Driving higher voltages with p-type MOSFET

driving the gate directly from a mictrocontroller that uses +5 and ground. The problem is that I need +24 and +19. I looked into "high voltage level translators", but they have been discontinued or are a non-stock item.

Safest is an optocoupler, you can get opto gate drivers; will need a zener regulated supply hanging down from +24V.

If the switching speed allows, you could you a blocking capacitor with a reset diode; no good for switched DC operation; beware the power on charging of the blocking cap.

You could use an NPN transistor with a resistor up to the gate, and a zener to limit gate voltage; fairly safe?

Lots of options, depending on desired switching rate and power.

Grant.

I'm driving the gate directly from a mictrocontroller that uses +5 and grou= nd. The problem is that I need +24 and +19. I looked into "high voltage lev= el translators", but they have been discontinued or are a non-stock item.

s.

Use a second transistor to control the voltage to the MOSFET?

Well, it's actually half of a PNP/NPN pair that is switching back and forth= with a common gate at about 20 kHz. The NPN is no problem: I can switch i= t with +5. It's the PNP that's vexing me. The opto idea sounds good, if I = can find one that can handle the current (20 amps) and has a low drain-to-s= ource resistance. It also needs to be surface mount and the board is very = tiny.

Thanks for the ideas so far!

Don

On Tuesday, August 21, 2012 5:21:29 AM UTC-5, fungus wrote: Use a second transistor to control the voltage to the MOSFET?

Yeah, but I need it to turn on the output on a "low" signal, so I'd still have to drive it with a PNP at +24 and +19, which presents the same problem.

It's part of a PNP/NPN pair and one circuit has to shut off when the other turns on with a common gate signal from the MCU. I could use two NPN's and invert one gate signal, but there are three of these pairs and I'm trying to keep my chip count down.

Don

A little resistor divider chain and npn at the bottom to switch it, sounds easy.

have to drive it with a PNP at +24 and +19, which presents the same proble= m.

Add an inverter between the micro and the npn as switch. Or code the micro differently? ... or yet another transistor?

George H.

r turns on with a common gate signal from the MCU. =A0I could use two NPN's= and invert one gate signal, but there are three of these pairs and I'm try= ing to keep my chip count down.

Use a resistor inverter:

24V | R1 | +---to MOSFET gate | BJT --R2--MCU | | GND

When the MCU switches the BJT "on" the gate of the MOSFET will go to ground.

When the MCU switches the BJT "off" the gate of the MOSFET will go up to 24V.

It seems too simple, I must be missing something...

I'd use all N-channel parts and a half-bridge driver like the LM5109 or any of the plethora of other parts out there.

--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com

driving the gate directly from a mictrocontroller that uses +5 and ground. The problem is that I need +24 and +19. I looked into "high voltage level translators", but they have been discontinued or are a non-stock item.

I think this is the simplest implementation:

PNP

24VDC>-----+--E B C-+ | | | [R1] | | | | | +----+ | | | [3.9R1] [LOAD] | | C | MCU>-[R2]--B NPN | E | | | GND--------+--------+

--
JF

--
Yup, you punched through the MOSFET's gate oxide.

driving the gate directly from a mictrocontroller that uses +5 and ground. The problem is that I need +24 and +19. I looked into "high voltage level translators", but they have been discontinued or are a non-stock item.

--
Oops, PNP should read "PMOS".

driving the gate directly from a mictrocontroller that uses +5 and ground.

Why do you want to use a PMOS? If you want to switch a 24V circuit ON with 5V logic low, the PMOS is going to be inconvenient (the gate and source which control its conduction are up at the +24V side of the circuit).

I'll assume that you want the load to always have one pin grounded, and want to do a high-side switch between +24V and its other pin; that's a good reason to switch with a PMOS device.

So, PMOS high-side switch, with source connected to +24, gate pullup resistor to source, and gate pulldown with an NPN transistor, base connected to +5V, emitter through resistor to your logic pin. You can put a resistor from NPN collector to the gate, instead of straight connection, if the PMOS gate doesn't tolerate 19V.

Optoisolators are expensive and low-performance by comparison with even the lowliest simple transistor.

Oh, yes... :-(

Ok, add an extra resistor in parallel with the BJT.

