I think I have a modest understanding of how an ESC for RC hobby crafts works, and how a BLDC motor works. I've been wondering about how to make an ESC that could run a motor with large current required like a motor for an ebike when the motor is not spinning fast like a quadcopter motor. Could a circuit like the one in https://www.instructables.com/id/Make-Your-Own-ESC/ be used in conjunction with an IGBT on each output? That way it could handle higher loads than the ESC could normally.

Edit: I've been trying to do some basic calculations on sizing a motor that would accomplish my desired acceleration, and top speed of an ebike, and I anticipate using a 72V 18S lithium battery. I think I would be looking at about 150A during peak acceleration to about 60 mph.

  • \$\begingroup\$ I picked IGBT as a very conservative guess because I saw some parts are capable of 400V and 600A, more than my application, so I would maybe need less cooling solution for the heat generated. Would a MOSFET, sufficiently large help accomplish the same task though? \$\endgroup\$
    – Jon Bot
    Commented Oct 21, 2019 at 18:39
  • 1
    \$\begingroup\$ Could you edit your question with specifics about the anticipated voltage and current draw? Folks kinda sorta know that eBikes run on 24-48V, but I'm not sure what current you're talking about. As soon as you give a definitive answer then if there are no surprises @DKNguyen can make his comment into an answer. \$\endgroup\$
    – TimWescott
    Commented Oct 21, 2019 at 18:41
  • \$\begingroup\$ I have a spreadsheet that I can post later today that shows some of my math I worked out to guesstimate what I would need for a motor and controller \$\endgroup\$
    – Jon Bot
    Commented Oct 21, 2019 at 18:44
  • \$\begingroup\$ If you're going to do this for production, remember that there's some regulatory standards revolving around voltage levels and safety precautions. If I remember correctly, the lowest threshold is 62VDC, but I can't say for sure (I just know that 48VDC is in, and 120VAC is out). If you're doing it for you, remember that those standards are there for a reason. \$\endgroup\$
    – TimWescott
    Commented Oct 21, 2019 at 18:56
  • \$\begingroup\$ I assume you are referring to electrocution hazard from higher voltage, I will be the only one planning to use this bike. I have been doing some reading about safe voltage levels, and 48V does seem like the safe limit for an ebike, but I have noticed that with 72V I can reduce my current to maintain same speeds. \$\endgroup\$
    – Jon Bot
    Commented Oct 21, 2019 at 19:01

2 Answers 2


You would not use an IGBT for this. IGBTs are more suited for very high voltage setups (>600-1000V) that switch quickly (but not as quickly as a MOSFET). For your purposes, you would just use bigger MOSFETs or parallel MOSFETs.

MOSFETs have a number of things going for them:

  • they are more forgiving to drive than IGBTs (IGBTs can behave eratically if you don't switch them off properly and sometimes require a negative gate drive voltage to do so which means more complexity)
  • they are easier to parallel than IGBTs
  • they switch faster than IGBTs

When IGBTs start coming into play is at higher voltages. To construct MOSFETs to withstand these higher voltages, tradeoffs have to be made that start to result in higher losses than if you used a fundamentally different approach (i.e. IGBTs). A low voltage MOSFET has less losses than a low voltage IGBT. As voltage is increased, there is a point when the IGBT has less losses than the MOSFET.

So at pretty much any reasonable eBike voltage and reasonable eBike current, you are better off using bigger MOSFETs or parallel MOSFETs. You might want to parallel MOSFETs even if singular, large MOSFETs are available just to spread the heat out over a larger area which makes it easier to manage. They are probably also more available too.

At increasing power levels, there is a point when the current become unmanageable for a given voltage. It could be that the wiring itself is unmanageable, or large MOSFETs are unmanageable, or parallel MOSFETs are unmanageable, or a number of other things. At this point, you increase the voltage so the current can be reduced to achieve the same power level and start using IGBTs instead.

enter image description here Taken from: "IGBT or MOSFET: Choose Wisely - Infineon Technologies" https://www.infineon.com/dgdl/Infineon-IGBT_or_MOSFET_Choose_Wisely-ART-v01_00-EN.pdf?fileId=5546d462533600a40153574048b73edc

  • \$\begingroup\$ That makes sense. I hadn't thought about running multiple MOSFETS in parallel. I suppose I could have the same wire connected to both Gates in order to turn on both MOS with one? \$\endgroup\$
    – Jon Bot
    Commented Oct 21, 2019 at 18:47
  • \$\begingroup\$ @JonBot Yes, but you need to make sure that gate drive loop is as small as possible (to reduce inductances and ringing which could cause damage) and also place a gate resistor or ferrite bead as close as possible to the gate of EACH MOSFET to dampen other current paths that could produce ringing. \$\endgroup\$
    – DKNguyen
    Commented Oct 21, 2019 at 18:54
  • \$\begingroup\$ @JonBot Remember, it is not always safe to parallel transistors. Some types of transistors do not inherently share the load current properly because the hotter a transistor is, the better it conducts which makes it take even a larger share of the load until it explodes. In these cases additional measures must be taken (the simplest of which is small load balancing resistors in series with each MOSFET) to cause the load current to be distributed equally. \$\endgroup\$
    – DKNguyen
    Commented Oct 21, 2019 at 18:57
  • \$\begingroup\$ I don't think I've ever heard of "ringing" before. Is that where I would have two lines with high frequency signals on them that are near each other, and one might induce a current in the other if they are too close? \$\endgroup\$
    – Jon Bot
    Commented Oct 21, 2019 at 18:58
  • \$\begingroup\$ @JonBot No. That is just induced noise. Ringing is a transmission line effect when the transition time of a signal is so fast compare to the length of the wire that you can no longer assume the electrical signal travels down the wire instantly. Each end of the wire can be at a different voltage. It's similar to pouring water down a pipe and having the water eventually splash back at you because the far end changes in diameter or is plugged. It takes time for the "info" about the far end of the pipe to work its way back to the inlet so that the water itself can adjust its own flow rate. \$\endgroup\$
    – DKNguyen
    Commented Oct 21, 2019 at 19:08

A quick look at the link you provided indicates that they have probably done quite well.
This is not certain without a deeper look but they are using

  • "Proper" sensorless control via the LM393's.

  • Current sense via U5.2 .

  • U2 driver - suited to motor used.

A higher power driver - or discrete drivers & FETs would allow a higher current motor to be used with no major conceptual changes. The current sense resistors would need changing.

They use U5.2 to provide an interrupt drive on/off current sense, with level set by R15, which allows switching at a set level but no "power control" as eg load varies on hills etc at less than full current. It would be "easy enough" to add analog input current measurement as well as interrupt controlled input if more 'nuanced' control was wanted.

This is their circuit:

enter image description here

From here


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.