OEM Design Choices for Chips, Cards and Drives

Abstract

The purpose of this paper is to illustrate how OEM machine designers can benefit from the different types of motion control components available today.

This paper compares how motion control solutions based on chips, cards, and drives fit into different ma-chine requirements, and how they affect the outcome of a machine design. This paper also examines mo-tion system architectures to illustrate the design trade-offs. Examples discussed include medical/biotech equipment, a 3D printer, and an industrial paper-cutting machine.

Topics include:

• The importance of early-identification of machine requirements.
• Motion component definitions.
• Motion component selection.
• Motion system architecture design.
• The effects that the choices of motion component and the resulting architecture have on cost, relia-bility, maintenance, and time-to-market requirements.

Introduction

Designing a machine that will incorporate motion elements is a challenge that requires knowledge of both mechanical and electrical engineering disciplines. Because its role is central to the machine’s function, the motion control system influences many important aspects of the machine, especially from an architectural point of view.

We will start by looking at all the possible motion requirements for a new machine. Identifying the requirements reduces the risk of costly redesign due to forgotten requirements. Throughout this paper, we will examine the different motion control architecture solutions and their advantages and disadvantages. We will also look at the affect the level of integration each motion component has on the architecture and overall design effectiveness including cost, maintenance, reliability, and manufacturability.

Identifying Requirements

There are many ways to identify system requirements, but there is no single “right” way. To help convey the importance of the requirements to the motion control design, we will present some of the critical steps that should take place during this initial phase.

External Requirements

There are, of course, requirements that come from outside the engineering department. Some of the common ones are:

• Cost (Production and/or Development).
• System size (dimensions).
• System performance (e.g. throughput).
• Time to market.
• Usability (the human factor).
• Manufacturability (Assembly).
• Maintenance.

These requirements are usually influenced by the customer, market place, and/or management, which means they are critical. Failure to meet one or more of the requirements can result in cost overruns or failure of the product in the market. So, it is important to always have these in mind as you progress through the design.

Technical Requirements

Now that we have the external requirements, we can start focusing on identifying the technical requirements related to what the machine has to do. The following questions can help us define the technical requirements. 

1. What is the machine designed to do?
2. What are the process steps that involve motion?
3. How much space is needed to do the process steps? (We are not including the electromechanical system needed to do the process.)

For example:

• Paper cutting machine.
• Max paper size A0, i.e., 1 m x 1 m.
• Max paper stack height 20 cm.
• Minimum length for process 2 m, since paper has to be able to enter and leave process.
• Minimum width 1 m.
• Minimum height 20 cm.
• Result: Minimum process envelope 1 m x 2 m x 20 cm. 

4.      What steps are needed to accomplish the process?

For example:

• Paper cutting machine.
• Move paper into machine.
• Move paper out the machine.
• Cut paper, i.e., move blade.
• Adjust paper stack, i.e., alignment on all four sides.
• Paper vertical holding around blade to avoid paper moving while being cut.
• Etc.

5.      What motions are needed to perform each of the steps in the process?

• Linear?
• Rotational?
• Torque and/or velocity control?
• Point-to-point? Etc.
• Etc.

6.      Based on the answers from question #5:

• How many axes are needed?
• Do they need to be synchronized?
• Are there time or event triggers?
• Etc.

7.      What kind of architecture makes sense based on the requirements that have been found?

8.      Are there any unique environmental factors?

Note that this is not an exhaustive list, but it is the minimum needed before starting a machine design involving motion. It is important that your requirements list be as exhaustive as possible to avoid any painful redesigns.

This includes receiving market/customer requirements from outside of the design engineering function, (i.e., directly from the customer, marketing, sales, management, etc.) or anybody who has influence on the product and the project so that you capture a full list of requirements before proceeding.

Motion Components: What To Use?

Now that we have the requirements, we need to decide how to perform the different motion functions in the system. Depending on the complexity of the process and the features it requires, a motor (rotational or linear), solenoid, or pneumatic component can perform the different motions. We are going to focus on the motor-based motions and, as the title of this paper says, we are going to compare motion-control solutions based on chips, cards or drives.

