Smooth motion at low rpm with VFD 7

[nadica]

Hi All,

I would like to pick your brains if I may.

I try to run a motor with FVD at very low speed, even just with a few Hz. I know this is not recommended and could lead to problems but ignoring that... I observed that the motor is not turning in a smooth manner rather like the second hand of a watch, stops momentarily and jumps, oscillating if that is the ride world.
I tested with two different VFD and the one with more sophisticated control (vector mode?) was better but what else I could do to improve the smoothness? The current motor is a 4-pole one. Does the number of poles influence this?

Thanks in advance!

 

[BrianPetersen]

Run the motor at a higher speed and couple it to your low-speed load using a reduction gearbox.

A motor with a greater number of poles would have a similar effect.

 

[Drivesrock]

Hi nadica.

Simple answer is, I think - To run low speed you will need an encoder with large number of pulses per rev fitted to the motor and the drive in closed-loop vector control. You might try a little voltage boost (fixed or variable depending on what your drive has to offer) but I don't think you will be 100% successful.

At low Hz the applied motor volts is also very low and now the small voltage drops in the cables and stator are a much larger proportion of the volts applied so the motor flux is compromised. With an encoder on the motor shaft the drive 'knows' exactly where it is in relation to the applied stator frequency and will/should adjust it's output to achieve the required speed and 'smoothness'.

This ignores any other 'problems'.

 

[itsmoked]

Motors are discrete devices. They have a couple of poles which are the sources of the torque or twisting effort. Because of that, they do not have a smooth continuous way of moving degree by degree. As they speed up the effort becomes smooth via the momentum of the spinning rotor. You are essentially trying to use a hammer as a screwdriver. It can be done but you must spend a lot of money to misapply everything effectively.

As was suggested above using a gear ratio gets you back to place where the motor can work effectively. You have gear boxes or belt drives to help you with this.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[jraef]

Although I completely agree that the best route to smooth operation is to gear down a faster motor, I'll also toss in here that not all "vector control" algorithms are created equal. There are more than a few bottom feeder VFD mfrs who use the term "vector control" in a manner "unbecoming an engineer". For Vector Control to properly function, the VFD must measure and create a mathematical model of the motor equivalent circuit. This can be a data set you can enter manually, but honestly, trying to get all the necessary info from a commercial-off-the-shelf motor supplier is like trying to get a government official to remember a campaign promise. So most good VFD mfrs provide whats called an 'Auto-tune" feature (or called something similar) in which the VFD does a test run of the motor to gather the necessary info. Some only do a "static tune" which means the motor is energized, but not spun, but the best data set will come from a rotating auto-tune procedure, which generally means disconnecting the load to do it. How you can tell when a bottom feeder is selling you a bill of goods on what they want to call a "Vector" drive is to read the manual and discover that there is NO auto-tune procedure at all and no way to enter motor data at all, so they can only be using a set of nebulous average data points. In my opinion, that belies the entire concept of a true vector control algorithm, but unfortunately there are no "spec police" keeping them honest, and people buy them because they are cheap. So before you judge all vector control drives as being incapable of smooth low speed operation, read your manuals carefully. If there is no auto-tune option, there is no true vector control. If they only offer a static tune, it could do better, but they chose not to tell you that. If they DO give you an option for a rotating tune, try that if you used the static tuning, it's better. If you DID in fact do a rotating auto-tune and you still don't like the outcome, but don't want to put an encoder on your motor, then use the gear box.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[nadica]

Many many thanks to all of you for your advise!

Jraef pointed out that not all VFD Vectrol Controls are equal. What about motors, could be a difference between motors running from identical VFDs?

I definitely will follow Jraef’s advise as I have to replace the VFD so I will look into more what I am going to get. The current one does static auto-tune.

Closed-loop control, Drivesrock, I will go down that route later but first I will see what is the best I can get out from a VFD with rotating auto-tune.

Gearing – it is partly done 1/9, couldn’t do more because there is a high rpm requirement, too. But the current motor is 4-pole, maybe use a 2-p motor and 1/18 ratio?

 

[mikekilroy]

Different motors will typically respond the same and the main factor will be the vfd capability.

Your research on different model vfds should include comparing min sensorless vector speed control speed or hertz. For instance, Hitachi says 0.5hz min control, which is typically much lower than most cheap ones. Your 'couple hertz' speed on the 1800rpm motor is only 60rpm. Many sensorless vector vfds will not be able to give any decent speed regulation here; this is about the slip of the motor itself and the 'fake' algorithms Jraef mentions, used by many low cost units, cannot operate here.

