# Problem with overvoltage protection circuits limits

I have a circuit which input voltage shouldn't exceed 50 V, I can't use neither of available overvoltage protection circuits because of the limits each have;

1. Crowbar circuit. While this is a good approach, the fuse in the circuit is 10 A. What if overvoltage condition occurred but the source supply can't provide 10 A to blown the fuse? For example if the input voltage is 60 V and supply can only provide a maximum current of 5 A, that's 300 W on the SCR and fuse is not going to open.

2. P-channel MOSFET based overvoltage protection. This circuits working condition is depended on the gate voltage and is not a sharp cut off like crowbar circuit. assuming the mosfet have 50 mΩ RDSon with 10 A over it, that's 5 W of power loss, not to mention the price on low RDSon p-channel MOSFETs.

• How do I protect the input from overvoltage condition and overcome the problems described? is there a better approach?

Input source can be a battery or power supply of any kind (DC), input voltage shouldn't exceed 50 V and maximum current is limited to 10 A both by a fuse and design.

• Although it also uses a PMOS, a possible solution has been partially addressed in another question (electronics.stackexchange.com/questions/538839/…), where a zener + bipolar are used to control the gate voltage of a PFET Dec 26, 2020 at 10:43
• @vtolentino P-channel mosfet is not a solution for the reasons I mentioned in the question. Dec 26, 2020 at 11:03
• Why would you use a 10 A fuse to protect a power supply that can't deliver 10 A? Dec 26, 2020 at 11:22
• @Transistor It's a dummy load, the supply can be any sort of PSU with different output capability. what defined is the load limits. Dec 26, 2020 at 11:34
• You misunderstand the operation of a crowbar. If the 60V supply can only supply 5A, that is NOT 300W across the thyristor. It's 2V or5V or so across the thyristor for about 10-25W. The remainder of that 300W is the PSU's problem... Dec 26, 2020 at 13:58

for example if the input voltage is 60 V and supply can only provide a maximum current of 5 A, that's 300 W on the SCR and fuse is not going to open.

That is simultaneously impossible; you can't have 60 volts across a crowbar if the crowbar has activated because the crowbar will clamp the voltage to 1 or 2 volts. So, at 5 amps that might be 5 or 10 watts but, it's more likely that it'll be much more current and, the fuse will blow.

How do I protect the input from overvoltage condition and overcome the problems described? is there a better approach?

Just simulate it and you'll see that the crowbar circuit has to clamp irrespective of what series resistance the incoming supply has.

From a comment: -

any sort of PSU can be connected to it. maybe it's a 60 V / 3 A PSU, maybe it's 12 V / 20 A PSU etc.

That puts a different light on things. This device is a 10 amp thermal fuse that operates at 73 °C. If it were thermally bonded to the triac/SCR it should offer some protection against sustained thermal damage on the triac/SCR. The two wires will be in series with the normal 10 amp fuse.

There are probably other variants that "operate" at higher than 73 °C and, maybe you might be able to replace the 10 amp fuse directly if you look into it more deeply.

• Does the triac fail short? I mean can I use it in sacrifice mode? Dec 26, 2020 at 11:36
• Off the top of my head I can't be sure - better to assume that this is not guaranteed one way or the other. In other words, you shouldn't rely on a component failing reliably in one way or the other (other than a fuse). Dec 26, 2020 at 11:38
• The basic problem isn't the MCU as far as I see it. Maybe you need some kind of fast disconnector circuit but, disconnecting several amps with the potential for big flyback voltages is problematic. I built one once for circa 100 mA that used MOSFETs to disconnect a 40 volt rated circuit when the supply rose above around 35 volts but your application is much higher current and this means different sorts of problems due to power supply connection loop inductance. Disconnecting rapidly can generate 1000s of volts hence, why I didn't suggest this path when u talked about it a month or so ago. Dec 26, 2020 at 12:21
• One more thought. Have you heard of thermal fuses? They should be able to handle the current you need but, will blow when they get too warm so, if you have one mounted on the triac it might work. Obviously, electrically its pins connect as per the normal fuse so it would be an extra fuse in series with the regular fuse. Dec 26, 2020 at 12:33
• @ElectronSurf Most commonly, if a semiconductor fails due to silicon overheating, it shorts out. If you fail to sufficiently cool the triac at 10W, it will most likely fail short. On the other hand, if you apply enough current to a semiconductor that the bond wires melt, it fails open. If you use a SCR that is rated for 15A continuous and 60A or more pulsed current, and your 10A fuse is fast-acting (type F, not type T), the fuse should blow before the bond wires melt, so failure is nearly certain to be "short circuit". No one is going to warrant it, though. Dec 28, 2020 at 8:26