Input/Output protection with Polyfuse and Diode

[markus_b]

I'm working on failure-proofing a power supply (a couple of amps in a couple of amps out). I'd like it to survive abuse unscathed (overload, reverse polarity, etc). One interesting option is the combination of a diode and a polyfuse. The diode (Anode to ground) would essentially short-circuit a reverse connected power source and cause the polyfuse to trip.

I will have an input current of up to 6 amps and a corresponding polyfuse.



Polyfuse 30R600 data sheet: http://www.littelfuse.com/data/en/Data_Sheets/Littelfuse_30R.pdf

I'm now struggling to find a corresponding diode which is small, cheap and can survive/absorb the current surge until the polyfuse trips.

Anyone with pointers or hints for good diodes ?

[free_electron]

SK153 will hold 15 ampere. DO214-AB package so pretty small.

[SeanB]

The ones in PS PSU units, cheap and with good ratings. Bonus is that they are ultra cheap ( and free if you strip a PC psu for parts) and the low forward voltage ( or here reverse voltage) is low enough that the semiconductors in the equipment most likely will not be turned on and damaged by reverse current. Just use both halves in parallel, they are rated for 35V at up to 10A or more. Use the ones on the 12V rail.

[nctnico]

Just be aware that you should use a polyfuse which is rated for twice the expected current and that SMT polyfuses are likely to fail catastrophically when they trip.

[Zero999]

What's the maximum short circuit current of the power supply?

You may want to consider adding a low value resistor in series with the circuit to limit the short circuit current to a sane level but you need to account for the voltage drop.

[markus_b]

Good questions and suggestions !

SK153 will hold 15 ampere. DO214-AB package so pretty small.

I had a look. Slightly more expensive than I would like and not stocked by Mouser. Oh well, you can not have everything :-).

The ones in PS PSU units, cheap and with good ratings. Bonus is that they are ultra cheap ( and free if you strip a PC psu for parts) and the low forward voltage ( or here reverse voltage) is low enough that the semiconductors in the equipment most likely will not be turned on and damaged by reverse current. Just use both halves in parallel, they are rated for 35V at up to 10A or more. Use the ones on the 12V rail.

I want to put together a 'professional' design, so no scavenging. Most high current rectifiers come in DPAK or similar packages and I want something smaller, too. The DO214-AB package above looks good.

Just be aware that you should use a polyfuse which is rated for twice the expected current and that SMT polyfuses are likely to fail catastrophically when they trip.

My polyfuse will have a trip current of 12 amps. So there is some margin. It is not SMT because I could not find SMT polyfuses for larger than 1 amp current. I'm somewhat astonished by your statement about catastrophic failure, this defeats the entire point of a polyfuse !

What's the maximum short circuit current of the power supply?

The power supply itself will be electronically limited to 3 amps. I'm using a LT3083 who limits to 3.3 amps internally, too.

The problem is to protect against the outside. If a user wants to charge a car battery and reverses the polarity accidentally, this should have no effect.

[rohitdesa]

Just be aware that you should use a polyfuse which is rated for twice the expected current and that SMT polyfuses are likely to fail catastrophically when they trip.

Oh I didn't know that. Thanks! Is there someplace online where I can read up about polyfuse selection?

[markus_b]

Oh I didn't know that. Thanks! Is there someplace online where I can read up about polyfuse selection?

There is the Littlefuse 'Product Catalog and Design Guide' (http://www.littelfuse.com/data/en/Product_Catalogs/Littelfuse_POLYFUSE_PPTCs.pdf) who explains polyfuses. It contains some explanations.

Essentially a polyfuse is a device, that when it heats up, increases its resistance (PTC). If you have too much current it 'trips', you get a runaway situation of heating up and increasing resistance until it reaches a high resistance/high temperature equilibrium. After cooling down resistance comes down too to its normal, low level.

[nctnico]

Just be aware that you should use a polyfuse which is rated for twice the expected current and that SMT polyfuses are likely to fail catastrophically when they trip.

