# How to select BLDC motor for high torque applications

For electric vehicle I need to choose BLDC motor. Rated power around 2 to 3kW 48 nominal voltage. I don't need high speed, but would like to have high torque at low rotation speeds. There will be reducing gearbox installed on top of motor 1:10. On what parameters to look when choosing motor? For example, is motor with lower RPM per V "slower" and thus more powerful, or does it means just its efficiency is worse?

• Efficiency will be largely independent of Kv. Less RPM/volt = more torque/amp in a reasonably fair exchange.
– user16324
Apr 27, 2021 at 19:41
• Consider torque motor.
– user76844
Apr 27, 2021 at 20:45
• I would look into high rotor-pole switched reluctance motors (HR-SRM) using VFR drivers for broad torque speed curves that may not require a transmission Apr 27, 2021 at 20:48
• "I don't need high speed," - what is the maximum speed you want, and what diameter are the wheels? Apr 28, 2021 at 6:04
• @BruceAbbott Hi wheels will be 13'' diameter, max speed needed is about 20km/h. What torque is needed: roughly car with 4 wheel drive should be able to climb 30 degress hill with total weight 300 kg Apr 28, 2021 at 12:20

In the comments you specify a top speed of 20 km/h, all-up weight of 300 kg and ability to climb a 30° slope. I put your figures (plus some estimates for drag and rolling resistance) into the online calculator at Built-For-Fun EVs, and it returned the following:-

Drive train output power required for 20 km/h on flat ground: 1913 watts.

Output power required to climb 30° slope at 5 km/h: 2459 watts.

This is within your motor power range of 2 to 3 kW.

Next you need to determine the Kv required for each scenario. First calculate motor rpm from wheel diameter and gearbox ratio, then divide rpm by battery voltage to get Kv, and finally factor in the loaded speed reduction (typically 10~20% at rated output power).

Immediately you will see that there is a problem - the Kv required for 20 km/h is much higher than it must be to climb at 30° while staying within the rated motor power. That means you either need a 2 speed gearbox, or a motor rated for more than 3 kW.

As a motor designer, I am going to give you a dimensional approach to selecting a motor. Torque is proportional to the rotor radius squared. Torque is linearly proportional to length. $$$$\tau \propto r^2l$$$$

What does this mean?? It means that you will want a motor that looks more like a pancake than a hotdog. I suggest you look at axial flux motors (YASA, MAGNAX are examples of companies that make these motors).

• Doesn't efficiency tend to suffer as you move from hotdog to pancake, though? Apr 28, 2021 at 4:59
• That is a difficult question to answer. You really have to compare efficiency maps of motors to determine performance differences. Regarding pancake motors I can provide some examples on how they improve efficiency. I know Magnax has a somewhat ingenious design where their stator pole-pieces are wound as solenoids (then grouped together to construct the stator). This is done to reduce end turns, which increases efficiency. I believe that Magnax also uses grain oriented electrical steel in their pole-pieces (typically only used in transformers) which lowers core loss in pole pieces. Apr 28, 2021 at 5:17
• Is not it that they (Magnax) have not yet started mass production? And their motor seems rather big for my vehicle. Apr 28, 2021 at 12:22
• I agree. Those example motors are way too big for your application, I was intending that they would serve as an example for the pancake motor geometry. I would suggest you look up "Frameless Motors". These motors need to be integrated into a housing but will give you the most flexibility/options when searching for a motor that meets your specifications. You can apply the dimensional technique I started above when comparing different motors. Regarding performance metrics. I would start by identifying your maximum torque requirements over the speed range you need maximum torque. Apr 28, 2021 at 19:09

There is a motor efficiency factor and a transmission (gearbox) efficiency factor. You choose the motor and gearbox, so that fulfills the constraints about torque/max. speed and efficiency. There are also many other parameters to consider:

• cooling capability
• mechanical forces on motor/gearbox. The transmission of type belt/pulley exhibits large radial force on the driving shaft, so you have to pay attention to not exceed those constraints.

A complete motor drive consist of a motor and transmission. You have to combine both together, so not only the motor is questionable, rather a combination.

This is a nontrivial problem. There are a large number of metrics to evaluate across, and to provide a generic answer to this question is impossible. However, there are some tricks to teasing out motor parameters from the wide variety of available brushless DC motors on the market with sparse specifications.

A lower RPM per V motor has a higher torque constant (see my answer to this question for how to precisely calculate the torque constant). A motor with a higher torque constant will produce more torque per amp. That doesn't necessarily mean it's stronger though; you'll need to factor in the voltage rating/current rating/winding resistance and the amount of heat you can siphon away from the stator. For instance this motor has a 140Kv rating, but the high winding resistance (required to get a motor this size to have a Kv rating this low) means you won't be able to operate it at a high current without burning it out. This motor has exactly the same torque constant, but is larger, has lower resistance windings, and is therefore capable of producing significantly more torque.

Efficiency will depend on motor construction and the controller you use, and will not be apparent in any ratings you might find from an online datasheet.

You'll want to look into sensorization options as well. To achieve efficient control with high torque at zero speed for a BLDC motor (and to do things like torque control and field weakening that appear in some of the other answers) you will need a controller that can get an accurate estimation of the relative angular position of the rotor with respect to the stator. For most practical applications this will require a continuous position sensor (usually hall effect or TMR). Many sensorized motors will have 3 discrete/binary hall sensors, which may not provide resolution you'd need for advanced motor control techniques, so if this matters you should be discerning about it. It's worth noting that there are some techniques in research that involve sensorless methods for rotor position estimation, but they're extremely difficult to implement and require meticulous fitting to your particular motor.

