Regulator + Voltage protection

[MobileWill]

I am working on a power supply design for a microcontroller and need some input. I wanted to have a wide input range in the smallest footprint. I was thinking I could use a 3.3v regulator with a max input voltage of about 12-16v. Then use a over protection circuit to cutout anything greater than that. The circuit is pretty low current, probably 100mA max if even that.  Mostly the source voltage will be 5V but could be higher in some cases.

Would a SCR over voltage circuit work for this application? I found this example http://www.radio-electronics.com/info/circuits/scr_overvoltage_crowbar/scr_over_voltage_crowbar.php Or does this require the power to be cut promptly. I would like to block power until it drops again.

I don't want to add to much cost to my circuit. Would I be better off get finding a regulator with a higher input voltage? I won't have room for any heat sinking or anything like that. Ideally I would like to handle 20-26v without needing to cutout. But if I have to at least I can provide 3.3v from as low as possible with good protection.

Thanks for the help.

[singapol]

You worry too much, most ic regulators have built-in protection of some sort but if you really worry just use a TVS (transient voltage suppressor) across it which is faster than regular protection conventional components.

[MobileWill]

I forgot about those. Would they withstand continues blocking? I would assume so since its a diode. Usually regulators have over current right? Not over voltage protection.

So would I pick one with a 16V clamp? To block above 16V

http://www.digikey.com/product-search/en/circuit-protection/tvs-diodes/655429?k=transient+voltage+suppressor&k=&pkeyword=transient+voltage+suppressor&FV=fff4000a%2Cfff80045%2C1140003%2C1f140000&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=25


I also found this great schematic. The TVS should be in place of the green diode right? Shouldn't there be a resistor so when it kicks in there isn't a short? Or does the TVS limit current?

http://www.infineon.com/cms/en/product/promopages/aim-mc/Selecting-a-linear-voltage-regulator/Voltage-Regulator-Input-Protection.html

[David Hess]

A crowbar circuit works by shorting the voltage to ground and relying on either blowing a fuse or a separate current limiting circuit.  A TVS as its name implies is intended for "transient" conditions; running a TVS as a shunt voltage regulator at high power levels is likely to short the TVS making it operate as a single use crowbar which may or may not be the intention of the circuit designer.

What I normally do is add a series switch at the input to the regulator and a circuit which shuts it off if the input voltage rises out of range.  This might be part of a cascode circuit which would otherwise increase the maximum input voltage during normal operation.

[MobileWill]

A crowbar circuit works by shorting the voltage to ground and relying on either blowing a fuse or a separate current limiting circuit.  A TVS as its name implies is intended for "transient" conditions; running a TVS as a shunt voltage regulator at high power levels is likely to short the TVS making it operate as a single use crowbar which may or may not be the intention of the circuit designer.

What I normally do is add a series switch at the input to the regulator and a circuit which shuts it off if the input voltage rises out of range.  This might be part of a cascode circuit which would otherwise increase the maximum input voltage during normal operation.

That is what I thought. I assume I can't use the EN pin of the regulator? I would think that the higher voltage should be stopped before it reaches the regulator? Having a higher input min voltage would be a problem as in some situations it could be pushing it already.

I could try and find a regulator with a higher max voltage but even at low current that voltage has to go somewhere and I don't have the space for a heatsink.

[David Hess]

I assume I can't use the EN pin of the regulator? I would think that the higher voltage should be stopped before it reaches the regulator? Having a higher input min voltage would be a problem as in some situations it could be pushing it already.

The enable pin will shut down the regulator but the maximum input voltage specification will still apply.

I could try and find a regulator with a higher max voltage but even at low current that voltage has to go somewhere and I don't have the space for a heatsink.

Exactly, disconnecting the input may not only be because of the maximum input voltage to the regulator.  It may be done to protect against excessive power dissipation as well.

If you design the input switch as a cascode to the regulator, then it will extend the input voltage range and distribute the power dissipation between the regulator and input switch.  Then if excessive input voltage is detected, the cascode or regulator can be shut off protecting both from excessive power dissipation.

[MobileWill]

I assume I can't use the EN pin of the regulator? I would think that the higher voltage should be stopped before it reaches the regulator? Having a higher input min voltage would be a problem as in some situations it could be pushing it already.

The enable pin will shut down the regulator but the maximum input voltage specification will still apply.

I could try and find a regulator with a higher max voltage but even at low current that voltage has to go somewhere and I don't have the space for a heatsink.

Exactly, disconnecting the input may not only be because of the maximum input voltage to the regulator.  It may be done to protect against excessive power dissipation as well.

If you design the input switch as a cascode to the regulator, then it will extend the input voltage range and distribute the power dissipation between the regulator and input switch.  Then if excessive input voltage is detected, the cascode or regulator can be shut off protecting both from excessive power dissipation.

Would the input switch be like a zener connected to a mosfet that turns off when the zener hits breakdown voltage?