If changes to hardware is the future, I would definitely go to a real closed loop encoder feedback vector drive instead of messing with gearing and 2 pole motor. Think of it this way: change all that hardware and keep lowly vfd and get possible 2x improvement vs just change vfd to closed loop, add encoder, and get 1000x improvement.

http://www.KilroyWasHere<dot>com

 

[jraef]

Here's one possible way to approach the selection; look for a drive that offers Sensorless Vector Control cability, but can ALSO provide full blown Flux Vector Control (FVC) or Field Oriented Control (FOC) by adding an encoder to the motor (and often an accessory card to the drive). If the DRIVE is capable of being upgraded to FVC/FOC, that means it has the brain power to accomplish it, which will factor into the performance of the SVC algorithm. That way you can try their SVC and if it works for you, you are done but if not, then get the accessory card and an encoder and I know for sure it will work. The "holy grail" of FVC /FOC performance is the classic "100% torque at zero speed" capability which was only attainable with DC drives for years. That takes a lot of brain power to accomplish in an induction motor drive, which does not come cheap. But it is attatinable for sure. In fact, I know of a couple of high-end drive products available that are capable of FVC without the encoder feedback now, because of improvements in the mP technology as well as the way current is sensed on the motor output (an integral part of the accuracy when not using an external feedback device). I wouldn't use that for servo-like positioning or for hoist control where you need torque proving, but it is excellent for torque or velocity control accuracy. And by the way, another way of determining the capability of a drive to get you there will be to see if it is capable of running servo motors instead of induction motors. I'm not saying you need to use a servo, but a VFD CAPABLE of running a servo will be one that is capable of giving you what you need from your motor.

What you need is out there, one way or another. The main point is that if you are operating at that low of a speed, you are at the margins of the typical working envelope for induction motor drives. The drive mfrs or products who lead with price rather than performance are the ones which fall off at the margins.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[jraef]

Oops, I forgot the motor.

If you are going to be operating at that low speed for any length of time above a few seconds, you definitely need a motor capable of that. The way to attain that is going to be indicated by the "turn down ratio" of the motor, meaning the range of speed in which the motor is DESIGNED to operate without sacrificing performance and/or service life. You will want to look for a motor with a turn down ratio of 1000:1, meaning it can SAFELY operate at frequencies down to fractions of a hertz. You never mentioned the size, but at small power ratings you may be able to get that with a TENV motor, but larger sizes will usually mean the motor is going to be "blown", meaning it will have a separate blower for cooling air that is NOT powered by the VFD output, so it stays at full capacity even as the motor is barely rotating. Those motors will also come with better winding insulation and better bearing designs to thwart other potential issues of running on VFDs.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[Skogsgurra]

Jeff, are you thinking NFO?

It is now available in the US. UL listing and all.

But it is not, to my knowledge, available separately as a drive. If you really want to test one, then talk to DeLaval, the dairy and farm equipment manufacturers. They may be able to let you test one.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[nadica]

I managed to improve the smoothness greatly with the current hardware. The story is long why but I changed the "Motor Rated Voltage" setting to 170V from 200V and the rotation at 5Hz VFD output is much much smoother. The VFD is plugged in a laptop and I had the following figures on screen with the two different settings:

"Motor Rated Voltage"
230V 170V
Output freq. 5Hz 5Hz
Output volt. 28.7V 18.9V
Output amp. 5.2A 2.4A
Output power 149.2W 45.4W
Output torque 27.1% 11.5%
Bus volt 315V 314V

I am wondering if these figures could be correct? Why I have a smoother motion at lower figures. There was no load on the rig just some inertia and friction from the mechanism.

Should I try to play with the settings in V/f mode? (i.e. not vector mode)

PS: I am in the UK but the NFO website is not working.

Many thanks again to all of you!

 

[jraef]

What is the nameplate voltage of the motor? 200V? If so, what you have done is lower the V/Hz ratio, which makes the motor create less torque. If you don't need much torque at this low speed, the lower saturation of your windings may make it appear smoother, but it would be at the cost of increased slip, which means more current for the work being performed. If you are monitoring current and it's still low however, there should be nothing wrong with that, other than the cooling issues as mentioned before.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[LionelHutz]

Yes, the power output would drop because the motor isn't constantly accelerating and decelerating. I would try V/Hz mode. There typically is a voltage boost parameter that increases motor voltage at low speed with V/Hz that you'll have to adjust lower I would expect.

 

[Nellysdad]

Get a Smart Motor from Moog-Animatics in Milpitas Ca

 

[notnats]

If you can change the gear ratio, it should be possible to run a 4 pole motor at speeds up to 120 Hz to meet the high speed requirement. This will let you run the low speed at a higher frequency. This worked for us in a special treadmill to train wheelchair athletes and cyclists up to 40 km/h but still be stable with 110 kg footballers at walking speed

 

[jraef]

If you can change the gear ratio, it should be possible to run a 4 pole motor at speeds up to 120 Hz to meet the high speed requirement. This will let you run the low speed at a higher frequency. This worked for us in a special treadmill to train wheelchair athletes and cyclists up to 40 km/h but still be stable with 110 kg footballers at walking speed

That's fraught with a LOT of pitfalls, the least of which is the dramatic loss of torque as you go above base speed. There are tricks of the trade to come up with work-arounds, but those are not for initiates in the drives world and should never be done without consulting the motor mfr.


