InstaSPIN for torque control of linear actuator

Hi, David

From the link it's not readily apparent the specifications of this motor

Voltage, current, speed range?

What is the motion profile this will be used for? I see many linear systems used for very low speed control. If this is the case I would say that for now you would want to stick to a control system that uses some sort of position sensor for feedback.  If the motor will be run at moderately higher speeds and you are only controlling torque and velocity then InstaSPIN-FOC could be used.

If these motors have a position feedback sensor then you may look at using InstaSPIN-MOTION. There is currently a lab showing velocity (using a rotor sensor) + torque control , and in a couple weeks we'll be adding Position control examples as well.

Hi Chris,

Thanks for the quick response.

The PSM12-C3 is a 2.97A, 41.2W continuous, 14.85A, 1030W peak motor (although we don't plan on operating at peak). Coil resistance is 4.5Ohms and max terminal Voltage is 100V although we will operate at closer to 50V. (After reading my question it guided you the wrong way, I am not so interested in the dev board specs against our motor, it is more about Insta SPIN. If the approach works we will design our own.)  Lets say about 0.5m/s although I don't really know as we don't measure velocity.

For our operation there isn't really a concept of position or velocity control at least not in a tight loop.  We really use torque control around the joints, hence current control. (When the leg is in flight phase we just let a spring return it to normal position (ie current = 0), and during stance phase we push at max continuous current in the direction of expansion.) See here for our small test leg http://www.youtube.com/watch?v=j_4blE9DpBc using a hall effect sensor BLDC linear motor however the series doesn't scale up in power.

So I am not so concerned with accurate position control at low or high speeds. (The only reason I am concerned with position is how well your InstaSPIN will track phases for timing the phase control.) I am very interested in how good you think torque control will be at low speeds, in particular at 'startup' as the motor will constantly be 'starting' from a movement point of view? Although note that current control will be continuous.

From reading online there are startup issues with sensorless control of BLDC motors. (We haven't been able to find any BLDC linear motors with hall effect sensors that meet our requirements.) I am wondering how good InstaSPIN will be to solve these problems. Upon coming across the IT InstaSPIN range it seems at face value to solve all of our problems. Perhaps another way to ask is for you to compare a sensorless BLDC + InstaSPIN versus a hall effect BLDC and a regular current controller for our application.  

Regarding motor identification, my concern was that it might have a fixed process that requires spinning a rotary motor which won't equate well to a linear motor as there is only a short travel distance from end to end (eg 100mm)

Regards,

David Ball

David,

I'm thinking that InstaSPIN-FOC by itself today may not get you the performance you are looking for.  I really haven't used linear motors so I'm not entirely clear on the exact use, but you are correct that most "sensorless" algorithms have start-up issues because they rely on measurements that only occur once the motor is moving.

In this case the FAST algorithm we use has similar limitations, but not the same limitations.

FAST does require phase voltage Bemf measurements for proper estimation, so the motor must be moving.  
Unlike other estimators though FAST very quickly converges in under 1 electrical cycle of rotation, most other techniques require speeds in the range of 20 Hz. 

All of these estimators require a way to start the motor turning to reach the necessary speeds. With most estimators this requires an extended time in "open loop" control, which makes system stability a real challenge, especially under load or with a dynamic load.  Because FAST can take over so much more quickly from the open loop (what we call ForceAngle) the InstaSPIN-FOC control system performs much better.

Regarding Full Torque, this is possible with the InstaSPIN-FOC system, though it is still challenging.  To produce full torque you still need to orient your stator flux precisely with the rotor flux (which is unknown).  Even Hall based start-up can't guarantee FULL torque, but they can usually get within 15-30 electrical degrees to produce near full torque.

With InstaSPIN-FOC there are several start-up options to help achieve a full load start-up. 

Here is a post on the topic.

Yes, the motor ID process typically requires that the motor spin unloaded for up to 2 minutes to identify the flux and inductance. There are ways around this if the motor can't spin (like a sealed compressor) but I'm still not sure how this would work on your linear motor.