I want to design a safety circuit for a battery powered device which measures human body impedance. It is intended to be powered only by an external battery pack (5V) via a USB cable. But I want to make sure that even if device is connected to wrong power supply by mistake (e.g. a faulty USB charger leaking 230V), the user is safe. I am also limited in PCB size. What kind of circuit or IC should I look for this purpose? I couldn't find any regulators that can tolerate such high input voltage.

  • \$\begingroup\$ measures human body impedance? Uhhh... wouldnt a mulimeter do that (still at lower voltages)? \$\endgroup\$ Commented Jul 10, 2014 at 11:42
  • 1
    \$\begingroup\$ Such devices should be designed so it is not possible to connect anything except the intended supply - except with vast purposeful effort. \$\endgroup\$
    – Russell McMahon
    Commented Jul 10, 2014 at 12:47
  • \$\begingroup\$ @RaghunathV, no, a two-lead device like a multimeter would not give you an accurate measurement. \$\endgroup\$ Commented Jul 10, 2014 at 13:14
  • 10
    \$\begingroup\$ I hope that your medical devices aren't designed just with advice from random people on the internet, irrespective of their reputation on this site. You must get a trustworthy independent review of your design for your own sake as well as the patients'. \$\endgroup\$
    – Martin
    Commented Jul 10, 2014 at 16:17

5 Answers 5


You need to study isolation. If you put a USB connector on the device, you'll need two means of patient protection, 5kV of isolation, and 250V working voltage to be medically safe. There are isolated DC-DC modules available. All of this is contained in IEC-60601. Study it. People are killed by designers that don't follow the rules.

  • 1
    \$\begingroup\$ I think zud will have to treat the USB connection exactly as if it was a mains connection (so isolation, maximum leakage current and so on) for safety purposes. Most DC-DC converters require a capacitor from input to output.. does IEC-60601 distinguish between reactive current (mains frequency current through the capacitor) and "leakage current"? \$\endgroup\$ Commented Jul 10, 2014 at 13:15
  • \$\begingroup\$ Treating it as though it were mains connected is where the working voltage comes from. Isolated DC-DC modules use a transformer as an isolation barrier. I believe a sufficiently rated Y5 cap counts as a means of patient protection. As to how the standard specifies leakage currents, I don't know off hand, this is all from one of my back burner projects that I will get back to in a couple months. \$\endgroup\$
    – Matt Young
    Commented Jul 10, 2014 at 13:25
  • 4
    \$\begingroup\$ +1 for naming the applicable standard and isolation voltage. \$\endgroup\$
    – user16324
    Commented Jul 10, 2014 at 13:32
  • 1
    \$\begingroup\$ @SpehroPefhany I've never seen an isolated DC-DC converter that requires a cap from input to output. Not that they're not out there, but there are plenty that don't. \$\endgroup\$ Commented Jul 10, 2014 at 13:39
  • \$\begingroup\$ @SpehroPefhany The standard US hospitals generally test to is NFPA-99. It doesn't care where the current comes from. FWIW, there are different current requirements for different types of gear. \$\endgroup\$ Commented Jul 10, 2014 at 13:44

To make this stand up to mains voltage on Vcc seems a bit misguided (as opposed to making sure it's never plugged into mains), but here's how I would approach it-- as @BrianDrummond suggests.

If that's not realistic for your case, I recommend completely isolating the stimulus and receiver by using transformers, and diode or zener-clamping both the inputs and outputs to the transformers. The concern is that if you have 230V AC on these diodes, they'll fry up and stop doing their job. I'd try to look for diodes that can hold up to that, but I'm not sure you'll have luck.

Thus, to be double safe I'd probably use an isolated DC-DC converter to generate Vcc, paying attention to the likely failure modes of the device you use. Mains would be unlikely to punch through. Hopefully, there'd be no way for the 230V AC to find its way to the isolated side, but the research on that would be up to you. Even if the mains can poke through for a little while, the idea is to have the diodes last long enough for the converter to fail.

Frankly, the easier way to do this is to just use an IEC 60601 compliant power supply or just build in a battery holder so there's no other way to plug in the wrong supply.


One approach against casual mis-connection is to eliminate the USB cable, and to use some other connector style as the connection from battery pack to equipment.

You can use a separate USB connector on the battery pack for recharging purposes, and a 2-pole mechanical "break before make" switch set to "charge" or "use" connecting the battery to either charging circuit or output. Then isolation can be guaranteed up to the voltage rating of the switch.

There are other approaches, but this has the benefits of simplicity and obviousness in its operation, which will help if you must make a safety case for it.


I think I would consider a lithium battery power supply and designing a mechanical interlock of some kind so that the unit cannot possibly be used while attached to the charger (perhaps the connectors for the probes physically cover a micro-USB port).

Also put an electrical interlock in there so even if a determined user defeats the mechanical interlock it can't be used while it is charging. And read the standards, make sure you're 100% compliant with safety standards and current best practices, and maybe then talk to a lawyer about product liability.

I would definitely treat the USB cable as a mains connection if it has any kind of connection to a human being that is different from casual contact, and especially if it's a medical-ish device.


This is a place where you could very very easily kill a patient. Install a battery holder, use the lowest-voltage battery that can possibly satisfy the circuit, build for the highest input impedance you can tolerate, and DO NOT provide any means of connecting the device directly to USB. If data transfer via USB is a requirement, use optoisolators.


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.