.

A line-powered switching regulator is a good choice, in my opinion. It does use one or two transformers, but they are high-frequency transformers, which can be much smaller and cheaper. That’s why they are so commonly used in consumer electronics. They are, however, noisy, so you will need to be sure to provide adequate filtering. Such power supplies may be available as commercial off-the-shelf products.

  • \$\begingroup\$ A line-powered switching regulator is already our first solution in the original design, but it's expensive (a 1000-V MOSFET, controller, transformer, etc) for the customer's special requirement. ...provided that the magnetic field will not cause the transformer to malfunction... In our tests, we use a magnet having a DC flux density of about 250mT. And it can "stop" the regulator because the flyback transformer gets kinda saturated. That's why we place hall-effect sensors around the electricity meter's PCB to detect and log the presence of an unexpected outer magnetic field. \$\endgroup\$ Commented Jan 25, 2020 at 12:49
  • \$\begingroup\$ And the customer expects resistance to heavier magnetic flux densities (increased from 200mT to 300mT). So we should use either a bigger transformer to prevent saturation easily, or use a power supply with non-magnetic components. That's why I go for a capacitive power supply. \$\endgroup\$ Commented Jan 25, 2020 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.