My polyfuse will have a trip current of 12 amps. So there is some margin. It is not SMT because I could not find SMT polyfuses for larger than 1 amp current. I'm somewhat astonished by your statement about catastrophic failure, this defeats the entire point of a polyfuse !

The power supply itself will be electronically limited to 3 amps. I'm using a LT3083 who limits to 3.3 amps internally, too.

Then I'd use a 6A polyfuse. But if reverse polarity protection is your aim use a series diode. A reverse connection shouldn't cause a lot of sparks. You could add a MOV for overvoltage protection. Automotive environments can be prone to high voltage spikes (google for load dumping).

[digsys]

I would differ slightly with the suggestions given. Use a TVS, Transorb, Zener instead (D4), that way you have BOTH
reverse protection and OV protection. Use a OP-IP Diode (D5) to pass any "missed" OV back to the Input safely.
S/Mount polys don't blow, not sure where that came from. BUT chosing the right value IS NOT simple -
For 6A protection, I'd use a 5A poly !!!! IF I want closer tolerance, I go with 4A !!! Why, check the curves -
A 6A poly (or any fuse, except special types) will take a day or so to blow at 6A, and SECONDS to blow at 12A (2X OC) !!!
At 3X OC (over-current), that drops rapidly to 20-100mS. IF you are running 6A continually into a 5A poly, it will be a
LITTLE warm, but perfectly FINE !! A 12A poly, as suggested, will be up at 25-30A before it acts properly like a fuse.
This is from a lot of experience.

[markus_b]

But if reverse polarity protection is your aim use a series diode. A reverse connection shouldn't cause a lot of sparks. You could add a MOV for overvoltage protection. Automotive environments can be prone to high voltage spikes (google for load dumping).

The power supply is designed to take a 12V-20V input voltage (car battery or random laptop power brick) and provide a regulated 0-30V 0-3 Amp output.

I'd like to avoid the use of a series diode because of the voltage loss. On the output side this will defeat precise regulation and on the input side I'll loose up to 1V/6W. On the input side I could use a MOSFET driven reverse polarity protection, but a PPTC is probably easier / cheaper.

I would differ slightly with the suggestions given. Use a TVS, Transorb, Zener instead (D4), that way you have BOTH reverse protection and OV protection. Use a OP-IP Diode (D5) to pass any "missed" OV back to the Input safely. S/Mount polys don't blow, not sure where that came from. BUT chosing the right value IS NOT simple - For 6A protection, I'd use a 5A poly !!!! IF I want closer tolerance, I go with 4A !!! Why, check the curves - A 6A poly (or any fuse, except special types) will take a day or so to blow at 6A, and SECONDS to blow at 12A (2X OC) !!! At 3X OC (over-current), that drops rapidly to 20-100mS. IF you are running 6A continually into a 5A poly, it will be a LITTLE warm, but perfectly FINE !! A 12A poly, as suggested, will be up at 25-30A before it acts properly like a fuse.
This is from a lot of experience.

I understand that the polyfuse needs a lot of power to blow fast, something like >3x the trip current. this is the root of my inquiry, because I need a diode, the can provide this trip-current to the poly without blowing out before the poly trips. I also want this diode to be low-cost.

I may want to change the design to use a MOSFET-driven reverse polarity protection on the input side and a PPTC on the output. The output is only 3 amps, so the PPTC can be smaller and I may find a 3 amp SMD version. For 6 amps I found only through-hole.

[digsys]

... For 6 amps I found only through-hole.

Polyfuses LOVE to be paralleled !! I do many designs that way. When one starts to become even
SLIGHTLY warm, the others take up the current, until equilibrium is reached. Also applies on a
sudden severe short, I test them that way.

[markus_b]

Polyfuses LOVE to be paralleled !! I do many designs that way. When one starts to become even SLIGHTLY warm, the others take up the current, until equilibrium is reached. Also applies on a sudden severe short, I test them that way.

