[Mariss Freimanis]

Please see PID_servo.gif and G320_PID_servo.gif for the results.

This is it. A step motor running as a PID servo and doing a darn good job of it too.

The PID_servo.gif yellow trace is the position error summing node. That is the difference between the command location and the motor's actual location. The scale is 14 degrees per volt, center line equals zero position error. The motor is commanded to turn 600RPM and the direction is reversed every 100mS.

The stepper servomotor develops a peak 15 degree following error on the direction change, overshoots by 2 degrees before settling to zero error in 28 mS. All this before I've even tried adjusting the PID coefficients.

For comparison, I added a G320_PID_servo.gif to contrast a DC servomotor under the identical conditions. The scale 9.2 degrees per volt, centerline equals zero position error. The DC servo develops a peak following error of 23 degrees, overshoots by 1.8 degrees before settling to a 1 degree following error in 45 mS. The following error is eventually taken up by the integral term but not in the 100 mS before direction changes again.

The step motor servo trace looks much better (even before PID optimization) by comparison. As a servomotor, the stepper accelerated more quickly and settled far more rapidly to zero error than the DC servomotor. This is exciting.

The servo control algorithms were developed in assembly language on a Rabbit RCM3720 and it is what's running the step servo in real time. These algorithms will be translated in VHDL and will go into a CPLD. This will bring a large increase in performance because execution will not be limited by the constraints imposed by a sequential machine like a microprocessor.

The final PID step motor servo control algorithm is unlike anything I had ever imagined at the beginning of this project. It is also unlike any other servo control algorithm I've ever dealt with before. I simply followed what the motor needed and this is the result.

The step motor is an MCG IH23014 3A/phase NEMA-23 run at 24VDC, the DC servomotor is an DGM D860-30B7 NEMA-34 24VDC rated 60W servomotor. Both motors are roughly comparable in power output at 24VDC.



Mariss

[Oldmanandhistoy]

Mariss,

I won’t pretend to understand most of that but note your excitement and satisfaction. I would just like to wish you luck with further developments and hope they aren’t too expensive so I can buy some.

Regards,
John

[Mariss Freimanis]

Then this should get you excited too. It probably won't cost much more than a G203V.

Mariss

[ger21]

Mariss, will this just be a standalone stepper drive, or will the G100 be required?

[Rube]

Then this should get you excited too. It probably won't cost much more than a G203V.

Mariss

OK...even though I just bought a set (3 each) of G203V and G320's...now I want some of these And silly me thought this hobby would be cheaper than my other ones...HAHAHAHAHAHA.

[Mariss Freimanis]

It will be stand-alone servodrive. It will make your motors two to three times as powerful compared to open-loop step motor drives. It also will have a % of available torque being used signal. This will go back to the G100 to signal it to decrease the command velocity when torque load nears 100%. That combination will make up the "unstallable step motor" project.

Mariss

[kreutz]

Congratulations!!

I have been thinking for a while about doing something similar with an inner PI torque loop (error signal = phase difference between rotor position and coil current) acting upon the peak current reference voltage, and an outer velocity PID loop (error signal = desired position vs current position), using a high ppr quadrature encoder.

As usual, you are almost finished while I am still sitting on the starting line...

Regards,

Kreutz.

[turmite]

It will be stand-alone servodrive. It will make your motors two to three times as powerful compared to open-loop step motor drives. It also will have a % of available torque being used signal. This will go back to the G100 to signal it to decrease the command velocity when torque load nears 100%. That combination will make up the "unstallable step motor" project.

Mariss

Estimated time of release?

Mike

[Mariss Freimanis]

Kreutz,

Your comment is especially appreciated since it comes from a peer. I have been at this for 3 years, 2 of which were wasted chasing a method I ultimately concluded could never work.

I cannot go into how it works because we are applying for a patent. The method is an unusual mix of analog and digital circuitry. Release of a commercial product is at least 6 months away, governed by refining the method and securing legal protections for it.

Mariss

[jahonen]

Kreuz,

I have been thinking about using a field oriented (vector)control method which I believe should do the job. That however requires a DSC, but that is not a problem nowadays. For example dsPIC30/33F, TI TMS320F28xx or Freescale 5683xx-series MSC would be a good choice.

The basic idea (at least how I understand it, and I'm not a motor drive expert) is to measure the motor phase currents using a simultaneous sampling ADC of the MSC and then do a a,b->d,q-transformation (we'll rotate the current phasor so that PID controllers see DC-values) to control q (quadrature, torque producing) and d (direct, field producing) components (which is set to zero). The angle required to do this transformation comes from the encoder attached to the rotor. PID-controller outputs (voltage phasor) is then inversely transformed to a rotating a,b-frame and so on. This regulation loop is best performed at each PWM period. The torque reference comes from the velocity error loop output, which in turn is controlled by the position error loop. This way, since the voltage phasor always tracks the rotor angle, we should not loose any steps, since the voltage phasor never looses sync with the rotor as long the encoder keeps track of the actual position.

Regards,
Janne

[bunalmis]

@jahonen

Yes, Field oriented control best method for electric machine.

8 wire stepper may good choice for sensorless closed loop applications.
2 half phase coil for excitier, other 2 half coil for positions feadback.

