0
\$\begingroup\$

First time posting :)

I am a Mechanical Engineer from Sienci Labs and for the past 6 months we have been developing a 800-1000W BLDC spindle for hobbyist CNC machines as a souped up alternative to the Makita RT0701C router.

During that time, our motor suppliers (in China) suggested and provided us with sensor-less motor + driver solutions that worked to a degree. Yet after multiple rounds of tuning, the motors' dynamic performance still leaves something to be desired. Specifically motor speed dips / overshoots by more than 15% when step loads are applied / removed with sometimes second long settling time. This is quite a bit worse than the Makita tool we are trying to match.

BLDC Dynamic Performance when load is applied and removed

Makita Dynamic Performance when load is applied and removed

While our suppliers are still hard at work on the problem, the team would like to sanity check the following:

1. If we are approaching the limits of a sensorless motor?

2. If so, will moving to a sensored motor would help solve our problem? Or is it time to change motor types

Be great if someone senpai on the Internet with experience with both motor types can comment :)

Thanks in advance!

Johann

P.S. We are aware that hall sensors are mainly used for commutation and the chief benefit is low speed speed regulation, but we're just wondering if it also is a critical factor in our use case.

\$\endgroup\$
5
  • \$\begingroup\$ If you don't get answers here, perhaps find the part numbers of the relevant ICs, look up the manufacturer's data sheets, the relevant Application Notes, and try to speak to an applications engineer at the manufacturer. \$\endgroup\$ Commented Apr 24 at 22:32
  • \$\begingroup\$ @technophile documentation from a Chinese company? Good luck with good support from a Chinese company \$\endgroup\$
    – Voltage Spike
    Commented Apr 25 at 0:31
  • \$\begingroup\$ In general, you have hit on the major weak point of sensorless motors: their poor reaction to load step changes, so you are making a good guess. No amount of tuning will address this problem. \$\endgroup\$ Commented Apr 25 at 12:58
  • \$\begingroup\$ @VoltageSpike often the ICs were designed by a company that wants to sell them, and thus publishes the datasheets. The boards they're on are another issue entirely, along with the question of whether the boards are using real or counterfeit parts, out-of-spec parts, etc. \$\endgroup\$ Commented May 3 at 21:17
  • \$\begingroup\$ @VoltageSpike a thought is that during manufacture of a component that must pass through incoming inspection, some parts will not meet specs. Then what? E.g. suppose you had a large batch of LEDs that did not meet minimum brightness, max. forward voltage, molding dimensions, etc.? You could pay to dispose of them - or, you could sell them off-brand, or put them in modules and sell those, and make a few bucks instead. \$\endgroup\$ Commented May 3 at 22:58

1 Answer 1

1
\$\begingroup\$

A brushless motor controller needs the shaft's instantaneous angular position for best operation. They provide optimum performance by maintaining the relationship between the shaft position and the stator field. Knowing the shaft position allows the controller to operate a control loop to control the speed and torque with PWM and phase frequency.

A sensorless motor has an additional control loop to infer the shaft position, This control loop has a time constant. When the load or speed changes, an error in shaft position must occur for the control loop to react and compensate.

In lathe operations, the motor speed often increases at a rate as cuts are made to keep the surface speed constant, and of course the required torque changes as the part's diameter changes and also when the tool makes contact with the part. When running at constant speed and constant load, sensorless motors perform well, but for the varying load and speed operation of a turning center, they are a poor choice.

If brushless or PMSM motors are in a top-of-the line machine like those made by Makita, they will employ sensors that continuously report the shaft position to their controllers, allowing them to keep the speed loop tight. Hall effect devices are a second choice, only providing shaft position at specific points where the halls are employed so their speed regulation will not be as good, although still superior to a sensorless approach.

\$\endgroup\$
2
  • \$\begingroup\$ Thank you your earlier comment and the very detailed response here. I wish we hadn't have to learn this the hard way. If you don't mind some follow-up, I am also curious to know how dynamic response is characterized and tested in industry? The impression we got from our suppliers is that they don’t typically do such testing? Lastly, any chance you can comment on the amount of improvement in dynamic performance we can expect going from a sensorless setup to a hall effect / position sensors (presumably some sort of encoders)? Oh, and are there any other gotchas we should look out for? \$\endgroup\$ Commented Apr 26 at 15:42
  • \$\begingroup\$ @JohannChung perhaps you can get some sort of idea of the improvement in dynamic performance based on the shaft rotation angle between position sensor updates, which will also affect the maximum practical servo loop update rate. No point in attempting recalculation when there's no new data. \$\endgroup\$ Commented May 2 at 2:37

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.