loading

How to protect OpAmp output?

Hi,

I have an OpAmp which controlls the Gate of a MOSFET.
When the MOSFET dies, sometimes the voltage it's running on (50V) somehow gets to stand over the OpAmp's output which destroys it.

What's the best way to protect the OpAmp from the high voltages that get to stand over the output?

Will a zener do the trick?
Trigger transformer won't be easy to use here because It's a PWM controller so I need nice block waves.
Diode on the output? but won't this be bad for delivering good current to the Gate?
...
Any ideas?

sort by: active | newest | oldest
gmoon5 years ago
Have you added the snubbing diodes per Steve and Iceng's suggestion? I would try that first.

You could also add an additional "driver" for the MOSFET, like Steve said. That could be as simple as a single transistor switching a higher voltage than the opamp... (which would invert the signal, but you know how to counteract that, if it's important). This would isolate the opamp from the MOSFET (and adding some series resistance between the opamp and the trans would be very practical and probably called for).

But that's not really the problem, is it? :-)
gmoon5 years ago
I would defer to Steve, as he's very knowledgeable about this stuff...

Still, just a few points--what's the switching speed? On the order of MHz? If it's low-kilohertz, I doubt the circuit would be hampered by a gate resistor (to a point).

But adding gate resistors won't help if the snubbing diodes in the circuit aren't correct, and that could be the root of your problem.
Put the damned diode in the right place, and the right way round, and stop blowing your devices in the first place. Sheesh.

;-)

Using a proper gate driver is the best way of course.
You're missing the point - the discharge path is the SAME as the charging path, only backwards. The opamp will SINK instead of SOURCE current.

Steve
NO, I mean the current path to the gate must have a very low resistance, period. You ain't going to damage the gate with an opamp
So build a signal condtioner on the front - take a look at a "real" scope front end - and copy that.

Mine "only" goes to 50V

Steve
NOW you're getting there.

And no, you need a VERY SMALL resistance to discharge the gate !

Do you understand now ? Its one of the reasons all these damned silly plasma speaker circuits are so crap. Only one of the reasons though.

Driving Mosfets is NOT trivial, and becomes increasingly harder, the faster you try. That's why a pro uses a special gate driver device.

Steve
Inductive kickback puts a HIGHER voltage on the top of the Mosfet, not a REVERSED voltage, so the diode doesn't help you at all.
iceng5 years ago
Sorry, got the impression the op-amp was expiring.

Mosfets can turn current off SO Fast ( if U let them ) that a 6" length of 
wire has enough inductance kickback to over volt the fet drain-source.

However a plain op-amp is usually to slow, it would take something
like ECL logic to cause that kind of current interrupt harm.

So inverse current can blow the fast intrinsic diode, that is why us old
inverter types will put a heavier diode to take over the inverse current
the fast diode has started.

Once the mosfet is inverse protected only current generated heating or
brake-over voltage will damage the fet, that is why I don't see the
reason for the series diode.

The diode across the inductor will prevent high voltage but slow the
inductive current fall.   If you don't want the diode then get a high Vp
fet and a 1N4007 diode.

A
fetsprob.jpg
+10
Nope. Look at the specification for a typical MOSFET device- I've chosen the 540 for the hell of it, it doesn't affect my argument.

http://www.irf.com/product-info/datasheets/data/irf540n.pdf

Look at page 2, and get back to me.

Steve
The device is, I suppose, "voltage controlled", but ask yourself, what is the gate, and what does it look like to my circuit....... ?

The school was screwing around. You can buy a new Rigel scope on ebay for a lot less than that. For this kind of work, a scope is vital, not just handy !!

I bought a high speed, high perfrormance PC based "PicoScope" for only about 400 E new a couple of weeks ago.

Steve
Sorry, I assumed it was a single opamp. Whatever pin you use for the output to drive the gate then, 1 or 7.

I think I've told you before that an opamp to drive a gate is a bad design choice anyway. They really don't have the output drive capability you need - ideally several AMPS at high frequencies. Use a scope to look at the gate voltage sometime, I think you'll be pretty surprised.

The slewrate of the TL072 is only 13V/uSec. I'd look for something of the order of 100. It can only source about 10mA.
gmoon5 years ago
Have you tried a current-limiting resistor between the Op amp and the gate?

Most MOSFETs have pretty high input impedance, and since they are "voltage in" devices, rather than "current in", a decent-sized gate resistor shouldn't hurt. Although Inductive spikes can be HUGE, way higher than the 50V supply...
iceng5 years ago
Did you not know that a mosfet has an intrinsic inverse high speed diode
as a "you understood" part of the fet structure ?

A
NFETtransistor.jpg
Put a lowish resistance diode from the opamp pin 6 to a diode clamp from the other end of the resistor to the op-amp Vcc and the same end to 0V. Add another resistor from there to the gate - as you know, you really need to keep these resistors small

What's killing your MOSFET ?

Steve
What are you driving ? If its a switched load, then using a proper gate driver is the way to go, if its a linear load, then you could isolate it with a fast opto-isolator

Show a circuit of your op-amp/gate connection.

Steve