I have not seen that argument in any data sheet, but it is very rational. Load-balancing is automatic. I also re-ran the parametric table (Mouser) and found there are 6 amp PPTCs, but not from Littelfuse and most are low voltages only (6-12V). But it looks that a single 6 amp 30V through-hole fuse is cheaper than two 3 amp SMD fuses. We'll see what I come up with.

Did you test PPTCs only or also in combination with reverse-polarity shorting diodes ?

I'm still afraid that in an extreme case the diode blows before the PPTC trips. This would render my protection circuit useless.

[digsys]

... Did you test PPTCs only or also in combination with reverse-polarity shorting diodes ?
I'm still afraid that in an extreme case the diode blows before the PPTC trips. This would render my protection circuit useless.

If choosing this method, you DO have to be CERTAIN that you can achieve 2X Poly rating !! Higher even better !! With a popular 1.5KE
device, no problem at all. !!
BUT, if the fault energy collapses to say 1.5-1.2X continuous, then the protection device will sit there and COOK !! and protect the Poly !!
So yes, you need to do the sums. On occasions, when I really have to use this method (motors etc), I put a Thermistor on the
Protection device, usually have one on the pass transistors, heatsinks etc anyway - as a master failsafe.
The 2nd method is a zener (or sensor) triggered FET etc to "blow the fuse", but that can also suffer from the same 2X 3X issues.
PLUS it had the added problem of false or inhibited firing due to RF / switching noise from the load (ie Inductors, motors etc).
The FET gate is a high gain amp after all.
The 3rd  method is "fast blow" or HRC type fuses, which can literally beat a semi (we use them in EVs), but they're expensive and
can literally blow if they think there's a short coming :-).
The 4th method I've used is a Saturated PNP Pass transistor (NO polyfuse), with the OV detect at the base, turning it into a current limiter
or latched off !! BUT you STILL have to worry about load being connected in reverse .. yadda yadda
So after years of trying ALL sorts of methods, I've settled with matching a poly with a transorb, and optional thermal sense.
If you're not going to sell 100s of these, or put them in critical locations, then just go with a fuse and Zener.

[T4P]

Um... crowbar protection circuit? with a Triac or SCR

[SeanB]

SCR can fail quietly as open circuit............

Still does not do well as reversec protection unless you use a triac, and then you will need a 25A device to have a good enough soa at 12V and 3A.

[Electr0nicus]

Hi

I came up with a quick circuit, that may help. This circuit will work if a precise cutoff in a overvoltage condition is not necessary. At least you can get within +/-1 V (Ube of the transistor and zener diode tolerances) of the desired cutoff voltage.

I'll give you a brief description what the circuit does.

First of all there is a zener clamp, built up with R2 and D1. If the input voltage is lower than the zener voltage, the voltage on the base of the PNP transistor Q2 equals the voltage on the emitter, so this transistor will be disabled. So the base of the PNP transistor Q1 will be pulled down to GND with R4. Because the voltage on the Emitter of Q1 is 0.7V larger as on its base the transistor will conduct and thus pulling down the gates of the MOSFETs M1 and M2.

If now a overvoltage occurs the zener clamp will hold the base voltage of Q1 at a constant voltage, during the voltage on the emitter increases. When the emitter- voltage is 0.7V higher than the base voltage Q2 will conduct and so puts the supply voltage- Uce(Q2) on the base of Q1 which will be disabled. So the gates of the MOSFETs get pulled to the supply voltage and the MOSFETs are disabled too.

If a reverse polarity condition occurs Q2 will not conduct, because now its Ube is positive.  The same applies for Q1. Because the difference between Gate and Source of M1 and M2 is nearly zero they will be in non conductive state and because they are back to back, there is no way, both body diodes can conduct bypassing the reverse polarity protection.

The circuit is not tested in real life, but at least the simulation looks promising. 

Advantages:

-> Fast response.
-> Nothing gets hot, due the fact that if a fault condition occurs the power is cut completely
-> reverse polarity protection and overvoltage protection in one circuit

Disadvantages:

-> more components needed
-> maximum voltage equals maximum Gate Source voltage of MOSFET so a TVS will be needed for higher voltage.

