2
\$\begingroup\$

I'm designing a low-side PWM driver for a 24 V brushed DC motor that draws around 15 A of continuous current. The motor is driven using PWM at a frequency between 2 kHz and 5 kHz, with a variable duty cycle depending on the operating conditions. The traditional approach is to place a Schottky diode in parallel with the inductive load, as shown below.

conventional flyback diode across the motor

While this solution is simple and reliable, I've run into a practical issue: Schottky diodes capable of handling this current are physically large and occupy a significant amount of PCB area. Board space is an important constraint in my design.

Since the main current path is already switched by a low-RDS(on) NMOS in a compact package (such as DPAK or D2PAK), I started wondering whether the flyback path could also be implemented using a MOSFET instead of a diode.

One idea is to use an actively controlled NMOS (possibly with a bootstrap driver or another control method) so that the MOSFET conducts during the freewheeling interval, reducing both conduction losses and PCB area compared to a high-current Schottky diode.

My questions are:
  1. Is an active MOSFET-based flyback path a practical and reliable alternative for a 15 A inductive load operating with 2–5 kHz PWM?
  2. What are the main challenges regarding timing, gate drive, dead time, and reliability compared to a conventional Schottky diode?
  3. Is there a commonly used topology for this purpose in DC motor PWM applications?
  4. If this approach is not recommended, what alternatives are commonly used to reduce the size of the flyback circuit for 15 A or higher inductive loads?

I'm interested in practical implementations, application notes, or examples used in commercial motor drivers.

\$\endgroup\$

3 Answers 3

Reset to default
5
\$\begingroup\$

Is an active MOSFET-based flyback path a practical and reliable alternative for a 15 A inductive load operating with 2–5 kHz PWM?

It is commonly used.

What are the main challenges regarding timing, gate drive, dead time, and reliability compared to a conventional Schottky diode?

The main challenge is ensuring you have sufficient dead-time (anti-shoot-through). Any MOSFET in the upper position will have an internal body diode that acts like the Schottky in your basic design so all the MOSFET is doing is shorting out that internal diode to maintain high power efficiency. There are specialist gate drivers that will fit your requirements. Look for high-side gate drivers.

Is there a commonly used topology for this purpose in DC motor PWM applications?

I would suggest that the topology is very similar to a synchronous buck converter (aka half-bridge) and, as such, the motor load return can be grounded or taken to the main positive rail. Taking this one step further and it becomes a full H-bridge with capabilities to spin the motor in reverse. That's another option to look into.

If this approach is not recommended, what alternatives are commonly used to reduce the size of the flyback circuit for 15 A or higher inductive loads?

This approach (half-bridge/synchronous-buck or H-bridge) is recommended.

Regarding the IR2104, this is what google AI says (and I confirm its validity): -

enter image description here

But, please do this yourself to see the full list of constraints and, if you give current and voltage details for the motor, it will recommend MOSFET types.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks a lot for the feedback. Could I perhaps use a half-bridge driver like the IR2104—with the low-side driving my NMOS and the high-side driving my flyback—given that the IR2104 already features dead-time control and is bootstrap-compatible? \$\endgroup\$ Commented 11 hours ago
  • 1
    \$\begingroup\$ @LucasMarinoTorres sure, the IR2104 is a standard part for this type of application. However, you need to ensure that the bootstrap capacitor is high-enough in value for 2 kHz operation. You also need to ensure that you have sufficient DC bus capacitance local to the MOSFETs to ensure that when recirculating energy from the motor, the BUS voltage does not rise too much. And, you must always use PWM i.e. high and low transitions are always needed to keep the bootstrap capacitor charged. This means not running at 100% duty. \$\endgroup\$ Commented 11 hours ago
  • \$\begingroup\$ You could use a P-MOSFET for the flyback, which doesn't require a bootstrap circuit, and can operate at 100% duty-cycle. Of course, dead-time requirements must still be met. \$\endgroup\$ Commented 9 hours ago
  • \$\begingroup\$ @CarlRutschow you could leave that as a pukka answer. But obviously the MOSFET driver chip would have to change. \$\endgroup\$ Commented 7 hours ago
3
\$\begingroup\$

  1. An active flyback path - a.k.a. a synchronous rectifier - is practical even in compact, MHz-switching converters. It is very common.

  2. I'd argue the biggest challenge is usually heat removal, and keeping the efficiency up.

    • If you're implementing on an MCU, there's plenty of them with built-in dead-time generators and hardware shoot-through protection. That might be the easiest for you to implement.

    • If the platform is an FPGA, then you have full control over the logic and can have proper hardware protections, i.e. ensuring minimal dead-time even if the timers are fed "gibberish".

  3. An H-bridge is typically used for brushed DC motors. The flyback current can then be controlled exactly the same as the forward current, and you can do recuperative braking as well.

  4. Size reduction requires reduction in heat loss, so moving towards more efficient topologies is the common way to make things smaller.

\$\endgroup\$
0
\$\begingroup\$

Others have mentioned the H-bridge already.

Perhaps you've considered this, and I'm not sure how tiny you hope your solution to be, but you don't need a 15A rated diode to survive repetitive, momentary 15A surges. How long does the flyback last here?

For example the SS3P5, 3A part, rated for 45A (non repetitive) surges that last 8.3ms. Non-repetitive avalanche energy is rated, so compare that to your circuit (inductance, capacitance, duration of the flyback event). Other/better pulse rated schottky or normal diodes can be found; this is just an example.

You can also "snub" some of it with a capacitor or R+C circuit across the FET or across the coil, to share some of the burden. Surge rated resistors will happily operate under repetitive surges and increased temperatures. You can also use a TVS/Zener diode in parallel with the FET which changes the flyback operation -- higher voltage and thus dissipates faster; whereas a schottky is "more efficient" and thus dissipates the stored energy more slowly. TVS are generally pretty sturdy and surge rated but of course do your own analysis.

Ultimately the same amount of energy must be dissipated, except for what ends up lost in the winding (or other series resistance) or recirculated into the VCC.
This will be true in the H-bridge as well, the difference here being that more of the energy gets burned in the windings or recirculated into the VCC supply, by virtue of being a smaller proportion of the resistance/voltage-drop.

\$\endgroup\$
1
  • 1
    \$\begingroup\$ This is a PWM application, so it's repetitive ratings that are needed, not non-repetitive. \$\endgroup\$ Commented 6 hours ago

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.