Over voltage protection

[kcs]

Hello,
I am designing a project which will require 20Vdc (max 20A). I want to make a continuous over voltage protection that could withstand max of 250V. For example, if someone accidentally plugs 250V in instead of 20V, some clever circuit (or IC) should disconnect the system. Can anyone suggest me how to implement this?

P.S. I don’t want to use relays in my design. And I would prefer not to use any fuses.

[Psi]

Not using any fuses makes it very tricky. (i'm assuming poly fuses are also out)

If you try and "detect" 230V then 230v is already on your circuit damaged something.
You need the 230V to never get onto the system in the first place.

If it was less current at 20V you could linear regulate 230v to 20V for a nanosecond and detect the huge current flow to trigger the disconnection. But since you need 20A at 20V it would require burning off 4kW to get 20V at 20A, which is stupid.

I think we need more info about your project and why you don't want a fuse.

[kcs]

I am not strictly against fuses. I just prefer not to use them. Poly fuse is fine, I already use those in my design.

I understand that timing is critical here. I believe that is where TVS could help.

I was exploring crowbar technique, but it appears to work only when there is a small over voltage.

[jahonen]

One possible alternative technique is to put a suitable FET switch (which can withstand the foreseeable overvoltage) which disconnects your downstream electronics in case the input voltage becomes too high.

Regards,
Janne

[ptricks]

Crowbar circuit with a fuse is probably going to have the fastest response time. I have never used one with a polyfuse though.
The problem you are going to have with anything non fuse related is if someone connects 250V you either have to shut it off or dissipate the power.
Dissipating the power would require something heat sinked probably.  Shutting off power means a relay and associated parts.

If it were me I would use a regular fuse rated over what the device would normally use and connect that to a crowbar circuit, this would not be the primary protection for the device, just the over voltage protection. Use a polyfuse in the same path for normal operation. Normal operation wouldn't blow the crowbar connected fuse, but a serious overvoltage situation would.

[Short Circuit]

9-500V surge stopper. works with external mosfet
http://www.linear.com/product/LTC4366

[PSR B1257]

if someone accidentally plugs 250V in instead of 20V

250V AC, DC or even both?

[kcs]

One possible alternative technique is to put a suitable FET switch (which can withstand the foreseeable overvoltage) which disconnects your downstream electronics in case the input voltage becomes too high.

That would be an ideal solution. I was thinking about that but couldn't come up with schematic or find it on the internet.

Crowbar circuit with a fuse is probably going to have the fastest response time.

But how can it withstand large voltage inputs? Gate of SCR will be exposed to very large voltage thus SCR will be damaged.

The problem you are going to have with anything non fuse related is if someone connects 250V you either have to shut it off or dissipate the power.

I would prefer to shut it off using a mosfet.

9-500V surge stopper. works with external mosfet
http://www.linear.com/product/LTC4366

It is a Surge Stopper thus it is not designed to withstand continuous over voltage. Correct me if I am wrong.

250V AC, DC or even both?

DC.

[kg4arn]

Not sure if this might be too simple for your application.  The current levels are far lower than 20A.  But you may be able to adapt the principle.

I had to do something similar once to protect some RF amps from over voltage on the bench.  I opted for a series switch because I needed the flexibility to power the amps at multiple voltages (thus a regulator was impractical).  A comparator simply watches the input voltage and switches the pass transistor off at 16V.

Here is my schematic and an oscilloscope tracing of an over voltage condition.  The blue trace is the power supply turning on.  The yellow trace is the voltage passed to the devices down stream.  Keep in mind that this was built as a "one off" with the parts that I had available.

Also, at the time, I found a National Semiconductor App note with a very similar circuit.  It uses a PFET as the pass transistor.

http://www.ti.com/lit/an/snva190a/snva190a.pdf

[T4P]

Well there's a zener there isn't it? I'm afraid you got it wrong 

And crowbars are just magnificent, no other method but crowbars

[Short Circuit]

...

9-500V surge stopper. works with external mosfet
http://www.linear.com/product/LTC4366

It is a Surge Stopper thus it is not designed to withstand continuous over voltage. Correct me if I am wrong.
...

These device has an adjustable timer, which is used to time the maximum duration of the surge dissipation. After that, it turns off.

The LTC®4366 surge stopper protects loads from high
voltage transients. By controlling the gate of an external
N-channel MOSFET, the LTC4366 regulates the output
during an overvoltage transient. The load may remain
operational while the overvoltage is dropped across the
MOSFET. Placing a resistor in the return line isolates the
LTC4366 and allows it to float up with the supply; therefore,
the upper limit on the output voltage depends only on the
availability of high valued resistors and MOSFET ratings.