[Electr0nicus]

There is even a simpler solution of the abovementioned circuit.

[Electr0nicus]

Also you can easily add a Autopower off / Softbutton switch- on circuit. The circuit below gives priority to the safety function. If a overvoltage occurs you wouldn't be able to switch on the mosfets using the switch on/off circuit.

[markus_b]

I think I'll use the following arrangement:

1) On the input side I'll use a P-Channel MOSFET to protect against reverse polarity. Here I may have up to 6 Amps, so the trip current will get very large.



2) On the output side I'll use three 1.1 Amp PPTCs in parallel with a diode for reverse polarity protection. I think this is the best compromise.



I think this gives me good protection on both sides.

[Poe]

I think I'll use the following arrangement:

1) On the input side I'll use a P-Channel MOSFET to protect against reverse polarity. Here I may have up to 6 Amps, so the trip current will get very large.



2) On the output side I'll use three 1.1 Amp PPTCs in parallel with a diode for reverse polarity protection. I think this is the best compromise.



I think this gives me good protection on both sides.

I like those circuits Electr0nicus. 

Markus_b,
There's no over voltage protection on the input, right?  Not that there has to be, just making sure I'm not missing something.
Additionally the 'output' on the output circuit is on the right side, correct?

I'm using something similar to both of those on a project right now.

[markus_b]

There's no over voltage protection on the input, right?  Not that there has to be, just making sure I'm not missing something.
Additionally the 'output' on the output circuit is on the right side, correct?

I'm using something similar to both of those on a project right now.

Yes and yes.

You are right that I'm missing an over-voltage protection on the input side. But the circuit, although designed with 20V in mind, should survive until 35V.

And the output is on the right hand side, but there was some more stuff and I did not want to distract from the distract from the discussion by leaving it in.

The entire thing will be a power supply, inspired by Daves video series from 6 months ago. When it gets ready I'll publish it all.

[Electr0nicus]

Maybe you would use a zener diode with a higher voltage, at least over 10V. If you use a standard MOSFET (no logic level one) you only get the lowest RDS(on) if you have a GS- voltage greater than -10V. The greater the better. There is no need to clamp the GS- voltage to -8V, because nearly all PMOSFETs you can get have a maximum Vgs of -20V.

And of course your circuit has no overvoltage protection. If you use some kind of TVS and polyfuse, those components have to withstand the large currents flowing in case of a fault. This means the components have to be large and expensive. Also the impedance of the plugpack, wall adapter etc., which you are powering your device with, has to be taken into account. F.e. a car battery can deliver many 100A, so your protection must handle that without failing. 
Anyhow it's your decision what you need and what you are feel up to spend. My credo when designing stuff, is to avoid any large currents on the PCB, which can be difficult to control. At least my simple version of the overvoltage/ reverse polarity protection circuit will cost 3€ maximum. I will design it in in a current project of mine. There I'll have to deal with currents up to 12A continuous. I will build the circuit up this week and do some extensive testing. Because simulation and reality are two different worlds

Cheers Gregor

[bilko]

Looks like your p-channel mosfet pins D & S are reversed

[Electr0nicus]

Looks like your p-channel mosfet pins D & S are reversed

The MOSFET has to be built in that way. Otherwise no reverse polarity protection, because then the body diode will conduct.
In this case you have to connect the drain to the input voltage.
If the power supply is connected for the first few milliseconds current flows trough the forward biased body diode. The MOSFET is off at the beginning, because the source is connected to the load. The MOSFET only switches on when the voltage on the source rises above the Vgs threshold voltage. Then the mosfet is really low impedance, so almost all current flows through the drain, nothing through the body diode. The body diode can withstand at least the same current as the MOSFET itself. The only problem is the diode forward voltage drop, which creates massive power dissipation and heat up of the device. So if your load is shorted out and the voltage on the load cannot rise above the Vgs threshold voltage, then the mosfet will go up in smoke sooner or later.