"You measure the size of the accomplishment by the obstacles you had to overcome to reach your goals" -- Booker T. Washington

 

[LionelHutz]

Motors are discrete devices. They have a couple of poles which are the sources of the torque or twisting effort. Because of that, they do not have a smooth continuous way of moving degree by degree. As they speed up the effort becomes smooth via the momentum of the spinning rotor.

I don't agree with this. AC motors are not the same as stepper motors. The magnetic field will rotate smoothly if you apply a sinewave current.

The root problem is that the vector control algorithm is not applying an AC current waveform to the motor, but rather a modified waveform that is supposed to create more torque. But without a sinewave current, the motor can begin to operate similar to a stepper motor at low speeds.

V/Hz mode should give a better sinewave to operate the motor more smoothly.

 

[cswilson]

Lionel:

I agree with you in part, but disagree in part as well. There are several interesting issues, so it is worth breaking them apart.

I concur that AC motors are fundamentally continuous, not discrete. This includes both asynchronous induction motors and synchronous motors like brushless servo motors and even stepper (yes!) motors. The windings on these motors produce at least approximately sinusoidal torque functions as the motor (or just rotor field) spins. With multiple phases, you can get very low torque variations if you put in sinusoidal waveforms.

Of course, higher harmonics from the windings, and reluctance/cogging torque in the magnetics will limit how low you can get your torque variations, but many AC motors are designed to minimize these effects, and it is possible to compensate for these in the controls.

You may be surprised that I included stepper motors in this class. But most stepper motors (I will exclude cheap VR steppers) are AC synchronous motors. With either open-loop microstepping control or closed-loop control (treating it as a high-pole-count brushless servo motor), you can get very smooth control with sinusoidal command waveforms to the phases.

But I fundamentally disagree with your assertion that "the root problem is that the vector control algorithm is not applying an AC current waveform to the motor." We've been doing vector control for almost 30 years now, and we have always applied AC current waveforms to the motor, which are sinusoidal functions of time in the steady state. The power stage for a vector control drive is really the same as for an open-loop VFD. Both synthesize AC waveforms from a DC bus by a modulation scheme, usually PWM. The differences are in the control schemes for synthesizing the waveforms.

Roughly speaking, there are three classes of control schemes (focusing now on induction motors):

1. Open-loop VFDs ("Volts per Hertz" drives): These command AC waveforms with no knowledge of what the motor is actually doing, relying on the electromagnetic feedbacks in the motor to make the motor (roughly) follow the command signals. As their nickname suggests, the output frequency and magnitude are at least roughly proportional. These are cheap and simple, but have the lowest performance. They have a lot of trouble at low speeds, because they have no capability to resolve the magnetization and torque component interactions as a vector drive would.

2. "Sensorless vector" drives: In the technical literature, these are called "shaft-sensorless" drives, because they do rely on voltage and current sensors inside the drive to try to figure out what the rotor is doing. Fundamentally, they attempt to back out the back EMF voltage component on the phases to compute rotor velocity and angle. This works well at high speeds when the back EMF is large, but is much more difficult at low speeds, when it is a small component of the overall values, and the signal-to-noise ratio is horrible. Still, it is significantly superior to open-loop control.

3. Full vector drives: These employ a high-resolution feedback device on the drive that directly measures rotor angle. This provides excellent performance all the way down to zero speed. In fact, it has been common for over 20 years now to use this to make induction motors operate as positioning servo motors. If a machine tool has a spindle that is capable of "hard tapping", it is almost invariably an induction motor under full vector control. And smooth motion at low speed is vital for this function.

I know of no performance disadvantages of full vector control compared to sensorless vector or Volts/Hertz drives. Of course, they are more complex and expensive.

Curt Wilson
Delta Tau Data Systems

 

[LionelHutz]

"I know of no performance disadvantages of full vector control compared to sensorless vector or Volts/Hertz drives."

You forgot a VERY important part - if properly implemented. Your dissertation about how vector control is so superior doesn't change the fact that the sensorless vector control in THe OP's VFD on THE OP's application is providing a crappy waveform to the motor that is causing it to step.

I guess you missed the part that I'm addressing the OP's question. I would have thought that was obvious since I'm responding in the thread where he posted his question.

 

[Ananym]

I have a similar problem.
My motor is highly rated for my application, so there are times at which I need to be able to run both at relatively low speed and low torque. However, this is a torque controlled application, controlled at the moment with sensorless vector control from a VFD. What I find is that the stuttering mentioned by the OP is present, but also that when torque reference is low the motor will not gain speed as needed, and I lose the tension in my load.

In sensorless torque control, the motor should speed up until the torque output is matched, so this implies that the system losses are enough to match the output torque. However, this seems unlikely given the numbers involved. Does sensorless vector torque control struggle to provide accurate torque output at low torque references? Will adding a motor shaft encoder for closed loop feedback help this issue? Are there any other obvious solutions? Cheers.