I've been looking into energy monitoring (240VAC mains) lately and I've come across seemingly conflicting info. On the one hand, a lot of advice I've seen online stress the importance of galvanic isolation between the HV section and the MCU/logical section of the circuit, for the safety of the circuit and its designer as well. This is usually done with a low-rated transformer to reduce mains to some more manageable voltage below 12 V, for instance.

On the other hand, I've looked up a number of energy-metering ICs (like the ADE and STPM series) and all the app notes I've examined so far (such as this PDF), show circuits where the IC analog front-end is connected directly to mains live (albeit through some heavy attenuation) and the neutral is directly connected to the AGND, which in turn is linked to the DGND of the control section (sometimes through an inductor to dampen transients apparently). No galvanic isolation whatsoever.

So is all this stuff really contradictory or am I missing something? What's the 'right' way to take these measurements?

  • \$\begingroup\$ You need galvanic isolation somewhere. The case can suffice. \$\endgroup\$ Commented Nov 17, 2017 at 0:06
  • \$\begingroup\$ the app note to which you posted a link (mouser.com/ds/2/609/ADE7753EB-246777.pdf) shows an optically isolated output \$\endgroup\$
    – jsotola
    Commented Nov 17, 2017 at 0:14
  • \$\begingroup\$ @jsotola Yeah it's isolated from the MCU optically. What I'm concerned about is the neutral-AGND-DGND connection. \$\endgroup\$ Commented Nov 17, 2017 at 0:32
  • \$\begingroup\$ @IgnacioVazquez-Abrams Case? \$\endgroup\$ Commented Nov 17, 2017 at 0:33
  • \$\begingroup\$ i see what you mean ... disconnect JP11 \$\endgroup\$
    – jsotola
    Commented Nov 17, 2017 at 0:44

1 Answer 1


The classic power monitoring circuit is the kill-a-watt. It is a 'hot power supply' design. Galvanic isolation is provided by the case (meaning: assume the traces on the inside are connected to the mains). The user is kept safe by virtue of the plastic case.

Hot designs are done to save cost and space (transformers are costly and big, even in SMPS designs). But you can't attached scope probes to them or you will trip your GFI. To work on them, put an isolation transformer between the device and the mains. Then your scope will work.

In hot designs, very often the ground planes is 'starred' so that separate sections of the design are isolated from one another. This is done by picking a central ground point and running ground traces out to different sections. This is how analog and digital grounds are connected together.

  • \$\begingroup\$ or put an isolation transformer between the scope and the mains, but be very aware that you have a "hot" scope when you connect it to the device (or use a battery powered scope, still "hot" though) \$\endgroup\$
    – jsotola
    Commented Nov 17, 2017 at 6:21
  • \$\begingroup\$ Is there no practical way to isolate the mains neutral (or live!) from AGND/DGND? \$\endgroup\$ Commented Nov 17, 2017 at 17:40
  • \$\begingroup\$ Hot designs have no magnetic field between the mains and the circuit. By design they are not isolated. They are used in situations where cost and size are a concern. To get isolation, throw in a power supply design that 'transforms' the mains current 'loop' into your own current 'loop' local to the circuit. \$\endgroup\$ Commented Nov 17, 2017 at 18:22
  • \$\begingroup\$ @TylerBrooks And the dangers to the circuit of making such 'hot' connections? Is the whole 'starred GND plane' + inductor between AGND and DGND, usually enough to protect the MCU from dangerous mains transients? \$\endgroup\$ Commented Nov 19, 2017 at 1:58
  • \$\begingroup\$ You have to use every trick in the book. For instance, put a TVS (transient voltage suppressor) from your power rail to your ground. This protects the circuit from voltage spikes but it also protects against 'ground bounce' which will momentarily send your ground potential above your power rail (this can really cause problems that are hard to find). Any lines that touch the power (like your power monitoring lines) should have some form of TVS protection. Also, use ferrite beads on sensitive lines like your reset line. Otherwise, you will get random resets that you can't explain. \$\endgroup\$ Commented Nov 19, 2017 at 5:21

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.