An adjustable overvoltage timer prevents MOSFET damage
during the surge while an additional 9 second timer
provides for MOSFET cool down. A shutdown pin reduces
the quiescent current to less than 14?A during shutdown.
After a fault the LTC4366-1 latches off while the LTC4366-2
will auto-retry.

[AndyC_772]

if someone accidentally plugs 250V in instead of 20V, some clever circuit (or IC) should disconnect the system. Can anyone suggest me how to implement this?

Am I alone in thinking that if someone manages to accidentally plug 250V into a 20V circuit, that they fully deserve a loud bang, a certain amount of smoke and a big bill as a reminder to be more careful?

I'm finding it hard to see how anyone could make this mistake by accident. 20V is what we get from a laboratory supply, or maybe a battery pack. It's hard to confuse with the mains.

Personally I normally use a combination of a transient voltage suppressor diode plus a resettable fuse of some kind to protect circuits from spikes and surges, but it's asking a lot to protect a 20V circuit from 250V.

[FenderBender]

Yeah I was just going to suggest using something like an LM431 and an SCR or something along those lines.

[SeanB]

I saw an acess system that was killed by the PSU. 12V supply that had the OHL caps die, and it put out 65V open circuit..... Supply was repaired, but the hotel had to buy a new card writer, at a big price. I did suggest adding a big 18V transorb after the fuse in the unit, don't know if they did so.

[Flunze]

You can use a Relay, to connect power to the circuit. But the Relay only connects, if you apply the correct Voltage. Pretty simple way to do that is to use a Microcontroler or so. The simple plan:

You need some resistors, to split up the voltage. You ever heared of 20 Led Voltage Meter? You can use this circuit, changed just a bit, to detect, whether you exceed the voltage or not. 1 Led goes on means, you only drive 25 Volts into the resistors for example. 2 Led goes on means, you exceed the Voltage a bit, so do not allow current to flow. This is VERY simple and works for 100%. I will give you a simple circuit for the LED Meter, you can sure do something with it and change it for you to match perfect.

>Clicking right here<

Just remember: Led 2 turns on, is a signal of 1, so if you want to use a Relay, you have 2 ways: Use one, that can switch between on1 and on2 or use a NOT-Gate, to turn the Relay on, if power is off.

The relay is inductive load, so make sure, you won't damage triggering components by overvoltage. Also, it has a delay, so it will not work, if you get a Squarewave HV Spike, cause it CANNOT trigger so fast. Than the Relay becomes useless. Maybe a HV and High Speed MOSFET etc. can do it better and react fast enough. But don't ask me about MOSFET's.

(Together with a current limitting resistor on the inpot and a Capacitor, you can smooth out the spike a bit and it propably won't be so bad for the circuit - this is, how I fixed the triggering spike of my 50$ Microphone - Resistor+Capacitor over the power in and it does not spike anymore.)

This are enough ideas for you, I think. Right?

[kcs]

I have received many suggestions. I will try to evaluate each one.
At the moment I can share my bad (not working; tried with 3 professional simulation software) schematic I am working on. What it supposed to do:
[Works]:
S1 - reverse polarity imitator
Q1, D2, R1 - reverse polarity protection

[Doesn't work]:
R2, R3 - voltage divider (output should be 0-25V; for simplicity I assume there is no voltage drop caused by D3 and D4), Q3 should switch on when 2V or higher voltage is connected and turn off Q4.

[Short Circuit]

Q4 probably turns off, but then there's the body diode, which continues to pass current.

Also, beyond 200V, the gate of Q3 will be destroyed, and the switching threshold (Vgs) is rather coarse.

[ToBeFrank]

I simulated the following app note in LTSpice, and it worked. Perhaps you can modify it for the higher current you need?

http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=snva190a&fileType=pdf

[jahonen]

How about something like this? Of course, you'll need pretty stiff fets. Also note that component values may not be optimal, or even sensible, so check everything before use.

Regards,
Janne

[T4P]

Where are you getting the 20ADC+ relay from? What happens to it when someone inadvertently applies >200VDC to the circuit?

That should teach the person something.
After all a 20A+ relay is not hard to find but they only come in AC or DC
But it's not cheap,
http://futurlec.com/Relays/SSRDC200V40A.shtml 27.90
And if it's infinite protection
http://futurlec.com/Relays/HF92B-120.shtml this heatsink required < 6.90$
And a SSR perspex cover to finish up everything 
http://futurlec.com/Relays/SSR_COVER.shtml

[Flunze]

What about my idea? Something like that woule be realy easy to do and should not cost so much.