I heard from someone that axial gap /pancake motors (lynch or etek) are better for servo control (speed and position) than standard radial flux DC motors. Both motors are brushed type.

I don't really understand the physics behind this claim , although I know how each one operate.

So can anyone please state if it is really better, equal or worse motor technology for servo drive, and why?

  • 1
    \$\begingroup\$ I would imagine this would relate to the physical size of the coils and distance from the center of the axis of rotation. Because position precision with a given field would be a percentage of the physical size of the field, this makes sense to me. They may also have been looking especially from the perspective of hobby motors/generators, and especially if you don't have factory precision available, a precise pancake motor is easier to build than a precise tubeaxial one. \$\endgroup\$ – K H Oct 4 '18 at 1:57
  • \$\begingroup\$ so a pancake motor is better ? \$\endgroup\$ – ElectronS Oct 4 '18 at 9:26
  • 1
    \$\begingroup\$ If I was certain enough to write it as an answer I would. It kind of matters in what way he was saying they were better in the first place. There are likely advantages to both motor types. \$\endgroup\$ – K H Oct 4 '18 at 9:31

One advantage i can think of is that the low inductance of axial-flux motors means that they have low electrical time constants, allowing current to flow very quickly into the armature for virtually instant torque production , which should translate to fast acceleration and deceleration which maybe desirable for high performance servo control..


However they are not widely used probably because of their higher price and manufacturing challenges explained in magnax motors white paper : WP- High Efficiency Axial Flux Machines - whitepaper v1.7

The pancake motor also might suffer from several problems:

1-High interia due to their shape which might not be desirable for rapid movments.

enter image description here

more details in :motor-sizing-calculations (source:orientalmotor.com)

2-low inductance cause problems in torque loop:

motor-inductance-effects-on-servo-drives (source is doc.ingeniamc.com)

Quote1: "Motor inductance, or more appropriately electrical time constant value affect the servo drives in many ways. While high inductance values may limit the system bandwidth, low inductance values can lead to control loop instabilities, inaccuracies in current readings, increased power losses and other problems. These issues are especially critical in high speed brushed motors with very low friction and fast dynamics.The problem is more notorious with brushed DC motors"

Since The current ripple can be expressed as : enter image description here

Quote 2: "At low loads, positive motor currents can be read as negative due to current ripple. This leads to unstable current loops that become uncontrollable!"

enter image description here

| improve this answer | |
  • \$\begingroup\$ really good details and a lot of it makes since , so there is no direct answer , it is better in some ways and worse in other ways \$\endgroup\$ – ElectronS Oct 6 '18 at 21:06
  • \$\begingroup\$ I am not accepting answer YET inorder to give more time if others want to add or share more details \$\endgroup\$ – ElectronS Oct 7 '18 at 20:45
  • \$\begingroup\$ I don't find any clue that axial motors have lower inductance than radial one. It's just about design if you want a low voltage, high current->low inductance or high voltage, low current ->high inductance, for both radial or axial. \$\endgroup\$ – Marko Buršič Oct 10 '18 at 12:00

If you want to do dynamic positioning, then you must take into account the load vs. motor inertia. With a perfect match, the load and motor inertia are almost equal.

Note that a pancake motor has higher torque but lower nominal speed as radial motor. It has also much higher moment of inertia.

A radial motor is often used with combination of gearbox, so you can match load and motor inertia. The transormed load inertia is then J'=J_load/p^2 where p is the gear ratio. So you can have high dynamic system using a low inertia motor with high reduction ratio gearbox.

A pancake motor meanwhile is more suited for applications where gearbox is not needed, but you need a high torque and low speed. These applicationa are typically direct drive, gimbal, ...high inertia load. It has also a possibility to have a hollow shaft, where you can put slip rings to supply sensors, other devices mounted on gimbal.

Now, what importance has the load inertia and what does it do?

enter image description here

Every mechanical setup has its own elasticity, like torsional elasticity. Having numerous axes, gears,...means that a load is spring like attached to the motor. We call this torsion spring. Now we have a rotor that is coupled by this torsional spring to the load. This system would have a resonance and an anti-resonance frequency. If the load and rotor have the same inertia, then those two frequencies are identical.

If inertias are mismatched, then you get a lower resonance frequency, which means it is harder to filter out, the whole system has to be tuned to a slower dynamics to stay away from this resonance which has to be damped.

enter image description here

Reference link

enter image description here Another resonance link

Let we have a radial motor with mounted encoder or brake, the rotor inertia now becomes the sum of all inertias: rotor + encoder + brake. Why? Becuase they are stiff mounted on rotor and no elasticity is between them. So now we have altered the motor rotor inertia. Anything on rotor, that is rigidly coupled becomes a part of rotor, not a load.

With a pancake motor, we could say that the mechanical setup is almost perfect, if the rotor is stiff mounted with a load. Therefore you get only one main natural resonance frequency that will determine the maximal dynamics of the system. This is one main advantage if you want a high dynamic system.


A closed loop system like servo , prefereably has to have low response time aka high dynamics. This can be acheived with increasing the overall gain - loop gain, by increasing controller proportional gain. Now, if you look at the Bode plot, you can notice that at node, the system has a considerable gain and yet more important, it has a phase shift of -180 degrees. That means, the system will begin to oscillate exacty at the resonance frequency. So, when tuning the servo, the gain has to be such that there is always a safe margin, to remain stable. As the node gain is smaller and it has higher frequency, the system can have more gain and thus have more dynamic response. The last way to improve the dynamics is to add low pass filter, notch filter,...If the rotor is stiffly coupled to a load, like for example a gimbal, we get a neat high resonance frequency that can be elliminated by inserting a notch filter.

Therefore, yes the inertia is very important for closed loop servo sytem due to these resonance nodes.

enter image description here

Refeerence - Siemens S120

| improve this answer | |
  • \$\begingroup\$ Good point you brought up regarding the load/motor inertia matching , i remember in Yaskawa servo drives you can choose a motor from 1:1 to 1:5 inertia with respect the load . So this must be a really important aspect of the servo system \$\endgroup\$ – ElectronS Oct 11 '18 at 13:52
  • \$\begingroup\$ what does happen at the lower resonance frequency does it cause mechanical vibration ? or does it make control difficult ? and how do you find this resonance frequency in reality not theroy ? \$\endgroup\$ – ElectronS Oct 11 '18 at 13:56
  • \$\begingroup\$ I have edited the answer \$\endgroup\$ – Marko Buršič Oct 11 '18 at 18:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.