24V | R1 | +---to MOSFET gate | /-+-\ | | R3 | | | | BJT --R2--MCU | | \ / GND

Now when the BJT is closed the voltage at the MOSFET gate is proportional to the values of R1 and R3 (hopefully low enough to keep the magic smoke inside the MOSFET).

--- No.

When the BJT is open the MOSFET's gate will see

(24V * R3) / (R1 + R3)

but when the BJT is conducting, the MOSFET's gate will be sitting at about 0.3V above ground.

That means that the drain-to-gate voltage will be about 23.7V, which is higher than the usual +/- 20V spec.

The right way to do it, I think, is like this:

. PMOS .24VDC>-----+--D G S-+ . | | | . [R1] | | . | | | . +----+ | . | | . [3.9R1] [LOAD] . | | . C | .MCU>-[R2]--B NPN | . E | . | | .GND--------+--------+

That way, when the input from the MPU is low the NPN won't be conducting, the PMOS gate will be at 24V and so will the drain, so the MOSFET will be off.

Then, when the MCU output goes high the NPN will turn on, connecting

3.9R1 to ground.

That'll pull the gate voltage down to about 19V, 5 volts lower than the drain voltage, which is what Don asked for. :-)

-- JF

--- Aarghhh...

. PMOS .24VDC>-----+--S G D-+ . | | | . [R1] | | . | | | . +----+ | . | | . [3.9R1] [LOAD] . | | . C | .MCU>-[R2]--B NPN | . E | . | | .GND--------+--------+

That way, when the input from the MPU is low, the NPN won't be conducting, the PMOS gate will be at 24V and so will its source, so the MOSFET will be off.

Then, when the MCU output goes high the NPN will turn on, connecting

3.9R1 to ground.

That'll pull the gate voltage down to about 19V, 5 volts lower than the source voltage, which is what Don asked for. :-)

-- JF

4VDC>-----+--D G S-+ >. | | | >. [R1] | | >. | | | >. +----+ | >. | | >. [3= .9R1] [LOAD] >. | | >. C | >.MCU>-[R2]--B NPN | >. E | >. | | >.GND--------= +--------+ --- Aarghhh... . PMOS .24VDC>-----+--S G D-+ . | | | . [R1] | | = . | | | . +----+ | . | | . [3.9R1] [LOAD] . | | . C | .MCU>-[R2]--B NPN | .= E | . | | .GND--------+--------+ That way, when the input from the MPU is = low, the NPN won't be conducting, the PMOS gate will be at 24V and so will = its source, so the MOSFET will be off. Then, when the MCU output goes high = the NPN will turn on, connecting 3.9R1 to ground. That'll pull the gate vol= tage down to about 19V, 5 volts lower than the source voltage, which is wha= t Don asked for. :-) -- JF

Yes, this is what I was looking for. Thanks John.

Don

--
My pleasure. :-)

what I was looking for. Thanks John. > >Don --- My pleasure. :-) -- JF

Well, wait. I guess this works, but it still doesn't do what I want.

This switches on the load when the output is HIGH. If I wanted to that, I could have just used an n-type MOSFET and controlled it directly with the MCU.

What I need is for the load to switch on when I give it a LOW. And as soon as you start thinking about using a PNP to switch it, you run into the same problem.

Am I stuck with an inverter?

Don

what I was looking for. Thanks John. > >Don --- My pleasure. :-) -- JF

could have just used an n-type MOSFET and controlled it directly with the MCU.

you start thinking about using a PNP to switch it, you run into the same problem.

+24 o--------------o---.-. .---- | | A | | ===== | | | | | | | | |_| | | | o-----' | | | | | |_| | |/ +Vdd o-----o-----| | |> | | ------- | | | | | CPU |o--' | | ------- | 0V o-------o---------

Now high on CPU output turns off the transistor, thus no gate voltage?

You might want a b-e resistor so the circuit is failsafe off while CPU starts, before the outputs are enabled. Voltage divider to prevent excess gate voltage as in other posts.

Grant.

what I was looking for. Thanks John. > >Don --- My pleasure. :-) -- JF

connect the NPN base resistor to the +5V supply and the NPN emitter to the MCU output.

--
?? 100% natural

--- Posted via news://freenews.netfront.net/ - Complaints to news@netfront.net