FET Based Motor Driver Is Better Than L298N

If you want to build a small robot with a motor, you are likely to reach for an L298N to interface your microcontroller to the motor, probably in an H-bridge configuration. [Dronebot] has used L298N chips like this many times. In the video below, he uses a TB6612FNG instead, taking advantage of the device’s use of MOSFETs. The TB6612 may be a little more expensive, but it’s clearly worth it.

You can get breakout boards for the tiny chips. [DroneBot] looks at several ready-to-go breakout boards. They are not drop-in compatible, though. For example, the L298N can operate motors from 4.5 to 46V while the TB6612 can go from 2.5 to 13.5V on the motor voltage. The L298N also handles more current. However, because of its relatively low efficiency, it needs a heat sink. The TB6612 boasts up to 95% efficiency and also has a low current standby mode. Of course, the TB6612 drops much less voltage which is great if you are using low voltage motor.

Assuming the new device is suitable for your hardware, the software isn’t really very different from L298N programs. If you know how to use the L298N, you can probably just snag a break out board, download the library, and be off to the races — no pun intended.

If you want more basics on the h-bridge, we’ve covered it many times. Of course, you can always use relays if you want real old school.

20 thoughts on “FET Based Motor Driver Is Better Than L298N

  1. This is way better that the “Old Standard” L298N chips!!
    Have you looked at the A4950?
    I’ve used these in a BBB driven design.
    A single H bridge at +/-3.5A and 40V. All in an “8-pin SOICN with exposed thermal pad (suffix LJ)” package.
    If I get around to it, I’ll do an Arduino shield for a couple or 3. That is something I’ve been toying with for a while.
    Yes, it is way past the time to give the L298N chips the boot!

  2. Why did manufacturers not modify the L293D and L298N motor drivers. There were many old ICs that are now produced with modified designs, and sometimes with more features added to it.

    1. May be you are the type tat don’t read datasheets or just have no idea. Part numbers are meant to *follow* their published specs, so products can be built over the year buying the same parts. So if a part rev. changes its specs, they should no longer same part number or they can no longer be tracked in the inventory. e.g. There are applications where older transistors were far superior because they can dissipate heat better than newer devices with smaller dies. God help you if your products relies on that and the engineers retired.

      There are cases where the same parts can no longer be manufactured as the process/chip fabs are obsoleted. Some of these are now built on newer often smaller geometries and often require resigning effort due to change in libraries specific to the process. Each process steps have their design rules/limitations and optimized layouts. If the parts aren’t popular enough, they are simply abandon or the remaining dies/mask sold off to some other companies that deals with obsoleted parts.

    2. Sometimes designers improve a product or fix a bug or qualify the part to a different specification and release a new rev of the part, and if they do they’ll give it a modified name: L298NA, for instance. (Atmega328P, atmega328PB, being a really obvious example.)
      But in general there’s no incentive to do that. If you make an updated part, you give it a new number, even if it’s just a design spin, because then you have a new part in your portfolio and that’s attractive to everyone: corporate, the sales team, the people buying it. It’s New! So it must be better.
      And usually it simply isn’t worth redoing an old part because your designers have moved to a new silicon process and have a newer state machine and newer control loop algorithms.

  3. The 13.5V limit was the big fail for me.. Can’t use a 14.4v three cell lipo.. Shame, as otherwise it would have been idea.
    The A4950 mentioned above looks really interesting.. Though I’d like to find a i2c based chip.

    1. You could use the chips from Trinamic such as the TMC2130, they are SPI steppers drivers, but you can use both H bridges separately by specifying your desired PWM value through SPI.
      So, two DC motor drivers, 30V max input voltage, and 2A max output current for TMC2130, they also have a 4A version called the TMC2660.
      There’s also integrated current limiting which is always a nice feature to have.

      I’ve used the A4950 a lot, it also have internal current limiting, it’s a very nice chip, well worth having to generate the PWM yourself.

      1. Higher directly off the charger or while on the charger. I do use the TB6612 but I’ve had them damaged with motor spikes and other things as everything is just a little too close for comfort and I can’t use a 4 cell.
        On 2 cells they do work very nicely though.

  4. There’s so many alternatives to ye ol L298 and its’ sibling L293 in all their form factors. Far cheaper and and greater Ampere ranges thanks to FET. Less voltage drop big plus and most have builtin protections that dont exist on either oldie. Built in PWM and choppers. Been playing with MX1508 H-bridge modules. 10V/1.5A supposedly 2.5A max. Havent tested for 2.5A yet. Not bad for most toy motor apps and cheap. Step Sticks (A4988 and other) pretty much out-moded classic h-bridge for small stepper motor use. Thanks 3D printer junkies!

  5. For those who aren’t experts in power electronics, I feel like the key point here has been glossed over.

    The L298 uses bipolar Darlington output transistors. They work fine, but they have a voltage drop of about 2 V each, which is results in low efficiency and high power dissipation.

    That was pretty much the only option many years ago, but more recently, they have invented power ICs with MOSFETs for the outputs. Those have much lower voltage drop (if you don’t run them at really high current), so they avoid those problems. If you want to shave nickels from your BOM cost, the old-school parts still come in a bit cheaper.

  6. L298N would be a good device to convert to a full CMOS solution that will proved higher efficiency, lower power dissipation, etc. The only issue is its extremely wide operating range of +4.5V to +40V. A CMOS device will have a narrower operating voltage range. I One can create a fully pin compatible device that works from +4.5V to +10V, then another from +9V to +20V, and another from ~ +18V to +40V. In fact, what would be the most common and most appropriate range? I am a chip designer and I would be happy to design a L298 compatible device. Of course, the interface will be fully logic (TTL or CMOS Digital) compatible for all of the devices. Just the output circuitry would require different output devices to achieve the higher efficiencies required.
    This device can be designed in about 6 mos to 1 yr, if there is a market that can justify committing volumes of >10K\k to 100k pcs per yr.

    1. There are 100’s of (maybe 1000’s of motor drive IC’s already available, it’s part of the maddnes of abundance in today’s world.
      Just recently I bumped into the L6234. It has 3 half bridges (so not pin compatible) but it runs from 7V to 52V and can deliver 5A peak currents (4A continous), and with it’s DMOS power transistors it has built-in freewheling diodes.

      I find it hard to imagine there would be a serious market for a pin compatible chip for the L298, mostly because of it’s lack of internal freewheling diodes. The popularity of that chip is probably mostly due to historic reasons, and people clicking on the “buy” button on shops targetting hobbyists who are only half aware what they are buying.

      Personally I have no knowledge of chip design, why would it be so difficult to make chips for a wide power supply voltage range?
      From the L6234 datasheet:
      It is realized in Multipower BCD technology which
      combines isolated DMOS power transistors with
      CMOS and Bipolar circuits on the same chip.
      By using mixed technology it has been possible to
      optimize the logic circuitry and the power stage to
      achieve the best possible performance.

      The datasheet of the L6234 is dated August 2003, so it is also not a new chip.

      Another development I’ve discovered recently is that (finally!) microcontrollers are coming into existence with built-in MOSfet driver stages.(STspin)

    2. I realise this is an old thread, but one use those h-bridges are put to (via an arduino motor shield) is for DCC++ model railway control. That requires 14.5 V (ish) so a +9 to +20 V would be ideal, with current from 1 to 5 A for a good sized layout before needing boosters. The 298 maxes at about 2A.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.