# +-33V Overvoltage and reverse polarity protection

I found on this forum a schematic (Zener + MOSFET overvoltage protection). But It does not work and I can't figure out why. this is the schematic:

I want to protect the power supply of an opAmp form voltage above 33V (and symetric -33V). It's maximum votlage is +-45V. For some reason M3 is always conducting and doesn't deactivate when M1 is conducting.. Which should bring M3s gate to the same voltage than M3s source, thus deactivating M3. I know there are many other circuit possible do achieve what I want to do. But I'm wondering why this doesn't work because thoereticly it seems ok to me. For my academic knowledge ^^.

Thanks for your help.

• Exactly where did you find this 'answer'? – Bruce Abbott Mar 30 at 9:48
• @BruceAbbott I found it in the answers of this post : electronics.stackexchange.com/questions/121670/… – Neeko Mar 30 at 9:54
• What do you mean by "opening" and "closing" - can you change those words to "conducting" (or activating) and "deactivating" because it is then less ambiguous. A hydraulic valve "opens" to let water through but a MOSFET "closes" (like a switch closing) to let current pass. The terms need to be unambiguous. – Andy aka Mar 30 at 12:57
• Why do you need a such circuit at all? Isn't a resistor + 2 diodes enough? This circuit can alter the performance of the opamp , thus making it badly working. – Marko Buršič Mar 30 at 12:58
• @vangelo thats the second part on which I haven't worked yet. but the goal is to add an undervoltage lockout for +-10V (opamp spec). The goal would be to cutt both rails if one has a fault. And because I'm also a gamer, some LED to visualy show what fault is cutting the source. – Neeko Mar 30 at 14:13

I can't reproduce your problem, it works for me. The green graph is the voltage over the load with increasing input voltage.

And for the negative side with the corresponding n-channel (and rotated diode):

• Well I guess I have an issue with my LTspice then. Am I right to think that this will aslo work in -30/0V if i change the mosfets? – Neeko Mar 30 at 9:57
• Yes, if you also rotate the diode ;) – jusaca Mar 30 at 10:17

As @Akohlsmith stated in his answer, this circuit is only good for up to 20v or so, due to the MOSFET's gate voltage limit, $$\\text{V}_\text{GSS}\$$.

If you look at the FDS4559 Datasheet, page 1 states $$\\text{V}_\text{GSS}\text{ Gate-Source Voltage ±20V ±20V}\$$, which puts an absolute max limit of +/-20v to either gate with respect to source.

Even if Zeners are used elsewhere to limit gate voltage, a ripple or pulse could cause the gate to go beyond ±20V due to parasitic lead/trace inductance. Most MOSFETs are very sensitive to gate overvoltage - even if it only happens once, for a nanosecond, through a 10k resistor, the MOSFET can be instantly destroyed.

The MOSFET's $$\\text{C}_\text{iss}\$$ (intrinsic gate capacitance) actually slows fast voltage changes, and so is a boon to fast switching. Many MOSFETs are employed in fast switching roles, so large gate capacitance is undesired and even detrimental. The FDS4559 only has ~700pF, which is relatively low, making it good for moderately fast switching. This low amount of gate capacitance also makes it more susceptible to gate overshoot than others.

I've seen a bidirectional TVS used to clamp gate voltage to ±15V with reasonable success. It must be placed as close as possible to the gate-source pins to reduce lead/trace inductance. Even TVS have a response time (shorter pin/trace lengths are better.)

You could add Zeners in series with R1 and R2 to limit the gate voltage, but then the minimum voltage let through the circuit is increased (and there is parasitic capacitance and inductance added.)

• Thanks for your answer! I used the FDS4559_p only to see if the real behavior of a mosfet is diffrent from the base model of a mosfet in LTspice. I had that issue in mind and was thinking of a zener. But first I wanted to solve my problem of the circuit not working. I tried by clamping the gates and that didn't solved my problem... So I first wanted to present to you my first draft and see if there was an issue with my thinking. – Neeko Mar 30 at 13:49

I too have simulated your circuit without problems, I saw that the Gate Source Voltages were "out of bounds" and added a couple more zeners to limit those voltages.

As for using this as an input to an OPAmp is not a good idea in my opinion.

• Thanks for your input. Somone else already told me not to use mosfets with opamps. But and don't get why since ,in my appication, the mosfets will most off the time be either conducting or diactivated. – Neeko Mar 31 at 8:01
• Its not that mosfets that are incompatible with opamps, depending on what you want to achieve it may be the ideal component. But this circuit along with what you describe you want to do does not seem to be the correct solution. This circuit introduces quite a few non-linearities into what would be your signal chain. As power supply management, this circuit can be useful, but in a signal processing path I would not recommend it. – Pau Coma Ramirez Mar 31 at 9:07
• For simple connection/disconnection I would use a relay, or an analog mux. These are ofcourse slow response devices and if you need to protect from short pulses, I would add infront of that some TVS (Transient Voltage Suppression) diodes, zeners, MOV (Metal Oxide Varistors),,.... protective elements to absorb the extra voltage/energy untill the relay disconnects the signal. – Pau Coma Ramirez Mar 31 at 9:12
• Oh, i guess I didn't explain my self well. This circuit is meant to creat the power supply for the opAmp, not the signal that the OpAmp has to work with. – Neeko Mar 31 at 10:24
• for power supply, initially looks like this could work, but perhaps instead of disconnecting you may just want to limit the overvoltage to your max voltage? – Pau Coma Ramirez Apr 1 at 5:48