VFD control: An introduction

In this post we introduce basic thoughts on VFD control. Some of you asked deeper questions about different motor control strategies and their benchmark. Such comparison will be subject of a separate post, resp. a dedicated motor control series. Today’s article provides a base for further description on VFD control functionality. It is an introductory part into a new series.

Note that “control” is a very wide topic. You will find many valuable books and papers on the topic. Some sources, such as [1], have several hundred pages. We could probably operate a blog dealing with VFD control only and we would have enough to write about. Therefore, do not expect that this blog post answers all your questions regarding VFD control. Rather than that, we invite you on a journey with us to discover a bit more about how the VFD works from control perspective. For a deeper study kindly refer to additional resources provided at the end of our articles as usual.

Control is the element that provides intelligence to the VFD. As a parable we can say that when power semiconductor is the heart of the VFD then control is the brain.

VFD control

So how would you define the control of a VFD? It is a system consisting of (control) hardware and software that processes measured and estimated signals as well as reference values and commands and reacts in a corresponding way, as defined in control algorithms and protection functions.

Well, that may sound a bit complicated and may not be a bullet proof definition, but hopefully you got the point.

Real time operation

VFD control, unlike for instance offline simulation, has to operate in real time. That sets fundamental requirements on the performance of the control system. Therefore, a good balance between the computation load and cycle time needs to be found. Almost all control systems work with several cycle times. The functions are divided into tasks. Most critical functions that require fast processing and high sampling frequency are executed in the shortest task. In medium voltage VFD control systems such fast task typically has a period of 25 µs to 250 µs. The choice depends on the topology of VFD, modulation technique or application in general (dynamic applications tend to have shorter task cycle). Obviously, it would be extremely demanding for the processor to execute all control algorithms within such a short time. There are other tasks with longer period of execution suitable for processes and algorithms that are less time critical.

Let’s imagine a fan drive for a while. Modulator and critical protection functions will be located within the fastest task with period of several µs as explained above. On the other hand, speed control can be executed in another, much slower task. The fan has a significant inertia and the speed physically cannot change too fast. Therefore, it would be a waste of resources to execute the speed control algorithms within a task cycle time of 25 µs. Instead, a reasonable task cycle for speed controller might be in the range of 2 ms to 20 ms (unless it is a very special case such as e.g. servo drivers – but that is a low voltage domain anyway).

Digital signal processors

The heart of the control hardware is a digital signal processor (DSP). Architecture of such signal processor is optimized for fast and efficient computation. There are DSPs specifically developed for motor control applications. They typically have several PWM modulators or input for optional speed signal.

The arithmetic used to be either fix point or floating point. With the growing performance of DSP the fix point arithmetic is becoming seldom.

Main tasks of a VFD control

Main task of a variable frequency drive is to control the speed or torque of supplied electric motor. Therefore, speed and torque control loops are essential in the motor control part of the inverter. We will look into them a bit closer in the upcoming articles within this series.

While the speed control typically consists of some sort of classical PI or PID controller, the torque control is more manufacturer-specific or product-specific. For simplest applications with very low dynamics there is no torque control at all. More sophisticated torque control algorithms are field oriented control (also called “vector control”) or direct torque control (DTC). Again, this topic deserves its own article.

VFD control loops

Cascaded structure of control loops is very popular. Although it is not the only possible structure, it is probably still the most common one. And it is definitely a good structure to start with as it is easy to understand.

There is an outer control loop for speed and inner loops to control the current or torque (or both). Some applications require position control which basically means one additional outer loop integrating the speed. Principally, some of the outer loops might be realized in upstream control system. The overriding control might take care of the speed and send a torque reference command to the VFD. Or it might control the position and send a speed reference, for instance.

The principle of cascaded control architecture is quite straight forward. The controller in the outer loop provides a reference for the inner loop. Outer loop is usually controlling less dynamic signals while inner control manages quickly changing signals.

Steady state and dynamic operation

VFD control must be able to manage not just steady state operation, but also “slow transients”, such as start-up or dynamic braking and finally all kinds of dynamic transients caused by internal or external disturbances or by nature of application. An example of such dynamic operation is a fault ride through mode described recently [2].

One brain or more brains?

So how many brains does a VFD have? That is a good question – and a matter of definition. There is one brain for the application. But there can be several smaller brains for the main parts. Inverter might have its own control taking care of motor speed, torque, protection etc. At the same time an active rectifier (that might by identical to the inverter from power hardware point of view) has its own agenda: the tasks are control of dc link voltage and grid side power factor control (talking about most common VSI drives – other drives, such as LCI, work a bit different [3]).

Control and protection

The control system is also responsible for protection. In fact, control and protection are so closely related that it might be difficult to separate one from the other. It shall keep the VFD operating within a range that is safe for the equipment and of course for the personnel.

In VFD driven motors the motor protection is normally covered by the VFD so that no extra motor protection relay is required. We will talk more about what motor protection functions belong to a solid VFD control.

Control and diagnostics

The VFD brain can also be used for diagnostic purposes. There is a shift from preventive maintenance to predictive maintenance (nothing really new) and the VFD ‘intelligence’ can be a major contributor to move this trend forward.

Yes, and there is definitely the whole world of AI and machine learning algorithms. That is surely also a topic on its own.

Sparked your interest? Then just follow this series. Sign up for a newsletter if you don’t want to miss future updates.

References

[1] P. Vas:  Sensorless vector and direct torque control,  Oxford University Press, 1998 (ISBN 0-19-856465-1)

[2] Undervoltage ride through, https://mb-drive-services.com/ride_through/

[3] LCI versus VSI, https://mb-drive-services.com/category/lci-versus-vsi/

[4] Power, speed and torque, https://mb-drive-services.com/power-speed-and-torque/

[5] Medium voltage AC drives, https://new.abb.com/drives/medium-voltage-ac-drives

[6] Motors and generators, https://new.abb.com/motors-generators