Switching a P-Channel MOSFET from 3.3v

I'm trying to use a P-channel MOSFET (FQP7P06) as a high-side switch to turn off power to my sensors/radio from a 3.3v ATmega328P.

The Vgs threshold is -2.0 to -4.0v according to the datasheet, the same as the FQP27P06 and FQP47P06.

The problem I'm having is that from a 3.3v supply its only putting out 1.9v so is barely even lighting an LED. From a 5v supply it lets through 4.93v as expected.

It shouldn't need a driver chip or transistor to run it as long as SOURCE and GATE differ by 2-4v, which they do as I can see on my mulimeter that the digital pin is switching between 0v and 3.3v.

Any ideas what the problem is here? Am I completely misunderstanding Vgs, rdsON etc? Do I need a different FET?

Am I completely misunderstanding Vgs, rdsON etc?

Yes

The Vgs threshold is -2.0 to -4.0v

Yes but the threshold value is just that, the value when the FET just starts to turn on, it is not the value when it is turned on fully. To see that you have to look at the RDS(on) figure. This shows what the gate voltage is when the device is fully on. You will see that this is 10V, so you can't switch on this FET with your 3V3.

Do I need a different FET?

Yes

You want a Logic Level, Low Rds, P-channel MOSFET.
This one should do nicely.

The part you selected is a Standard Level P-channel MOSFET, looking for a 0-10V swing on the gate. Logic Level parts have lower requirements.

You can find parts with even lower gate voltage requirements, like 1.8V, but they will be surface mount parts:

i knew you were going to say that. even at 5v it seemed a bit unsteady (LED was flickering almost unperceptably).

would a NDP6020P be better - VGS threshold is -0.4 to -1.0v and RDS(on) from 2.5-4.5v, its about the only one still on sale with those sort of specs, about a quid each:

http://www.fairchildsemi.com/ds/ND/NDB6020P.pdf

edit: CrossRoads, thanks, you selected the same part i found, the rest are all SMT as you say or EOL like the FDP4020P

my NDP6020P has arrived but things are behaving oddly.

i can see the voltage to my nrf24l01+ is flipping between 3.3v and about 0.8v (i'm a bit surprised that's not 0v) but what's odd is that after the first "power off" the radio never comes back on again. i can see the 328p has a constant 3.3v as that's not going via the mosfet.

if i comment out the code to set the digital pin high to turn off the power, everything runs ok.

i assume its something to do with the radio, as if i make a basic led circuit on breadboard the led goes on and off just fine.

code and circuit (sorry its fritzing)

sej7278:
I'm trying to use a P-channel MOSFET (FQP7P06) as a high-side switch to turn off power to my sensors/radio from a 3.3v ATmega328P.

Why? The nrf24l01+ only consumes 900nA when you put it in standby.

fungus:
Why? The nrf24l01+ only consumes 900nA when you put it in standby.

i'm not convinced, that's datasheet speak for the chip alone, not measured by anyone in a circuit with pna/amp etc; as far as i know.

something in my circuit is eating 2xAA in a week, its not the MCU, so i'm powering off the bmp085, dht11, ds18b20 and nrf24l01+ using the mosfet (or trying to!)

something in my circuit is eating 2xAA in a week

So find out what it is with a multimeter....

MarkT:
So find out what it is with a multimeter....

easier said than done. that's not really the issue anyway, the issue is why doesn't the rf24 recover after being turned off with a mosfet. i think i've got it fixed now, got to test a bit longer but it seems to be as simple as calling the radio.begin() etc. in the loop

Some devices are sensitive to the speed with which the rail rises on switch on - using
a FET means the rail rises in a few us, rather than ms - you might find increasing the
gate resistor value could have some benefit here.

It may also be that the leakage current, which is keeping the modules supply at 0.8V
when "off", causes the device's reset/power-on detection circuit to malfunction. Add
a 10k resistor across the device's supply perhaps to pull that 0.8V down to millivolts.

sej7278:

fungus:
Why? The nrf24l01+ only consumes 900nA when you put it in standby.

i'm not convinced, that's datasheet speak for the chip alone, not measured by anyone in a circuit with pna/amp etc; as far as i know.

something in my circuit is eating 2xAA in a week, its not the MCU, so i'm powering off the bmp085, dht11, ds18b20 and nrf24l01+ using the mosfet (or trying to!)

Five minutes with a multimeter and a switch will reveal the truth...

fungus:
Five minutes with a multimeter and a switch will reveal the truth...

for some odd reason none of my meters work as ammeters properly. and really it would be an utter cow to use a ammeter in a soldered circuit given that you have to connect in series.

any suggestions on a good ammeter that's not megabucks?

sej7278:
for some odd reason none of my meters work as ammeters properly.

You probably blew the fuses by not being careful. Take them apart and look.

fungus:
You probably blew the fuses by not being careful. Take them apart and look.

hmm maybe, they seem to display a non-zero reading even when not connected to anything, and it doesn't change when they are connected.

edit: ah yes you're right, the fuse is very smokey looking!

whilst i've got it open - any idea what the 201 potentiometer is for on a velleman DVM850BL ?