I am looking to add overvoltage protection to a circuit of mine.

A standard way of doing this is with a crowbar circuit with a fuse to blow when tripped, thus disconnecting the circuit. However, I don't want to have to deal with replacing blown fuses. This leads to looking at resettable polyfuses.

However, I'm concerned about the condition where the overvoltage is maintained for a period of time - this would lead to the intended effect immediately, followed by a period of no current through the polyfuse, allowing it to cool, at which point we eventually get to a point where enough current flows to raise the voltage past my tripping point and the process repeats until the overvoltage is removed - this seems like unnecessary stress on the components.

Is there a better way of managing this situation? I've not really used polyfuses before - how long do they take to "reset"? The datasheet of the one I'm looking at doesn't mention this at all, but perhaps the period is long enough that it wouldn't really matter.

Simplified circuit for reference: Reference Circuit

  • \$\begingroup\$ What duration of overvoltage are you worried about? Seconds? Minutes? Hours? \$\endgroup\$
    – marcelm
    Jun 12, 2019 at 14:15
  • \$\begingroup\$ In the order of minutes to hours - essentially someone plugging into the wrong voltage and taking the time to realise something isn't right when the device isn't working, at which point hopefully they remove the power! \$\endgroup\$ Jun 12, 2019 at 14:19

3 Answers 3


Polyfuses don't actually go fully open like regular fuses do. When a polyfuse trips, it goes into a high-resistance state, continuing to pass current. This current (which is far smaller than the fault current, of course; that's the point of using one!) is sufficient to keep the polyfuse warm. It won't cool down until the power is completely removed and the fuse allowed to cool.

  • \$\begingroup\$ So this relies on the polyfuse not increasing its resistance enough, such that the current is limited to the point where the SCR turns off, otherwise there is nowhere for current to flow - how can I be sure this happens? Also assuming a current does still flow - you are saying that even though this is below the hold current of the polyfuse, it still maintains its tripped state? I guess the higher resistance balances the lower current for the heating... \$\endgroup\$ Jun 12, 2019 at 15:17
  • \$\begingroup\$ Right, the resistance and current balance out naturally. If I understand polyfuses right, as it heats up, its resistance will increase until the current reaches the hold current, at which point it has found a stable equilibrium and will stop changing. If this hold current is below the thyristor's hold current, though, you may have a problem, as the thyristor will switch off, allowing the polyfuse to cool again. Just make sure your polyfuse's hold current is more than your thyristor's and you should be fine. \$\endgroup\$
    – Hearth
    Jun 12, 2019 at 15:23
  • \$\begingroup\$ I don't think I can design for this though - the current the polyfuse sees will depend on the short circuit current of the supply used, which is what is going to affect the initial heating of the poly. Whilst I might be able to do some calculations to try and figure out what the steady state will be, I don't think I can guarantee it won't go over this and turn of the SCR \$\endgroup\$ Jun 12, 2019 at 16:03
  • \$\begingroup\$ Okay - the bit I was missing was the Pd Typ rating of the polyfuse (Power dissipated from device when in the tripped state at 20°C still air) - using this with the fact that the polyfuse will see more or less the entire rail voltage, you can calculate the "leakage" current in the tripped state for a given voltage. \$\endgroup\$ Jun 13, 2019 at 15:53

Polyfuses don't oscillate like that. They maintain their high-resistance state (allowing a small current to flow) indefinitely, until the current is interrupted altogether by some other means. You'll be fine.

  • \$\begingroup\$ This assumes that the high-resistance state of the polyfuse doesn't limit the current below the hold current for the SCR though - how can I guarantee that? \$\endgroup\$ Jun 12, 2019 at 15:42
  • \$\begingroup\$ Compare the specifications in their datasheets. \$\endgroup\$
    – Dave Tweed
    Jun 12, 2019 at 16:35
  • \$\begingroup\$ Also, there are other ways to protect against overvoltage that might be more applicable than a crowbar anyway. For example, you could use a P-channel MOSFET in place of the polyfuse and simply shut if off if the input voltage is too high. \$\endgroup\$
    – Dave Tweed
    Jun 12, 2019 at 16:41

PTC's do not fail or operate this way. They can change R by several orders of magnitude.

However, this design is no longer resettable if you crowbar with an SCR if the PTC current limit exceeds the holding current. This would require manual power cycle and is your choice. A better design would have OVP, OTP and auto restart like all LDO's, but many of these have a high dropout voltage, (>2V) unlike modern MOSFET LDO's which may be preferred.

PTC Resettable Fuses act like a self-contained thermistor-controlled oven and are designed for these characteristics well above the maximum operating temperature such as 85'C max and transition at 130'C. This means will be burning hot and are made of flame-proof polymer materials and are suitable for Flame-Retardant substrates like FR4. They come in SMD and radial ceramic packages.

Since the proportional gain or sensitivity of the curve is high, the difference in temperature for a wide range in currents is small. So it acts like an analog controlled constant temperature ceramic heater. They must have TUV/UL fire safety approvals.

They are defined by these parameters;

  • Current - Hold (Ih) (Max)
  • Current - Trip (It)
  • Time to Trip
  • Resistance - Initial (Ri) (Min)
  • Resistance - Post Trip (R1) (Max)
  • Operating temperature range (normal)

Sample datasheet

My only concern is MTBF if there is a short. Often, datasheets will say " Life time-up to 10 times longer" , longer than what?? a fuse ? (lol)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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