Hi,

I have been looking around for a simple 5V USB overvoltage protection circuit. I dont need anything special, I want it to be cheap and simple. The internet is full of circuits with a zener diode and P mosfet. I understand the idea behind this circuit but... What if the overvoltage event is NOT a mili-microsecond pulse, but lets say it stays as long as the circuit is being powered?

I my understanding of how the zener diode works - zener diode is frankly an open circuit until the zener voltage (with respect to diode's anode) is applied. Then the zener diode starts to conduct in a reverse polarity, so far so good. But the resistor is also needed to limit the current not to fry the diode, absolutely logical... However, does the zener diode maintain a constant voltage (zener voltage) drop across itself when opened, or it just settles at around 0.6-0.7V of voltage drop? If the diode maintains only the usual PN junction voltage, all the voltage would drop across the resistor which would result in unwanted MOSFET operation...

In other words, does this circuit work for sure if I apply anything above 6V and keep it for considerable period? (10s of seconds lets say) Its my first time doing anything with zeners, so all the observations are welcome. Thank you in advance.

 

It "sounds-like" You need a "Low-Drop-Out" Voltage-Regulator, ( LDO ).

How high do You expect your Input-Voltage to go ?
.
.
.

 

It "sounds-like" You need a "Low-Drop-Out" Voltage-Regulator, ( LDO ).

How high do You expect your Input-Voltage to go ?
.
.
.

Well, I guess no USB charger can possibly get higher than lets say 20 V, but how about the zeners? Is my assumption true or the circuit I have provided you with is for TVS protection?

 

What is a "USB-Charger" ?
You need to be more specific.
What problem are You trying to solve ?
.
.
.

 

does the zener diode maintain a constant voltage (zener voltage) drop across itself when opened

When the Zener is conducting (open to me means not conducting) in the reverse direction it maintains relatively constant at the Zener voltage.

The circuit you posted will not limit the output voltage.
I'm not sure what its purpose is.

Another approach to you problem is to use an SCR crowbar circuit that clamps the output to a low volage if the maximum is exceeded.

 

When the Zener is conducting (open to me means not conducting) in the reverse direction it maintains relatively constant at the Zener voltage.

The circuit you posted will not limit the output voltage.
I'm not sure what its purpose is.

Another approach to you problem is to use an SCR crowbar circuit that clamps the output to a low volage if the maximum is exceeded.

What about tunel diode then? What

When the Zener is conducting (open to me means not conducting) in the reverse direction it maintains relatively constant at the Zener voltage.

The circuit you posted will not limit the output voltage.
I'm not sure what its purpose is.

Another approach to you problem is to use an SCR crowbar circuit that clamps the output to a low volage if the maximum is exceeded.

I had really unpleasant experience with SCR's (leakages etc...) Do you have a trusty ready to use circuit? My allowed Vmax is 5.5V

 

As @crutschow said a crowbar circuit will work , here is an example, chose the Zener voltage to suit.

 

As @crutschow said a crowbar circuit will work , here is an example, chose the Zener voltage to suit.View attachment 292161

What are the requirements for ZD? (leakage current, capacitance, recovery time etc...)
What SCR should I choose? What are the parameters? (Like I said, I have been playing with those on the breadboard, that was some terrible experience in terms of stability)
What kind of diode (what parameters) is SD1?

I would love to use this circuit, but this time I want to know that it will work 100% the first time I test it...

 

I would love to use this circuit, but this time I want to know that it will work 100% the first time I test it...

Then I suggest you run it in a simulation before building it

 

Then I suggest you run it in a simulation before building it

Then I suggest you run it in a simulation before building it

However, this circuit will require a complete power-off to restore the state... Is there any solution that the circuit would turn on by itself when the overvoltage condition is removed?

 

How do You know that You need a "Crow-Bar"Circuit ????
Or any kind of Over-Voltage-Protection ????
.
.
.

 

What are the requirements for ZD? (leakage current, capacitance, recovery time etc...)
What SCR should I choose? What are the parameters? (Like I said, I have been playing with those on the breadboard, that was some terrible experience in terms of stability)
What kind of diode (what parameters) is SD1?

I would love to use this circuit, but this time I want to know that it will work 100% the first time I test it...

Explain why you need an overvoltage protection, maybe there is another way to achieve what you need?

 

Explain why you need an overvoltage protection, maybe there is another way to achieve what you need?

I am designing a USB isolator with external USB supply. One port USB isolator would have 2 inputs: 1 for the main USB (data and power) and another USB port would be used if greater power levels at the output are required (second USB port would have only the VCC and GND pins connected to the PCB.)

It is very likely, that the one using isolator with both inputs connected would connect the ,,power USB" from the wall adapter. 5V 1A wall adapters are almost gone, now they are designed to provide fast charging at 9V, 18V etc...

So the idea behind OVP is that in case of incompatible wall USB adapter is connected to the power USB port, I would not fry my ISOUSB111 chip (I am using that as main isolator chip.) Of course I am not expecting the chip to output 2 A - I have transformers, transformer drivers and the LDO at the output, dont worry about that, my goal is to protect the input.

 

A simulation cannot prove a circuit will work 100% of the time, a simulation can't even prove a circuit will work at any time.

A simulation is a great help to design work, but nothing can replace an actual circuit.

If you want a circuit that will reset after the overvoltage automatically, you are probably moving beyond the "simple" premise stated in the OP.

 

Is there any solution that the circuit would turn on by itself when the overvoltage condition is removed?

Not without adding some additional sensing circuitry.
The current must be completely removed from the SCR to reset it.

Below is an LTspice simulation of high voltage cutoff circuit with the TL431 programmable voltage reference used as a comparator:
The circuit cuts off transistor M1 when the voltage is too high and turns it back on when the voltage drops.

The TL431 cathode goes low when its Ref input is above 2.5V, as determined by the R2, R6, and U2 voltage divider, which shuts off the two transistors.
A pot is used to adjust the cutoff voltage to the exact value you want, to compensate for the various component tolerances, which otherwise could put you outside the desired limits.

With nominal values and pot U2 at the 50% setting, the output (yellow trace) shuts off when the input (green trace) goes above an indicated 5.4V.

The M1 P-MOSFET must be a logic-level type [max Vgs(th) ≤2V] that has an on-resistance low enough to not significantly affect the voltage for the maximum load current.

 

A simulation cannot prove a circuit will work 100% of the time, a simulation can't even prove a circuit will work at any time.

A simulation is a great help to design work, but nothing can replace an actual circuit.

True. A circuit breadboard must always be built to prove that it works.

But a simulation often shows a design error where a circuit will not work.

 

Yup, I would never go back to design work without a sim.

Unless I can't get the damn models I need.

 

I used a sim a few days ago to check on something I pointed out a few years ago on this board, and was told I was wrong, turns out I was correct, but I didn't have the sim at the time.

Or the component to BB.

 

Yup, I would never go back to design work without a sim.

Unless I can't get the damn models I need.

hmm
What if I put just a single zener across the load? Something like a 10 W one? I guess thats the easiest and the cheapest way...

 

hmm
What if I put just a single zener across the load? Something like a 10 W one? I guess thats the easiest and the cheapest way...

If you can find one with the needed voltage and tolerance.