I have heard many people talk about using a charge pump circuit instead of a bootstrap approach when supplying MOSFET gate drivers with a biasing level for the high side. I have yet to see anyone show what that would actually look like. I have already looked at 100% duty cycle drivers from Analog Devices with built in charge pumps. They are FAR too expensive and they don’t care to provide any useful information in their functional diagrams about how they are wired. My application absolutely must remain on indefinitely.

Here is the IR2101, a common gate driver for MOSFET transistors I have used before. Notice the bootstrap circuitry. I want to design a charge pump that operates up to the full 600v, all the way down to 15ish volts. I am tired of solutions that only work for one exact supply voltage or else they would become ineffective or exceed the Vgs rating of the MOSFET. I don’t want any extra ics, preferably just a few discrete components.

My biggest issue is exceeding the rating of the FET while the source is floating. My application has no guarantee that the low side MOSFET will ever come on at all. This means it would be best to float the charge pump off of the source, switching the oscillating capacitor terminals between the source and ground, just like a bootstrap circuit, only without using the main switching FETs. I do not know how to do this with transistor switches; switching between Vcc and GND would be easy.

enter image description here

  • \$\begingroup\$ I'm probably not reading as well as I should. But what's wrong with the TI's UCC2771x family? For example, UCC27710? \$\endgroup\$
    – jonk
    Commented Dec 27, 2019 at 6:27
  • \$\begingroup\$ I want to design a charge pump that operates up to the full 600v, all the way down to 15ish volts. What voltage are you talking about? Is this voltage across the FET's or across the load? \$\endgroup\$
    – Huisman
    Commented Dec 27, 2019 at 8:12
  • \$\begingroup\$ When the source voltage is floating, wouldn't the gate voltage float equally on top of the source voltage? \$\endgroup\$
    – Huisman
    Commented Dec 27, 2019 at 8:14
  • \$\begingroup\$ You could use a small isolated DC-DC converter \$\endgroup\$
    – bobflux
    Commented Dec 27, 2019 at 11:36
  • 1
    \$\begingroup\$ The bootstrap method relies on low side PWM activity but is also a Charge pump method per se. \$\endgroup\$ Commented Dec 27, 2019 at 14:43

2 Answers 2


Let's make a charge pump...

enter image description here

Voltage source V1 is a 12V 100kHz square wave.

"SW" is the SW node from your MOSFET driver, ie the source of the top FET.

This will charge the bootstrap cap of your MOSFET driver to about 11V above the SW node, so it should work.

Now, when the FETs switch, SW node will go from 600V to 0V (or the other way around) very quickly, so the flying capacitor C2, will source or sink a nasty current spike into the square wave generator output. I added D3 D4 and R1 to dump this into ground or into the 12V power supply decoupling caps.

This means C2 should be as small as possible while still supplying enough current to the driver.

If the FETs switch frequently or at high frequency this will increase losses and pump charge into the 12V rail, possibly raising its voltage.

These issues have to be weighted against the alternatives, which are using a small isolated DC-DC converter to power your top driver, or use a PMOS for the top FET.


This is the one most people are probably referring to. See the app note for how it works.

It is a bit confusing to trace out how it works. Basically it is a normal bootstrap augmented with a charge pump which only runs when the switch is on. It turns on like a regular bootstrap without the charge pump's help since the pump is deactivated. The 555 timer powers up when the high side switch is on. The 555 charge pump floats up charge to the bootstrap cap periodically to refresh it. The 555 timer's zener supply is referenced to the HV line and hangs 15V below the HV rail instead of sitting 15V above ground like usual.


enter image description here

  • \$\begingroup\$ This seems like what I am looking for. Do I need to use that special 555 timer or will any old 555 work? \$\endgroup\$
    – Hackstaar
    Commented Dec 27, 2019 at 18:48
  • \$\begingroup\$ @JudsonHudson I think either CMOS or TTL 555 timers will do. The main spec to look for is how much current the OUT terminal can sink or source. \$\endgroup\$
    – DKNguyen
    Commented Dec 27, 2019 at 18:51
  • \$\begingroup\$ @JudsonHudson I looked at Texas Instruments and their TTL 555 timer can sink/source 200mA while the CMOS 555 timer can sink 100mA but only source 10mA. That still sounds like plenty though and the CMOS 555 has better performance. \$\endgroup\$
    – DKNguyen
    Commented Dec 27, 2019 at 18:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.