simulate this circuit – Schematic created using CircuitLab

The aim is to design a synchronous boost converter using a microcontroller ATMega 2560 switching at 31.25 kHz.

5V PWM signal is fed from the microcontroller to the high and low side driver (IR2101).

My current output is about 15A max and I read somewhere that ideally, diode current should be limited to 7A max to prevent high power dissipation.

I would like to minimize power dissipation through the diode and maximize efficiency.

I tried turning on the high side switch with NOT gate signal from the low side PWM signal.

High side mosfet turns on when the low side is turned off but the voltage remains at 4 volt and does not increase.

What could be the cause of this problem?

  • \$\begingroup\$ It is much easier to use a diode than to implement a high side switch and all of its complex circuitry. \$\endgroup\$ – Harry Svensson Jul 12 at 13:01
  • \$\begingroup\$ My current output is about 15A max and i read somewhere that ideally, diode current should be limited to 7A max to prevent high power dissipation. I would like to minimize power dissipation through the diode and maximize efficiency. \$\endgroup\$ – DesHam Jul 12 at 13:42
  • 2
    \$\begingroup\$ If you're going to have an output current at about 15A, then you want a good answer, a good answer requires a good question, your question needs to be improved a bit, such as, where is the PWM coming from? Is it straight from the ATMEGA 2560? Are you properly turning off the low side transistor before turning on the high one, and then properly turning off the high one before turning on the low one? - You are sitting on all the information, share as much as you want, the more you share, the better answer you'll receive. - Do not add more information in comments, edit your question instead. \$\endgroup\$ – Harry Svensson Jul 12 at 13:52
  • \$\begingroup\$ One question I had: how do you ensure that the high and low N are not overlapping at the same time? Does the ATMega PWM generator do this? If not, maybe add a non-overlapping circuit to guarantee this doesn't occur - that is, something that inserts a dead-time in the switchover. \$\endgroup\$ – hacktastical Jul 12 at 20:50

You're trying to use an N-channel MOSFET as a high-side switch. That isn't going to work without a special gate driver that includes a boost circuit.

Besides which, you've got it the wrong way around — think about which way the body diode is pointing!

You should probably be using a P-channel device at M2, but even then, you have to make its gate drive relative to its source, which is the output voltage, not the input voltage.

  • \$\begingroup\$ I used a high and low side driver in my simulation to switch the mosfets \$\endgroup\$ – DesHam Jul 12 at 13:44
  • 3
    \$\begingroup\$ Then you need to show that in your question. I can't answer something you didn't actually ask! \$\endgroup\$ – Dave Tweed Jul 12 at 13:51
  • \$\begingroup\$ N-FETs have lower Rds(on) in a given package, which is why most folks doing high-power work go to the trouble of using them on the high side instead of p-FETs in spite of the special biasing needed to lift V(gs) high enough to make them work. \$\endgroup\$ – hacktastical Jul 12 at 19:35
  • \$\begingroup\$ And he says he's using the IR2101, which does the boosting. \$\endgroup\$ – hacktastical Jul 12 at 20:46

All half bridges using dual NFETs must use PWM only needed on low side with series cap to Vdd diode clamp to create a boost voltage >2Vgs(th) above Vdd to drive the high side gate voltage.

The high side then can be controlled independently for say a Boost / Buck conversion with an inductive Load by storing + or -ve inductive Current energy.

  • the clamp diode only conducts when the the low side turns off after charging the series capacitors then the high side pulls up the capacitor to forward bias the clamp diode. There must be some xxx ps or ~1us of deadtime depending on reactance if Load L/R=Tau decay time and turn off recovery time of switches which is often sped up with a shunt diode across Gate resistor between LO and Low side gate.

You can't use M2 like that. MOSFETs have an anti-parallel body diode and yours is upside down freely conducting. You also need a high side gate driver.


There’s nothing inherently wrong with using an n-FET on the high side (and it's the preferred solution since n-FETS typically have lower Rds(on) for a given size), but watch the body diode (you have it backwards in your design). Also, to drive the high-side FET you need a bootstrapped high-side gate driver to make a high enough Vgs to turn on FET M2. (Huh, I said 'high' three times. Must have something on my mind...) I see that you've done that with the IR2101 so you're halfway there.

Anyway, having addressed those issues, here’s an idea: have a look at buck-boost and boost controller circuits that use n-FETs that deliver currents in your target range. Yes, you’re implementing just the boost part of buck-boost, but at least you will get an idea of component selection and circuit topology.

Buck-boost example: https://www.analog.com/en/products/lt8708.html#product-overview

And... Sync-boost example which has an input disconnect too, something your circuit could use: https://www.monolithicpower.com/en/mp9184a.html

One more thing: Some of the designs show a Schottky diode in parallel with the high-side FET, probably for speed-up of the overall switching time.

One other, other thing: are you ensuring dead-time (non-overlap) in your PWM drive between high and low side to prevent shoot-through?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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