• It is possible to provide vector control with no position sensors at all. Startup can be challenging, as you say. But even starting can be done without position sensors given certain assumptions. Apr 28, 2021 at 5:17
• This is true, but the performance will vary wildly depending on the technique used. I.e. bemf based is simple, but it means low-zero speed performance is dismal due to the poor angle estimation (making it useless for most robotics applications). More advanced techniques (such as the one in the paper I linked that uses rotor saliency) can provide good low-zero speed rotor position estimation, but these techniques involve significantly more design effort and less overall generalizability than simply mounting your own sensor to a motor. Apr 28, 2021 at 17:52
• I am of the opinion that if you want to use one of those low-zero speed sensorless techniques, the 'sensorless' aspect should be part of the point, and not a means to the end of just getting a working motor controller (for applications that require low speed performance, such as a car or a robot) Apr 28, 2021 at 17:54
• With a robot, if precision locomotion is needed, I think closed loop control with a shaft encoder would be needed anyway. But with a vehicle, especially scooters, bikes and skateboards, it may not be necessary to have good starting ability (because the user pushes to start). And sensor-less can work well. Also drones. Full torque is not needed at startup for propellers, and it is OK if the propeller jerks slightly backwards prior to achieving forward rotation. It is a key design decision that you need to make early on. I agree with that. Apr 28, 2021 at 18:18

RPM per volt is a specification that describes the motor in DC commutator motor terms. A motor and controller combination that is designed to emulate DC commutator performance will probably not provide what you need. A system like that will provide constant torque capability over the speed range.

With an engine-driven vehicle, the transmission gear change provides increased torque at low speeds with lower gear ratios. Electric motors can provide rated torque down to zero speed, but a motor with a higher than necessary power rating would likely necessary to provide the torque that you would like to have at low speed. For that reason, drive systems are often designed for constant torque up to some speed and constant power (decreasing torque) above some speed. That generally requires a permanent-magnet AC synchronous motor and an inverter-based controller that controls the frequency, voltage and current based on dynamically measured motor performance.

Motor performance with the most sophisticated motor control techniques is still limited by motor characteristics. There are a number of motor designs that can extend the constant power speed range. Tesla's permanent magnet / hysteresis hybrid is one example.

The simple way is to figure out the maximum RPM you need to hit top speed. Then divide by battery voltage. That is the Kv you need (approximately...). Kv * Vbat gives you a rough estimate of how fast your motor will spin.

Then figure out the maximum torque you need for worst case hill-climbing and vehicle loading. Convert torque to Nm. Convert max speed from rpm to radians/second. Multiply torque (Nm) x max speed (rad/sec). That is the motor power you need (in watts).

When you do this you will discover that the motor power is quite large and the motor is too big and too expensive.

So then the tradeoffs begin. Maybe you can use a lower torque motor, but overdrive it at low speed for a short time to avoid overheating. Maybe you can use a low Kv motor, but then use a field-weakening controller to achieve top speed anyway.

There are a lot of tradeoffs.

Here is how id look at it.

First, figure out your form-factor. The same number of amps and neodymium does double the torque at double the distance. So make the most of that.

Somewhat confusingly; the number of poles does not at all matter to maximum torque or gap shear stress. But it does matter, in the sense that smaller poles have smaller flux paths, which require less neodymium and copper to get that given amount of gap shear stress. So if you care at all about torque/kg, you want small poles. Pole count inversely relates to rpm. So assuming a fixed weight budget, poles do trade rpm for torque.

So if you want a lot of torque, and you have a finite weight budget, you want many small poles. Your air gap does not scale down along with your poles though, and there are manufacturing considerations, limiting how small you can really make them. But if rpm limits are not yet in sight, more poles is better.

Then there is the question; how much copper do I intend to pack in my slots? What is a reasonable current density, and rate of coil temp rise, for your application?

This is very application dependent. For a stationary efficiency focussed application like a powerplant, you will see lots and lots of copper, with very low current densities. But this will determine your slot depth and overall stator weight. Then you need to match that with the thickness and grade of your magnets, so they will be able to withstand the stator amp-turns, without demagnetizing.

With regards to amp-turns and magnet depth, its a simple more-is-more game, as it relates to torque. Though it needs to be balanced; no point in packing copper that will just destroy your magnets, or packing neodymium thats barely being 'worked'; just doubling your magnet thickness relative to an optimum will barely get you any additional torque, since the magnets are close to their flux saturation at their optimal working point anyway.

Scaling your motor thickness radially (at equal gap radius), or making your motor longer axially, are both mostly linear operations in torque. Actual practical magnet geometric and winding considerations dictate some reasonable range of values for radial thickness; if you want to scale everything linearly, its usually easier to scale axially if you can, and keep the radial direction looking 'reasonable'.

As for the thickness and number of turns of your wire, you need to match them to your drive electronics; often the reason you see torque curves level off at low rpm is either the driving voltage not being able to squeeze out more amps, given the phase resistance. Or it can be a demagnetization limit; where a motor optimized for high-rpm and modest torque does not pack the neodymium required to push these high amps safely. Most of the time its a 10s thermal limit though, as in 'this level of current will cook the enamel in 10s'.

So its not a simple story, there are many different limits that can kick in, and its a joint optimization process to make optimal use of all your materials. But hopefully the scaling laws sketched out above are of use to someone; I notice a lot of conflicting or incomplete information about it out there.