I'm designing a 24 V, 350 A, 70 kHz PWM DC brushed motor controller one direction only.

Here is what I came up with until now: Pwm circuit

I will use Pi Pico as the microcontroller it can generate 3.3 V PWM with variable duty.

Then signal gets fed in the left to the non inverting input of the LM318 op amp it was choosen becuase it has super high slew rate of 50 V/us so it will not take more than 4 us rise and fall times. I added a reference at the inverting input of 2.5 V. I think it will help suppress noise in the non inverting and make it require much more than just above 0. I will use a single op-amp to feed the whole circuit signal so I don't worry about synchronization, also it provides isolation for the MCU pins.

The signal will be around 12 V square wave it might be a bit less but I only need 10 V at the gate of the MOSFET. The signal will get fed to push pull circuit as a MOSFET driver. (PS, I couldn't find a suitable driver so I have to use this instead.) The push pull uses TIP122 and TIP126 as they can handle up to 5 A and I will need a maximum of 5 A for fast switching, however it wouldn't be a continuous 5 A it will just be short pulses so the average power loss is small and won't require cooling. I also used 2 kilohm to reduce the load on the op amp since it will be driving 20 push pulls and those transistors are Darlington with a gain of at least 1k.

Now we finaly get to the MOSFET it will be an N-chanel MOSFET IRF1404 which is the lowest RDSon I could find at 4 mΩ, I will use 20 in parallel with a shared heatsink to make sure they last long, however I think it isn't really necessary, each one will have around 18 A of current and that translates to 18^2 * 4 mΩ = ~1.3 W of conduction losses for each or 26 W for all of them, each one will have a temp rise of 80°C above ambient so even if I don't use a heat sink they can function in free air conditions. However there are also some switching losses so the heatsink should take care of that without me getting to carried away with the math, I also choose a resistor above the minimum of 2 Ω to prevent gate oscillation and allow good current flow, finally the 1 Ω resistor at the right is similar to load.

This circuit seems OK but I believe I'm missing something. I need decoupling and bypass caps but I'm cautious as I don't want to place a cap that can make the MOSFET turn on more than I need to.

Also I will add NTC to monitor heat and monitor current across all MOSFETs. I also will be using a regenerative braking using a different group of MOSFETs and a separate signal from the Pi Pico making sure I have some dead time or only allow the circuit to work if all MOSFETs at the main part are off.

I got into details so others can understand if they are attempting to make a similar circuit and also in case I was incorrect with some of my assumptions. I will build this circuit as soon as I feel confident it will work properly, and I will share the results here as well.

Edit: I built the circuit up to the push pull I didn't connect a MOSFET yet here are the results, the inverting input has 2 V from a 5k/1k voltage divider from 12 V source.

1- op amp output at 50% duty, 70 kHz

enter image description here

2- output of push pull at 50% duty, 70 kHz

enter image description here

3- push pull at higher duty cycle, 70 kHz seems to work well at higher duty.

enter image description here

4- push pull at lower duty ~40%, here some issues happen the low starts to rise sooner than expected.

enter image description here

5- push pull at very low duty ~10%, now it becomes a mess with another pulse coming from nowhere.

I don't know what causes this weird behavior. I will try a comparator and share the output.

  • 1
    \$\begingroup\$ @All - Although well-intentioned, the long comment chain has exceeded what is reasonable for comments. Therefore comments so far have been moved to chat & should be continued there (see link in the comment below). || As this bulk moving of comments to chat can only be done once per question, any further comments posted here may be deleted without notice. Keep it in chat now, please! || Any improvements to the question decided during the chat, should be made via a question edit, not as a comment. When someone has enough information to post an answer, please answer it as usual. TY \$\endgroup\$
    – SamGibson
    Jan 22 at 13:21
  • 1
    \$\begingroup\$ Comments have been moved to this chatroom and discussion should be continued there. Please do not continue the discussion in comments here. \$\endgroup\$
    – SamGibson
    Jan 22 at 13:22
  • \$\begingroup\$ thank you I will make sure to move comments to chat if they get big \$\endgroup\$
    – AhmedH2O
    Jan 22 at 14:13

1 Answer 1


If you are paralleling MOSFETs you need to think very carefully about how well current will be shared between each device, especially during switching transients. Unless you get that element right, you will find you will suffer cascading failure as one-by-one the device which switches on first takes all the load and burns out.

There are a number of good app notes on this topic, including this one and this one from Infineon.

There are also many single piece devices that would be more than capable of taking the full load, and these are likely to be much easier to integrate into a working design. A quick parametric search of digikey or the like will show these up (253 according to a search I've just made.).

For instance Infineon's IRL40SC is good for 478A continuous, and is just $4/piece.

  • 1
    \$\begingroup\$ I don't think a single device would be able to do so 350A is a huge current, if it did exist I would love to know about it, however probably I cant get my hands on it I live in egypt and importing is very hard. \$\endgroup\$
    – AhmedH2O
    Jan 22 at 14:26
  • \$\begingroup\$ @AhmedH2O Egypt has no semiconductor fabs so you are importing yours regardless. \$\endgroup\$
    – winny
    Jan 22 at 14:39
  • 1
    \$\begingroup\$ Even then it will take 2 monthes for delivery and another 2 for getting reviewed by authority to make sure it isn't going to be used illegaly, and even then you will pay a huge amount of duty maybe 3 times the price or more , so it is totally not worth it, the other importer knows how to deal with this either by bribing or getting a lawyer to pay the fair duty since they never follow the rules \$\endgroup\$
    – AhmedH2O
    Jan 22 at 15:06
  • 1
    \$\begingroup\$ @Jens That's still much easier than having to parallel 20 MOSFETs. When one of the two inevitably switches a little earlier,it won't have much trouble surviving the full load for a few nanoseconds. \$\endgroup\$
    – TooTea
    Jan 22 at 19:13
  • 2
    \$\begingroup\$ 360 A flowing in a pcb is not trivial. For the first tests I would prefer something with big screws like IXTN600N04T2 and a much lower PWM frequency. \$\endgroup\$
    – Jens
    Jan 22 at 21:02

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.