I found this regulator
https://www.digikey.com/product-detail/en/on-semiconductor/LP2950ACDT-3.3RG/LP2950ACDT-3.3RGOSCT-ND/2120661

I could use that and depend on the thermal shutdown? I usually try not to rely on the regulator thermal/current shutdown protection.

[Gyro]

A reasonably sized TVS or Zener in combination with a series PTC thermistor might give you the resetable level of protection that you need. Less aggressive than a crowbar and includes reverse polarity protection. Combine that with an automotive grade regulator which is more tolerant of input overvoltage and you should be well protected. Worst case, diodes tend to fail short circuit unless hit by a really big current spike (failsafe in your application).

[MobileWill]

A reasonably sized TVS or Zener in combination with a series PTC thermistor might give you the resetable level of protection that you need. Less aggressive than a crowbar and includes reverse polarity protection. Combine that with an automotive grade regulator which is more tolerant of input overvoltage and you should be well protected. Worst case, diodes tend to fail short circuit unless hit by a really big current spike (failsafe in your application).

I was planning to use a PTC anyway. I whipped up an example, how does this look.

[Siwastaja]

You worry too much, most ic regulators have built-in protection

Um, no, they don't. They blow up fairly easily on overvoltage, unless otherwise specified. You must be confusing this to output short circuit and thermal protections almost all regulators do employ.

[Gyro]

A reasonably sized TVS or Zener in combination with a series PTC thermistor might give you the resetable level of protection that you need. Less aggressive than a crowbar and includes reverse polarity protection. Combine that with an automotive grade regulator which is more tolerant of input overvoltage and you should be well protected. Worst case, diodes tend to fail short circuit unless hit by a really big current spike (failsafe in your application).

I was planning to use a PTC anyway. I whipped up an example, how does this look.

Yes that's what I was thinking. You'll want to size the PTC so that it trips as soon a possible but doesn't nuisance trip. You obviously need to put in whatever caps the regulator requires. Another protective feature to include is a diode from the regulator output to input so that the output capacitor(s) don't discharge through the regulator if the input gets shorted.

[David Hess]

In my experience PTCs are too slow in many applications.  By the time it heats up, the TVS may have already failed as a short so the PTC prevents further damage but the device is still dead.  Crowbars do not have to dissipate as much power as a TVS because of their low on-state voltage drop.

A couple years ago I rebuilt a 200 watt Kenwood HF transceiver power supply where this exact failure occurred.  The TVS which protected the HF power amplifier from over voltage conditions did its job ... by shorting out taking the power supply pass transistors and emitter ballast resistors with it.  The TVS failure was probably caused by the poor load regulation of the power supply which allowed the output voltage to spike whenever the power amplifier supply current dropped.  I suspect the Kenwood designers added the TVS for exactly that reason but they should have fixed the power supply which had several design mistakes.

Integrated regulators usually include safe operating area protection which will lower the maximum output current when the input to output voltage becomes too great but this does *not* normally protect the regulator from excessive input voltage.  Instead it protects the power transistor from secondary breakdown.

[MobileWill]

So it sounds like I should just use a higher max voltage regulator and get on with reverse voltage protection. I don't have enough voltage on the low end to add a in series diode unless I can find a diode with a low forward voltage drop. I will have a PTC on the power input which is better than nothing.


The other possible issue is dealing with two power sources at the same time. It won't happen often but it can. Currently it has been okay but only because both sources has been 5V. Going forward not so much. One could be 12V but at the same time there could be a 5V input. I need to source from either into the regulator.

I think I am going to run out of room for all this.

[David Hess]

Would the input switch be like a zener connected to a mosfet that turns off when the zener hits breakdown voltage?

There are lots of different ways to do it with a zener being one of them.  I have actually seen this done using a zener and SCR like you would find in a crowbar circuit except the SCR latched on turning off the transistor switch until input power was removed.

If you use some type of comparator on the input voltage, then hysteresis or something should be added to prevent oscillation.  The state should be on or off and not in between.  Sometimes these types of circuits will have a restart timer.

I could use that and depend on the thermal shutdown? I usually try not to rely on the regulator thermal/current shutdown protection.

The thermal protection in an integrated regulator is not really something which should be relied on during normal operation.

[Siwastaja]

This is not alchemy where using random components and hoping for the best is the only way  .

You just need to read some datasheets to pick a PTC with a known response curve, then pick a TVS which can handle said energy over said time. Sure, this solution may be expensive, huge TVSes are not cheap.

PTC should be thermally coupled to the TVS so that it also protects from continuous slight overvoltage causing excessive heating in the TVS, but not enough sustained current to blow the PTC open alone.

[Gyro]

You just need to read some datasheets to pick a PTC with a known response curve, then pick a TVS which can handle said energy over said time. Sure, this solution may be expensive, huge TVSes are not cheap.

Adding a low value series resistor would reduce the peak current before the PTC trips and allow a cheaper zener to be used. The OP is only talking about a <100mA current drain so normal drop would be minimal. Keeping the PTC, resistor and Zener in close thermal proximity (he wants compact) should make a simple 3 component reliable solution.