Charge pump to drive a MOSFET

Thread Starter

sumeryamaner

Joined May 29, 2017
114
I have a (maybe stupid) question.
I am designing and building various little circuits for RC aircraft. I have built some electronic switches for different purposes. For example to control the ignition of a gasoline engine or to switch ON/OFF the entire power supply of an aircraft.
In the former it is very straightforward. I am using an Attiny85, an opto coupler and an N Channel enhacement type logic level MOSFET for low side switching.
For a main switch this approach is not preferred as the usual way of controlling the power is high side switching. High side switching with a P channel MOSFET has known problems. First of all, P channel MOSFETs have a higher RDS ON values. But a more serious problem is that in high side switching the switching performance depends on the input voltage. For example, if I use an input voltage of 5 V (which is a standart value for RC systems) the P channel MOSFET would not conduct fully even if I tie the gate to ground because -5V VGS is not sufficient to fully open the MOSFET.
Of course there are numerous solutions for MOSFET gate drive but I want to experiment a little bit using discrete components which I already have.

My plan is to pull the gate of the P channel MOSFET to a lower level than the GND and achieve saturation of the MOSFET as much as possible. I think I can construct a simple charge pump for a negative voltage and feed it with an output from the Attiny85 with a PWM signal (duty cycle 50%). This negative voltage will be connected to the emitter of an NPN transistor (like 2N2222) which has it's base connected to ground and it's collector connected to the gate of the P channel MOSFET.

The proposed schematic is attached.

I really don't have the time to test this circuit on a breadboard until mid April. So I would like to hear your comments about this circuit. Thank you very much.
 

Attachments

Thread Starter

sumeryamaner

Joined May 29, 2017
114
example: P Channel Mosfet
Id = 1.3 amp
Rds On = 0.20 ohms @ -4.5 V Gate
https://www.onsemi.com/pub/Collateral/FDN336P-D.PDF

Small & lightweight in a SuperSOT-3 package

How many amps does your project need?
Thank you very much.
My current demand is about 5A with peaks up to 10 A.
This seems to be a very nice MOSFET. But I am an amateur and my PCB technology (!) can use only through hole devices. Of course I can try to use an SMD MOSFET but it's avalability in my region is questionable.

BTW I think I could connect two or three of them in parallel.
 

Thread Starter

sumeryamaner

Joined May 29, 2017
114
Maybe this one:
https://www.mouser.com/ProductDetail/ON-Semiconductor/SFT1342-W?qs=sGAEpiMZZMshyDBzk1/Wiwkux6%2B2oWnVK/h5PSOTplFeci60yDp1GQ==

Does your power supply typically hold voltage under those 5-10A loads? If the voltage sags to 3V or such, that could raise the Rdson of the MOSFET
The circuit is being fed by a 2S LiPo battery (4000 mAh / 25C) and 10A is not much for it. :)

The SFT1342 is not available in my region. At least not in my usual online sources in Istanbul
 

dendad

Joined Feb 20, 2016
4,451
I have used a charge pump driven by a PWM output to generate the gate drive for an N channel FET for high side switching.
This was on a Hula Hoop running addressable LEDs. The current was not very high, just 100mA or so.
The hoop ran off a couple of LiPo cells in parallel so it was less that 4V, and the logic level N FET gate was raised close to 4V above the battery +ve supply. It works quite well.
This is just for on/off power, and will not switch fast. If you want that, extra switching, like an opto coupler could be added.
 

Thread Starter

sumeryamaner

Joined May 29, 2017
114
I have used a charge pump driven by a PWM output to generate the gate drive for an N channel FET for high side switching.
This was on a Hula Hoop running addressable LEDs. The current was not very high, just 100mA or so.
The hoop ran off a couple of LiPo cells in parallel so it was less that 4V, and the logic level N FET gate was raised close to 4V above the battery +ve supply. It works quite well.
This is just for on/off power, and will not switch fast. If you want that, extra switching, like an opto coupler could be added.
Switching speed is not critical but I somehow don't like the idea of N Channel MOSFET in high side switching. I want to switch a voltage between 5V and 8.4 V with this device. Supply voltage of the logic stage will be 5 V. So I think I will need a multi stage voltage doubler at the PWM output.
 

dendad

Joined Feb 20, 2016
4,451
The FET gate volts will be 5V (for example) above the battery volts, irrespective of the battery level as the diode voltage multiplier "0V" connection is the battery +V, and the PWM is capacitivly coupled to that. It would be an idea to add diode clamps to the PWM pin to clip any transients that may appear on the battery form getting back to the processor.

I went with N channel FET as I have many of them and it worked well.
 
Last edited:

Thread Starter

sumeryamaner

Joined May 29, 2017
114
The FET gate volts will be 5V (for example) above the battery volts, irrespective of the battery level as the diode voltage multiplier "0V" connection is the battery +V, and the PWM is capacitivly coupled to that. It would be an idea to add diode clamps to the PWM pin to clip any transients that may appear on the battery form getting back to the processor.

I went with N channel FER as I have many of them and it worked well.
I'll give it a try. Thank you.
 

Ian0

Joined Aug 7, 2020
9,667
Just a thought. . .
Analog Devices does a device called LTC7001 which will drive a N-channel MOSFET to do the same job. Not cheap, but neither are P-channel MOSFETs!
 

crutschow

Joined Mar 14, 2008
34,280
Here's the LTspice sim of a driver for a high-side N-MOSFET that uses a CD4993 single IC oscillator/driver to generate a gate-source voltage of about 0.6V below the supply voltage, which requires just a DC level control signal:
The 5V enable signal (green trace) turns on the oscillator (blue trace), which generates the MOSFET Vgs voltage (red trace) that turns on the MOSFET (yellow trace).
The turn-on time for the MOSFET shown with a gate-charge value of 18nC, is <20µs and the turn-off time is <1ms.

For 5V operation, the N-MOSFET can be any logic-level type [max Vgs(th) ≤2v] with the appropriate voltage and on-resistance rating.

If you want to use a micro to generate the oscillator signal, then you can eliminate the IC and just connect the ≈50% duty-cycle PWM signal to C1.
The value of C1 depends upon the signal frequency and may have to be increased if it's a frequency lower than about 100kHz (here about 240kHz).

Bordodynov uses a more complex voltage-tripler circuit to generate a higher Vgs from a 5V signal, so you can use standard (not necessarily logic-level) MOSFETs.

1695752286189.png
 
Last edited:
Top