Polarity Protection MOSFET

[JFace]

I have read the excellent article on geofex regarding polarity protection using MOSFETs:
http://www.geofex.com/article_folders/mosswitch/mosswitch.htm

I have seen other articles describing the same design, except they leave out the resistor connecting the grid to ground, and connect the grid directly to ground. What is the purpose of this resistor, how does one decide what value to use, and is it safe to connect the grid directly to ground?

For reference, I have purchased a TO-251 P channel mosfet, FQU11P06TU.
Datasheet: http://www.farnell.com/datasheets/1783742.pdf


Using a resistor or a short works for this particular mosfet on the breadboard.

[R.G.]

I have seen other articles describing the same design, except they leave out the resistor connecting the grid to ground, and connect the grid directly to ground. What is the purpose of this resistor, how does one decide what value to use, and is it safe to connect the grid directly to ground?

The purpose of the resistor is that I get this itchy feeling between my shoulder blades at connecting a control lead directly to ground or power supply, and it's is a current limiter for a part that is not shown in that article, now that I go back and check my own work... 

It is OK to use a resistor of zero ohms - that is, a direct connection to ground.  The circuit works. But it bothers me that the MOSFET gate is now open to damage from transients on the power supply, and things like some junior genius thinking that if 9V sounds good on a power supply, 18V would be better and 27 volts would sound TERRIIFIC!! 27 V would of course kill the MOSFET, whose gate is rated for +/- 20V from the source.

The gate of the MOSFET is a piece of glass 20 volts thick. There is no commercial resistor you could put there that cause much of a difference in operation other than slowing down the MOSFET switching on and off. I would not hesitate to use anything between 100K and 10M.

The part that is missing is a 10V -15V zener with cathode to the gate and anode to the source of the MOSFET. This ensures protection from transients, including the junior genius with the adjustable power supply or stacked batteries.

Using a resistor or a short works for this particular mosfet on the breadboard.

It should work for any similar MOSFET. The resistance of a MOSFET gate is absurdly large, so the value of a resistor in series with it changes the operation of the MOSFET only in that the voltage on the gate is slowed down by the resistor charging and discharging the gate capacitance. With any reasonable resistor, this is unnoticeable. However, with a protection zener in, the resistor limits current through the zener to a safe value. This, too has a wide band of "works OK".

[JFace]

Thanks for the clarification, that helps. I wanted a solution that had the same parts count and footprint of a conventional diode, and adding a resistor doubles that. 

It is OK to use a resistor of zero ohms - that is, a direct connection to ground.  The circuit works. But it bothers me that the MOSFET gate is now open to damage from transients on the power supply, and things like some junior genius thinking that if 9V sounds good on a power supply, 18V would be better and 27 volts would sound TERRIIFIC!! 27 V would of course kill the MOSFET, whose gate is rated for +/- 20V from the source.

I thought of this when picking out this particular MOSFET. The gate to source max voltage is 30V. If the user puts in more than 30VDC, he has to pay the stupid tax.

[rankot]

The part that is missing is a 10V -15V zener with cathode to the gate and anode to the source of the MOSFET. This ensures protection from transients, including the junior genius with the adjustable power supply or stacked batteries.

I used this MOSFET protection for my circuits, but it seems that MOSFETs are very fragile when experimenting, since shortcuts in circuits easily destroy them. Is there any way to protect MOSFETs from that kind of damage? I didn't use that zener diode, just to mention.

[Rob Strand]

but it seems that MOSFETs are very fragile when experimenting, since shortcuts in circuits easily destroy them.

Protecting against shorts isn't a requirement really!  It's something that
needs to be handled outside the box:
- The best solution is to have a current limit on your power supply.
- The easiest solution is not load the MOSFET until you have finished!
- An intermediate solution is add a 10 to 100 ohm resistor in series with the power rail.
  Higher the better but too high on some circuits will cause the DC voltage to drop and lead
  to other weird behaviour.
- Build a current limiter circuit for prototyping.

It is possible to add a transistor and a resistor to the MOSFET circuit to turn it into a current limit.
Along the lines of Q2 and Rs in this circuit (R2 is like the existing resistor):
http://i.stack.imgur.com/rZL3F.png

Here's this one:
http://m.eet.com/media/1100963/fig2a.jpg

R1 = 0.7 / I_limit ; you might find it turns out to be 0.55/I_limit due to the large existing R2 value.
For I_limit = 100mA, R1 = 7 ohms.
Circuit will add a voltage drop of R * I_load

[PRR]

"Shorts" would not be a problem if you didn't have shorts.....

Or today you can switch from the dinky MOSFET to a fat one. 30Amps or more for a few bucks. Then you are more likely to blow-up your power supply than the MOSFET when shorts happen. I don't see why it would not work the same.

IPD220N06L3 $0.52
PSMN022-30PL,127 $0.57
IRFB7546PBF $0.84
BUK758R3-40E,127 $1.04

[rankot]

Then why complicate and not protect like this:



This 1N5819 (or any other from 1N5817 till 1N5822) diode have really small Vfd (0,2V) and this is safe in case of shortcuts. I use LM317 based regulated power supply for my pedals, but sometimes MOSFETs burn if I make shortcut in a circuit during test phase, although PS has built in overcurrent protection.

[Rob Strand]

Then why complicate and not protect like this:

There's nothing wrong with PTC's.

Using the zener like that probably isn't so great.  Zener voltages aren't precise it will probably flatten the battery and possibly trip-put the PTC in normal use.   A higher voltage zener will prevent bleeding current in normal use.   For over voltage the issue still remains that you might blow the zener before the PTC trips.    Better than nothing just the same.

The original ckt opens the connection and avoids any angry behaviour.

How Boss did the PSU and battery protection worked well it had resistance between the AC adaptor and the ckt but not the battery.   For effects pedals I'm happy with a reverse diode.   For medical products I'd be using something more elaborate.

although PS has built in overcurrent protection.

That protection protects the power supply itself from cooking when you have a short circuit but it doesn't necessarily protect the circuit you are testing.   Professional power supplies have a variable current limit that lets you set the current and protect the *circuit* you are debugging/testing:
http://www.bellnw.com/media/images/products/Agilent/E3610A.jpg

In the old days the power supplies sometimes had a switch which let you set the current limit to high or low eg 0.1A /1.0A.  It doesn't have the control of a current limit pot but it does work well.

You could for example put a 150mA PTC in your power supply and have a high/low current limit switch that bypassed the PTC for high current.  That's not how commercial supplies would do it but it would work.

Protection required for a project *in use* is different from protection require when developing the project ie. where you stuff things up.  Internal protection of a product usually covers faults and maybe some forms of foreseeable abuse.  It doesn't have to cope with slipping with the multimeter probe and shorting out the $50 IC.  They are different things. 

Interestingly I worked on a project where a certain chip kept blowing up.  I couldn't see an obvious design fault.  The problem stopped when people the number of people probing the boards went down.

[rankot]

OK, I have added zener diode to original circuit mentioned in the first post, so I presume it shall look like this:



Does it protect from overvoltage?

[bool]

I don't understand the fuss.

What is wrong with a schottky diode?

In all practical scenarios, you can use two small schottkys in parallel (f.e. some common-cathode duals in smt SOT23 package) in series with some 10-ish ohm resistance (fe. 4x 47ohm in 0805 package paralleled) and be perfectly safe (with a small-ish footprint and simple trace pattern).

Acceptable and reliable from design-pov and from manufacturing pov.

(Although theoretically nad practically, paralleling two semiconductor junctions is a very bad design practice, in such small-power circuit, the above "just works").