This is only idea.

[kreutz]

Kreuz,

I have been thinking about using a field oriented (vector)control method which I believe should do the job. That however requires a DSC, but that is not a problem nowadays. For example dsPIC30/33F, TI TMS320F28xx or Freescale 5683xx-series MSC would be a good choice.

The basic idea (at least how I understand it, and I'm not a motor drive expert) is to measure the motor phase currents using a simultaneous sampling ADC of the MSC and then do a a,b->d,q-transformation (we'll rotate the current phasor so that PID controllers see DC-values) to control q (quadrature, torque producing) and d (direct, field producing) components (which is set to zero). The angle required to do this transformation comes from the encoder attached to the rotor. PID-controller outputs (voltage phasor) is then inversely transformed to a rotating a,b-frame and so on. This regulation loop is best performed at each PWM period. The torque reference comes from the velocity error loop output, which in turn is controlled by the position error loop. This way, since the voltage phasor always tracks the rotor angle, we should not loose any steps, since the voltage phasor never looses sync with the rotor as long the encoder keeps track of the actual position.

Regards,
Janne

Yes, Field Oriented control is probably a good way to get it, DSPs are recommended to do the transformations on the fly, otherwise you will need big tables and lots of RAM, Although there is a lot of recent noise about going sensorless, I think that a sensorless approach is adequate for speed control, but still consider that positioning applications require a high resolution encoder for feedback.

Regards,

Kreutz.

[kreutz]

Kreutz,

Your comment is especially appreciated since it comes from a peer. I have been at this for 3 years, 2 of which were wasted chasing a method I ultimately concluded could never work.

I cannot go into how it works because we are applying for a patent. The method is an unusual mix of analog and digital circuitry. Release of a commercial product is at least 6 months away, governed by refining the method and securing legal protections for it.

Mariss

Mariss,

I respect your silence, it is required in order to get the patent. Take your time for the patent's documentation, be very creative, and generalize your claims as much as possible. Be also creative on the application fields. Congratulations again.

Regards,

Kreutz.

[307startup]

Will this new drive have any/all of the "enhanced" (or Vampire) features that the G203V has? If so...this would be perfect for a non-electronic inclined person such as myself to integrate into their project. Me, me, me...me, mine, me, mine...yes, I am a selfish sot, always thinking of what's good for me. Of course if it benefits others as well, it's beer & skittles.

[307startup]

One more thing...will I still be able to micro-step with this new drive?

[Mariss Freimanis]

To make full use of the feedback device (encoder), the drive must have resolution (not accuracy) in excess of the encoder. I intend to keep the drive's native resolution at 10 microsteps for a number of reasons including having the servodrive revert to an open-loop microstep motor if the user chooses.

I also plan to have it work with 1,000 line encoders (4,000 counts per rev). So I'm using an interesting technique I'm calling sub-microstepping to get the needed resolution from a 10 microstep drive. In essence each microstep is further divided into interpolated sub-microsteps to get additional resolution.

Mariss

[DonW]

Um, ok, you all seem mighty enthused about this, and I think I can see why. Sort of... except that I'm electronically challenged, unlike the rest of y'all. So please humor me if you will...

Does this mean that the four (4) 203V's I just bought on sale and haven't used yet are going to be obsolete, or something I bought that I will wish I hadn't?

Does this mean that I would be able to get servo-like continuous torque from steppers, and would these new drives require the specific motors you built this around? What is the Oz/In rating on them if this is so?

Will they need to be run at high speeds, and can I assume they will need to be geared down like servos? How will running them at high speeds affect their life? Won't this require running them pretty hot?

Sorry for the elementary level questions, just trying to learn and understand.

Thanks

[MrWild]

Does this mean that the four (4) 203V's I just bought on sale and haven't used yet are going to be obsolete, or something I bought that I will wish I hadn't?

Does this mean that I would be able to get servo-like continuous torque from steppers, and would these new drives require the specific motors you built this around? What is the Oz/In rating on them if this is so?

Thanks

I think this means Mariss is going to set the entire industry on it's ear. This is so new and full of WOW factor, he may be able to hang up his shop coat and move to an island of white sand beaches, beautiful girls, and those drinks that sport umbrellas.

Congratulations Mariss.

[DonW]

Gee, I hope not! Mariss is too valuable a resource for folks!
If I'm grasping things, then it sems this inded is pretty revolutionary, and a major breakthrough in stepper motor driving. Now who was that guy who said it couldn't be done in that other thread?

Congrats Mariss, I hope it's all ou say and more.

[Mariss Freimanis]

Contrasted with other servomotor types, a step motor servo remains a high-torque, low-RPM motor. Contrasted the same way, the power output of a stepper servo is now on par with other motor types. This is a good thing.

Motors have to drive loads and it's the motor's output Watts that drives the load. Power is torque times RPM. A step motor is a 50-pole motor. The rotating magnetic field that drives all motors has 50 simultaneous pull-points on a stepper's rotor vs. 6 for a BLDC servomotor. That is why a stepper has so much more torque than other motors.

Loads usually turn at lower RPMs than motors. Steppers develop full power at lower RPMs than other motor types. The advantage goes to servoed steppers because less or no reduction gearing is needed, simplifying the mechanism.

Mariss