mosfet failure reasons?

How's this link work for a shematic, seem correct to you? http://www.cadvision.com/blanchas/hexfet/npch-sw-sch.htm
im gonna use an irf9520 I think i and an irf 540
My plan to drive it is from a 555 timer, im gonna set up a dual supply opamp to take the 555 output from 0/12v to -12/+12,

If what you're trying to do is drive a speaker or similar load with a 5KHz square wave, you might be better off using bipolar transistors instead of mosfets. Use a cmos 555 (which drives the output right to the supply rails, unlike the bipolar version), and replace your N-channel source follower with an NPN small signal transistor driving a PNP power transistor (see Sziklai pair - Wikipedia but also add a resistor between the base and emitter of the PNP transistor). Then add an opposite pair (PNP small signal transistor driving NPN power transistor) to pull the output down to ground. This isn't quite as efficient as using a mosfet half h-bridge (because you will have maybe 1.5v drop across whichever power transistor is turned on), but it avoids the problems of shoot-through that you tend to get with home-made h-bridges.

If the power required in the load is very large, then the mosfet h-bridge is the most efficient solution; but in this case I recommend you buy a mosfet h-bridge driver chip. You won't need any p-channel mosfets because these driver chips are always designed so that both the upper and lower switches in each half h-bridge are n-channel.

I don't like that because it is possible to turn both FETs on at the same time.
You should look for a circuit that can't do that even by accident.

Yeah looking at that id rework it so it is both driven from the same wire, so +12 activats the nchannel, -12v will activate the pchannel
Anyone know any good examples?
And one of the things I plan on doing with this circuit is playing around with homeade higher frequency transformers, so ill probably want like 1-2 amps available

So basically this is how id wire it, should I add 10k resistors to ground for the gates?

winner10920:
How's this link work for a shematic, seem correct to you? http://www.cadvision.com/blanchas/hexfet/npch-sw-sch.htm
im gonna use an irf9520 I think i and an irf 540
My plan to drive it is from a 555 timer, im gonna set up a dual supply opamp to take the 555 output from 0/12v to -12/+12,

That schematic is for slow-speed switching rather than PWM - you need to reduce those 10k resistors between gate/source so the MOSFETs turn off fast - with 10k they are likely to spend 20 to 200us in turn-off, all the time in the linear region dissipating lots of power. For a 5kHz signal this clearly isn't sensible. Remember power MOSFET input capacitances can be around 10nF or so. To switch off in 1us you might need the gate-source resistors as low as 100 ohms (which would mean using 2W resistors!)

Proper MOSFET driver chips use active drive to the gate rather than one-sided in order to get good switching speed both on and off.

wouldn't the switch from +12 to -12 actively drive it?
just realized tho if I do it the way in that last picture it'll put too much voltage on the gate since the gate will be the opposite rail which is atleast 24v difference, being over the 20 vgs spec, or is my thinking missing something somewhere

you might need the gate-source resistors as low as 100 ohms (which would mean using 2W resistors!)

A 2W resistor to switch a 10nF capacitor, that sounds a bit excessive. It is not a continuous current rating we are working with.

just realized tho if I do it the way in that last picture it'll put too much voltage on the gate

Yes it will, your thinking is correct.

A voltage divider would work right? From the gate 1k to ground and 1k to my input should be good, unless I need something lower to make it faster?

Grumpy_Mike:

you might need the gate-source resistors as low as 100 ohms (which would mean using 2W resistors!)

A 2W resistor to switch a 10nF capacitor, that sounds a bit excessive. It is not a continuous current rating we are working with.

But if the switching time is otherwise 100us and you're sending a 4.5kHz square wave to it, the FETs will be in linear mode the whole time again and possibly dissipating a lot more than 2W. Certainly 10k gate-pull-down is not really compatible with PWM at kHz rates

BTW all we know so far is that the FETs are getting 4.5kHz and running off 12V - have no idea what the load is, would be nice to know.

It does seem to me that a 1/2 H-bridge MOSFET driver chip is going to solve these problems (do many of them support n+p channel though?)

winner10920:
A voltage divider would work right? From the gate 1k to ground and 1k to my input should be good, unless I need something lower to make it faster?

You don't need a divider if you have a single 12V supply - the motor switching circuit you reference is for split +/-12V supplies, hence the divider to the lower FET's gate. You do need to ensure both FETs cannot be on at the same time though.

The load is basically homemade high frequency transformers, its all just for playing around and maybe learning a bit
But im expecting like 1-3 amps draw but it will probably vary alot, its being fed by a +/-12v supply I made
id also like to be cheap and make this with what I got already, an irf9520 and an irf540
I've never worked with pchannel mosfets or negative voltages before so this part is new to me,
Also the input will be square wave and the frequency will probably be as high as 500khz max more likely in the 1k-200k range

Instead of a +/- 12v supply and mosfet half bridge, have you considered using a single 12v supply and a full bridge?

Making a home-made H-bridge capable of working at high frequencies is not easy. I think you would be better off getting a chip that does it all for you, such as http://www.farnell.com/datasheets/1383760.pdf.

It would definetly be easier but I like the challenge, and the lack of funds kinda supports to use what you got method

But so do u think the 1k divider will work? Where could I find the formula that could tell me my max switching speed?

The main constraint on the mosfet switching time is charging and discharging the gate. The mosfet datasheet usually gives a figure for total gate charge. This value varies with the drain voltage you are switching (the higher the voltage, the higher the total gate charge). To estimate the switching time, use the figure on the datasheet. The time taken to switch the gate is the total gate charge divided by gate drive current. The gate drive current is the excess gate drive voltage divided by the series resistor. The excess gate drive voltage depends on the mosfet and the driving voltage, but for a rough estimate you could assume one third of the gate drive voltage.

Example: 5V Arduino driving a STP40NF10L (total gate charge 46nC typical, when switching 80V @ 40A) through a 220 ohm series resistor:

t = 46nC * 220ohms / (5V / 3) = 6072ns

So you can see that for fast switching, you need to drive the gate from a low resistance source, which is what mosfet driver chips are for.

That goes by gate threshold voltage I guess? Or the 10v they usually spec it at?

The gate voltage plateaus somewhat above the threshold voltage as the gate charges/discharges. A typical non-logic level MOSFET will have a plateau about one third of the 10V gate-fully-on voltage, about 3 or 4V. The exact voltage of the plateau varies with device and drain current and voltage though.

Notice that driving through a resistor means that as the plateau is closer to 0V than 10V, it will discharge more slowly than charging (if plateau at 3V, then there is 7V across resistor when charging, -3V across the resistor when discharging, over twice as slow to switch off.

For this reason you'll see MOSFET driver chips usually provide more current sink than source since they want to correct this imbalance of switching times. My favorite MOSFET driver is probably the MIC4422 which can source or sink upto 9A, Yes, 9A... It makes it a handy MOSFET half-H-bridge in its own right (in fact I've used it as a class-D audio amplifier output stage!).

Wow that's pretty nice, how fast can that switch?