Drive 5V with p-channel mosfet controlled by 3.3V

Hi!
I'm currently designing a driver for my LED cube and I want to control the anode layers with p-fets. On digikey I found the "AO4813" dual p-fet. I want to control the fet with a PIC32, so I will have 3.3V to drive it. But the voltage I want to switch is 5V. The gate threshold voltage on the mosfet is typically -2.0V. Am I right assuming it will work since 5V-3.3V are 1.7V, so driving it to 3.3V will shut the mosfet close?

Thanks for answers!

That's not what people consider good engineering practice. If you don't care about some leakage current and inadvertant turn-on during power supply sequencing, you might use this approach though. Minimal AO4813 Vgs,th (250 µA Id) is 1.5 V, b.t.w.

You might spend a little time looking for "load switch" IC
products, which you might find in this supply range and
having normal digital CMOS inputs - and perhaps less BOM
cost than a power PMOSFET plus gate level shifting / drive
network.

dick_freebird said:

You might spend a little time looking for "load switch" IC
products, which you might find in this supply range and
having normal digital CMOS inputs - and perhaps less BOM
cost than a power PMOSFET plus gate level shifting / drive
network.

Click to expand...

Would the MCP 14E10-E/SN be a reasonable choice? In the PMIC Load drivers section of digikey I only found drivers with horrible rise times...

Using a Vgs=2.5 PFET with a 3.3V supply will work but gives very little margin.

Using a Vgs =1V is more reliable.

Here is a dual P-FET
https://www.digikey.com/product-detail/en/SI5935CDC-T1-GE3/SI5935CDC-T1-GE3TR-ND/2441705

Couldnt i just use an inverting buffer running at 5v to drive the fets?

Depends on the FET gate capacitance. Earlier you did not
like the risetime of the load switch you found; you'd probably
like even less, the transition times of a power PMOSFET
driven by a 5mA logic buffer (which might peak out at 50mA
or so, but a 5000pF power MOSFET wants 1A-range peak
current to slap it around).

And in that vein, look at the conditions at which those
load switch transition times are spec'd - especially if the
output network is a big C, big or infinite R, the turnoff
transition may have almost nothing to do with the part
and all to do with the load. Like you'd see on a TPHZ/
TPLZ tri-state or mux spec.

And for a LED display it would be surprising if your scan
rate had to be anything close to transition-time-limited.

What, in fact, is your LED current? Maybe low enough
that the drive strength of a logic level shifter IC could
throw enough to satisfy?

Thanks for the answer! The current of the LEDs will (obviously) vary. one LED pulls 20mA, there are 3 per led (rgb) and 64 are active at one time (max) since it's an 8*8*8 cube. I don't expect a very high averag current since it will also be pulse width modulated^^

I am vaguely recalling a class of product called LED
segment drivers, from back in the day when 7-segment
(plus decimal) was the display norm. Have you run into
any of these in your searching? They might be appropriate
(or not, maybe they are all 5V inputs and more needy than
they're worth).

CMOS octal bus transceivers commonly were rated at
24mA drive (74xx245, et al) and I think I've recently
seen some that claim level shifting inputs for logic supply
domain crossing. That's 24mA at something like 10%-Vdd
voltage drop. Could be more suitable than the ordinary
combinational logic. You can find families down to below
1.5V so you should be covered if this would do.

Pretty bad idea to try what you suggested although you are correct that it might work. The problem is that the Vgs threshold is very poorly specified. Generally you want off to be well below it and on to be well above it.

Generally a good way to do what you want is to simply add an NFET to drive the PFET. Take your 3.3V I/O and drive a small "Logic Level" NFET gate with its drain pulled up to 5V. Then connect the drain to the PFET gate. Your I/O will do a good job turning on/off the NFET and the NFET will do a good job turning on/off the PFET.

The other thing, suggested above, is to get a proper "Load Switch". For 5V there are tons of parts specified for switching 5V USB power (0.5A to 2A) on/off from a 3.3V input signal. And they come with bells and whistles like current limits, thermal shutoff etc.

Try this:
https://www.ti.com/lsds/ti/power-management/usb-power-and-load-switches-products.page

asdf44 said:

Pretty bad idea to try what you suggested although you are correct that it might work. The problem is that the Vgs threshold is very poorly specified. Generally you want off to be well below it and on to be well above it.

Generally a good way to do what you want is to simply add an NFET to drive the PFET. Take your 3.3V I/O and drive a small "Logic Level" NFET gate with its drain pulled up to 5V. Then connect the drain to the PFET gate. Your I/O will do a good job turning on/off the NFET and the NFET will do a good job turning on/off the PFET.

The other thing, suggested above, is to get a proper "Load Switch". For 5V there are tons of parts specified for switching 5V USB power (0.5A to 2A) on/off from a 3.3V input signal. And they come with bells and whistles like current limits, thermal shutoff etc.

Try this:
https://www.ti.com/lsds/ti/power-management/usb-power-and-load-switches-products.page

Click to expand...

I assume I should look into the "Integrated Load Switch"es and the "Power Distribution Switch"es? They look promosing, thank you!