0
\$\begingroup\$

What is the best way to drive a MOSFET with low voltage?

I have encountered several projects where power from a battery is used to drive motors or source a DC/DC converter.

In most cases, we used an IC fitting our needs, but I always wondered how to drive a MOSFET half-bridge at low voltages.

I'm used to bootstrap drivers working with 10V+.

Is it worth it to boost voltage high enough to drive the gate driver with a small PMIC, or are discrete solutions better for low voltage applications?

As an example, what solutions would you recommend for a synchronous boost converter with the specs:

  • Uin = 1.8V
  • Uout = 4.2V
  • Iin(max) = 2A
  • f = 250kHz

I'm more comfortable with two N-ET and a driver, but are there other solutions?

I'm not interested in integrated controllers for DC/DV converters.

I'm looking for an answer to the question: "How do you fast switch a MOSFET with low voltage?"

\$\endgroup\$
3
  • \$\begingroup\$ It entirely depends on the MOSFET chosen. Choose your MOSFET. \$\endgroup\$
    – Andy aka
    Commented Aug 30, 2021 at 16:11
  • \$\begingroup\$ Or I choose a driver frist, because i think there are more fitting mosfets to a driver than vice versa \$\endgroup\$
    – Shirokuma
    Commented Aug 30, 2021 at 16:19
  • \$\begingroup\$ This is what you wrote: but are there other solutions? hence, why I said choose a MOSFET (that can be driven with a gate source voltage as low as 1.8 volts and still provide low on-resistance suitable for a 2 amp load). \$\endgroup\$
    – Andy aka
    Commented Aug 30, 2021 at 16:28

1 Answer 1

1
\$\begingroup\$

There are over 50k types of MOSFETs.

overview

Many sites have precise design software with layout and BOM. TI, Microchip LTI Have your tried that? , then choose based on cost, size, performance.

How much do you want to trial and error or follow the shoulders of giants and still possible fail from hidden assumptions like step-load range, EMI, crosstalk, start-surge overload and overvoltage tolerance. The details are how you define all your expectations and assumptions.

** make a list of ALL specs BEFORE choosing/ designing Every & Anything.**

Ticky tacky details

  • For a ** Boost converter, a Half-Bridge with deadtime control, logic level and high speed.** is needed. and that's exactly what you could use above in a search engine and find solutions from ; ST, ON, TI, INFINEON, NXP, ROHM

My preference for easy use would be a Half Bridge with 1.2V input threshold with 50 mV hysteresis and a single input and enable rather than LI HI for LO HO outputs. So you don't need a charge pump or Boost cct, but that is also possible as long as you guaranteed dead-time and snub during that time with low L or ESL.

For this application, marketing calls them "Logic Level" FETs. Engineers specify Vt, to be 30% to 50% typically of Vdd to achieve a reasonable RdsOn. 2 to 4V= Vt types need Vgs=2.5 * Vt , while FETs designed well below this only need ~2 * Vt typically.

Output Load = 4.2V/2A= 2.1 Ohms
Vgs = 1.8V-10% = 1.62V
Choose RdsOn < 1% of load =< 21 mOhms @ < 50% of 1.62V = 0.81V as a search in Disti website.

But the datasheet will specify Vgs with Max Rdson for typical supplies -10% and Vt, aka Vgs(th)

Search for anything rated at your Vdd-10%. there are many < 1V choices mainly in SMT.

\$\endgroup\$
2
  • \$\begingroup\$ I found a lot of N-Fets around the specs you gave. This seems fine, even when I would prefere a stronger driver than my logic outputs. But what is about high side? With logic-level a P-Fet seems fine. But how can i do the level shifting to turn the fet off. And how is it done fast. Therefore i'm asking if in the end the integrated "overkill" version with a driver and a drivers supply is the better way to go. \$\endgroup\$
    – Shirokuma
    Commented Aug 30, 2021 at 16:27
  • \$\begingroup\$ Many sites have precise design software with layout and BOM. TI, Microchip LTI Have your tried that? , then choose based on cost, size, performance. How much do you want to trial and error or follow the shoulders of giants and still possible fail from hidden assumptions like step load range and overvoltage tolerance, \$\endgroup\$ Commented Aug 30, 2021 at 16:34

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