Motion Component Definitions

To start off, we need to define the motion components in question. Each requires a certain level of integration, depending on how much of the design and assembly is done by the supplier.

Motion Control Chips

When using a chip-based solution for motion control there are a few options.

• You can use standard off-the-shelf products, like MCUs (microcontroller) or DSPs (digital signal processor), and develop your own motion control algorithms. This is the least integrated solution you can buy, and it can be the right choice if you have high enough expected product sales vo-lume and experienced design staffing to make this feasible.
• You can use preprogrammed motion control ICs or dedicated ASIC (application-specific integrated circuit) solutions. With these motion control ICs, you can avoid the time-consuming task of develop-ing everything yourself. You will still have to develop a printed circuit board for this chip and code to run on it, but using dedicated motion control ICs will save design time. The chip may sit on the same PCB as the host processor/controller, saving you space.

The following table compares these two approaches.

Table 1: Motion Control Chip Options

 

The number of axes per chip may vary depending on the complexity of the signals needed or supported per axis. Some 4-axis chip solutions come as two-chip solutions and support several features, like commutation of 3-phase brushless DC motors, DC brush motors and stepper motors.

This makes it nearly impossible to put all of the needed signals in a single package. Then, there are some that limit their drive signals to pulse and direction outputs for each axis, which then makes it feasibly to use a 144-pin package, for ex-ample.

Motion Control Cards

Dedicated motion control cards come in different form factors, but usually there are PC-bus based, i.e., PCI, ISA, PC/104, etc. Depending on the manufacturer, the cards normally have 4 or 8 axes per card.

Motion control cards are easier to integrate into your system as compared to chips. Motion control cards are a more general purpose and standardized solution using a PC-bus architecture. They are designed to provide a much more complete solution that is flexible and scalable. If those are requirements you are trying to meet, then cards could be a good choice for your design.

Motion control cards are also available with stand-alone functionality. These cards look similar to their PC-bus cousins, but, instead of a PC-bus, they usually have a high speed serial bus, like Ethernet, CAN-bus, RS232/485 or a variation of them.

The stand-alone cards are always programmable, which allows them to do some tasks autonomously. This feature is important since these cards usually have a slower communication interface with the host processor, compared to the parallel bus available with the PC-bus-type cards.

Motor Drives

The motor drive combines motion control functions that a chip/card supplies with a power amplifier stage for a single axis. It is similar to the stand-alone card in that you communicate to it through a serial bus. Until recently, programmability was a special feature, but is now becoming a standard feature.

The drive is an integrated motion control component requiring minimal skills to set up and install. Drives usually come with simple setup instructions or an easy-to-use GUI running on a PC-based platform. If they do not include some stand-alone functionality, they at least come with a simple command set that can be sent over the serial bus to control their function and read out data.

Their power rating can be from 50W all the way up to tens of kW, though most of the motion applications are in the <10kW range. The motor drive allows for simple installation of motors in a PLC based system, since they are nearly always fieldbus based. In this way the fieldbus communication allows for distributed real-time control architecture.

Fulfilling The Requirements

Now that we have defined the motion components, we need to see how they fulfill the different requirements that we defined in the beginning of this paper. Table 2 summarizes the pros and cons of implementing each of the motion components with regard to the requirements. The following sections provide de-tailed information about using each type of motion component.

Chips vs. Requirements

Choosing a chip based solution gives you the flexibility to meet nearly all of your requirements, but there are certain requirements that you may not be able to meet with this solution. Specifically, the design cost and time will be much higher.

Also, the requirement of small size most likely will not make the design easily scalable, which, in many cases, makes the architectural choices dedicated and limited. Table 2 summarizes some of the obvious advantages and disadvantages of a chip-based motion control design.

