P channel power mosfet IRF9540 or similar

The International Rectifier datasheet I looked at had +/- 20V as an _absolute maximum_ rating, so -25 doesn't seem like too good an idea.

-- John

Wouldn't that depend upon "the rest of the story"? I guess if he hooks the drain to -12V, then it should be ok, but if he has ideas of hooking it to

Right thanks, as I mentioned, i am a real noobie with FETs and hadn't spotted that maximum on the datasheet, i can change the voltage hopefully simply, I was trying to svae real estate on the PCB, but it wont take too many components.

Thanks for the adice

pete d

He did say he was talking about gate-to-source voltage (Vgs), but maybe the actual picture is more complicated.

-- John

I can see why you think I want to connect it to a static voltage, I missed part of the explanation, what I realy menat was mmy PIC would switch a transistor to connect g to the -25. I generate a more suitable gate output. These FETs have a steeper learning curcve than bipolar.

thanks both

pete mc

Driving P-channel MOSFETs is not a trivial task, and there are not nearly so many drivers available as there are for N-channel. Depending on the frequency at which you will be switching, it is very important to drive the gate hard and fast to reduce switching losses.

Linear Technology has a P-channel driver LTC1693-5, but it needs a 5-12 V nominal Vcc. They have a high voltage high side driver LTC4440, but it is designed for N-channel, and utilizes a boost circuit and level shifters to drive the MOSFET at 80 V on the high side. TI has a complementary driver UC3714 which can drive bith N-channel and P-channel, but the P-channel drive appears to be for an auxiliary switch in a forward converter.

You can make a three-transistor level shifter to produce -12 V gate drive from a 5 V logic level, but it will be inefficient and slow, unless you can create a -12 VDC rail.

Depending on current requirements and speed, it may be better to use an optoisolator or a small relay to switch the negative voltage.

Paul

thanks for that but-

Wow that sounds complicated, am I trying to make it too simple with

| | | ||-+ | '--------. | === | Load GND .-. | | | | '-' | | |\\ ___ |/ -| >O----|___|--| |/ |>

GND (created by AACircuit v1.28.6 beta 04/19/05

This seemed easy and I nicked the concept off somebody elses circuit, did I misunderstand the concept?

pete d

Ahah! With a +25 VDC supply, that changes things immensely. Your circuit will work if you add a resistor across the zener to bleed off the gate charge and turn off the MOSFET. In fact you can just use two equal value resistors (about 1k each) to apply 12 VDC to the gate when the NPN transistor is on. The resistors and gate capacitance will cause slow gate voltage change and high power dissipation during the transition through linear range.

The simplest way to apply power to the load would be to use an N-channel logic level MOSFET and have the load from drain to 25 VDC, if it does not need to be grounded.

You could also use a high-side driver and an N-channel MOSFET. They come in small SMT packages and cost about $1, and are almost essential if you are switching at high speeds.

Paul

I see, thanks fr the clarification, there are several different voltages all referenced to 0V so the load does really need to be grounded, it is interesting about the slow switching, on this particular circuit the load will only be switched occasionally so it isn't really a problem, but another project I am working on will be PWMing a solar panel to charge 24V lead acid battery (yes 2 or actually 3 jobs on at the same time) so I guess the switching tine will become a factor then that will mean going to one of the methods you suggest. the high side driver sounds interesting I'll go look for one and study the datasheet see if I understand how to use it. FETs are a bit of a mystery to ne, so there's lots of learning to do.

Many thanks again for the advice and for taking the time to give such thorough answers.

pete d

it's mostly right, but

How is the mosfet that supposed to turn off?

when the NPN transistor turns on current flows through R1 and pull the voltage on the MOSFET base down to 15V, that's enough to turn on the MOSFET and led current flow into the load.

but.

when the transistor turns off nothing happens, well if you wait long enough the MOSFET will catch fire - the problem is there's nothing that can pull the gate back up to 25v to turn the mosfet off,

if you're unlucky the zener will leak a little and the voltage on the gate will increase slowly until the mosfet is turned half way on, then half the power goes to your load and the other half overheats your mosfet.

here's one way to fix it.

Change the zener to a resistor equal to R1. That way the two resistors form a voltage divider giving the gate half of 25V. and when the NPN is turned off the gate can discharge back to 25V through the top resistor.

Bye. Jasen

Thanks, that makes a lot of sense, i have added the resistor but left the zener in as it is just possible that if the connectons to the circuit are made in the wrong order there could be a lot more volts than

1. I think from what Paul said that is okay to do.

Thanks again

pete d

remove the ZEner, and put a resistor there., remove the resistor from the collector and run straight to the Gate.

Thanks, but unfortunately I have to allow for the possibilty of a much higher voltage on the 25V line temporarily and I think that would allow the entire possible max volts across the gate source junction, only for a very short time, but i don't know hopw the FET would cope with that. As usual the devil is in the detail and I only put a small part of the circuit up

pete d

ok, Keep the zener along with the Resistor then. but do the rest as i suggested. if protection is what your worried about, you may want to consider using a TVS diode (AC Type) because it's also possible you could get a pulse from the common side. We had that problem not to long ago. the TVS took care of that. We no longer have FETS dropping out mysteriously.

Okay thanks I'll look into TVS diodes, not come accross those before.