There are a lot of "classic" zero crossing detectors out there that drive an optocoupler more or less precisely around the mains voltage zero crossing. I am not so much interested in precision, rather repeatability, but mostly I am wondering if other methods of isolation are being used (inductive, capacitive) and if they allow the power consumption to be lower than 50 mW.

For instance, capacitive digital isolators like iso7710 probably use less than 1 mA @ 2.5V if transmitting short pulses at mains frequency.

Do you know any designs that use other methods of isolation (i.e. not photoelectric)?

  • \$\begingroup\$ Is this for something practical or is just for theoretically possible methods? You're not likely to find much which is isolated and lower power than an optocoupler. \$\endgroup\$
    – jonathanjo
    Commented Sep 14, 2022 at 10:28
  • \$\begingroup\$ Noise pulses on mains can make capacitive-type remote sensors a problem with repeatability, especially if counting cycles is your primary interest....too many crossings are detected. \$\endgroup\$
    – glen_geek
    Commented Sep 14, 2022 at 12:39

3 Answers 3


If you need very low current on the HV side, you can use a MOSFET to shortcut a small transformer during one half wave of the mains. This circuit consumes around 11 mW on the mains side, I think it will also work with 10 Mohm -> 5.3 mW.

On the secondary side of the small transformer you can feed it with a high frequency. The voltage at the secondary side is only present at the negative half wave of the mains (M1 not conducting).

A simple BJT can rectify this, and provide a (more or less) square wave signal of the mains frequency at its collector. The duty cycle is not 50 %, so this does not give exact zero crossing marks at the edges. For this feature a full bridge rectifier on the input side and some other modifications are required.

The high frequency can be almost anything created by a MCU pin, as long as the small transformer's inductance is high enough to leave at least 1 VAC for the BJT drive at open primary side.

The body diode of the MOSFET is a load during negative half wave, but the remaining amplitude on the secondary side is high enough.

There are well isolated small 1:1 transformers in the range of 20-100 uH.

Sometimes it is possible to omit C3 and D2, directly driving the base with the transformer.


simulate this circuit – Schematic created using CircuitLab

An alternative circuit using a D-flipflop worked for me as well:


simulate this circuit

Short positive clock pulses trigger the flipflop, R1 and C1 introduce a little delay and define the sampling point.
If SW1 is open, the voltage at the secondary side winding can rise and the flipflop reads high.
If the transformer is shorted, the flipflop reads low.
R3 and D1 protect the D input on the falling clock edge.

The selected values work for a 1 mH transformer, a lower value for R2 may be needed for lower inductance coils.

  • \$\begingroup\$ I have used common mode chokes as low power high frequency transformers, so they probably can work in this way. They are pretty cheap and commonly available. I used them with a 40 kHz 12V square wave to make an isolated 12 VDC IGBT/MOSFET gate drive supply. \$\endgroup\$
    – PStechPaul
    Commented Sep 15, 2022 at 1:03
  • \$\begingroup\$ Thanks, I've tried this successfully with a tiny 200 µH Ethernet transformer (rated around 1kV) and 9 MOhm in the range 10-100 kHz. Timing could be adjusted with R2, but it's also affected by the temp coefficients of the BJT+diode rectifier. \$\endgroup\$
    – handle
    Commented Oct 11, 2022 at 6:42
  • \$\begingroup\$ Flip M1 so it is on while the body diode is conducting and 100 % off when not. \$\endgroup\$
    – jp314
    Commented Oct 11, 2022 at 21:13
  • \$\begingroup\$ @jp314 Good point, I will try that \$\endgroup\$
    – Jens
    Commented Oct 12, 2022 at 13:16

If cost, complexity and maybe size/weight are of little concern, there are plenty of other isolation methods- transformer (pulse or mains), GMR (Giant Magneto Resistive), capacitive etc. There are packaged isolators with all those methods embedded (but often you'll also have to provide a power supply on the hot side, which means another transformer and more power consumption or some kind of dropper circuit).

If the only issue is power consumption, a better circuit with a reasonably good optoisolator might be a better approach. After all, 5mA for 100us every 1/120 or 1/100 second is only 50 or 60uA.


A small-signal FET with a weak biasing scheme could have about 20MΩ input impedance. Couple this to a small "antenna" and insulator, and you have an inductive field sensor, colloquially called a "chicken stick." Run the output through an op-amp band-pass filter at 55Hz (to cover both 50Hz and 60Hz mains) and/or 110Hz to remove everything else of no interest.

FETs and op-amps can be micropower devices.

A drawback becomes that this must be physically close to the measurement point. It can also pickup "interference" from intentional radiators, such as a WiFi router or other (unintentional radiating) nearby devices.


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.