Cards vs. Requirements

With cards, size is both a pro and a con. This is because of the different number of axes per board and form factors available. For example, a system that needs a lot of axes and uses two 32-axis PCI boards can meet the right size requirements since the mother board only needs two slots for the motion cards and one or two slots for I/O and/or communication cards.

Also, a 16-axis PC/104 card stacked on top of a SBC (single board computer) makes for a compact design, but, relative to some other approaches, the PCI and ISA cards require large chassis.

Motion cards are based on industrial standards, which make it easier to get pre-made components with a lot of software support from the supplier; this improves time-to-market factors. Also, being PC-based solutions, the performance is usually very good and allows for scalability.

However, being PC-based can also be a limitation, since it requires a host PC to implement and therefore limits the opportunities to optimize the architecture, and cost is therefore significant. Another benefit of the motion cards is that most manufacturers add general purpose I/Os on the board that the host processor can access directly. These I/Os can be used as part of the motion control process or for any other system need.

This can help the designer to avoid having to add dedicated I/O cards to the system and in such a way save cost and additional programming. Again, Table 2 summarizes some of these advantages and disadvantages.

Drives vs. Requirements

The advantages of drives are mostly related to the fact that they are a highly integrated product. They allow for scalability and a distributed architecture, which allow for certain flexibility in optimizing the overall system architecture. The drive’s high integration also helps to minimize the time to market and re-duce design cost. This, on the other hand, increases the component cost and size.

The serial bus communication that is normally used in drives, i.e., fieldbus, has up to now been rather slow and therefore, the performance has been a disadvantage for the drives. But with the emergence of Ether-net based fieldbus solutions such as EtherCAT, performance will most likely get close to what we see in the chip- and card-based systems. Table 2 summarizes the advantages and disadvantages of using a drive.

Table 2: Motion Component Choice and Design Process Impact

 

Motion System Architecture

There are two categories of architecture:

• The main system/host processor and the motion control IC are separate entities, with a possible I/O processing unit also as a separate unit (Figure 1). This is the most common architecture.
• The main system/host processor also doing the motion control, but I/O processing is separate (Figure 2).

Figure 1: Basic Machine Architectures – Separate host and Motion Control Processing

 

Figure 2: Basic Machine Architectures – Combined Host and Motion Processing

 

Now let us examine each of the motion control components and see how they affect the architecture and choices for optimization.

Chip-Based Architecture

The two main approaches to chip-based architectures follow the aforementioned main system architectures. The first one is a multi-chip solution and the second one is a single-chip solution. Let’s start by looking at the second one first.

Single-Hhip Solution

As shown in Figure 2, all of the system and motion control is in one chip. This chip must be powerful to deal with all the tasks that are involved with a multi axes system. Figure 3 shows a common implementation of the single-chip solution.

Figure 3: Single-chip Solution

 

As we mentioned in the definition of the motion control chip, often the host processor, shown in Figure 3, is a standard off-the-shelf MCU or DSP for smaller systems. If so, there might not be a need for the FPGA to do the signal conditioning. But, if the design is more complicated, such as more than four axes, then an MPU is needed. In that case an FPGA is most likely mandatory to handle the additional peripherals (timers, PWMs Etc.) and data I/O requirements.

This solution allows for a compact and optimized design and can be cost effective in production. Conversely, because of the use of a single host processor, you run a high risk of not having enough processing power, especially if new requirements and features are added late in the project. It is possible to design in some head room for this, but that would come at a cost.

The complexity of the firmware is also a consideration, as it can become challenging to deal with as the complexity of the motion system grows. The development of the motion control system is one of the most complicated tasks in a design. This solution will make your development time significantly longer. A single-chip solution is a good choice if there is enough volume to benefit from the low per-unit cost despite the possible increased development time.

Multi-chip Solution

The diagram in Figure 4 reflects the architecture of the multi-chip solution. Here, the motion control is put into a separate chip(s). This offloads the host processor, to allow it to focus on the overall system control, including possible information processing, e.g., biotech sample analysis. Figure 8, in the “Architecture Ex-amples” section, shows an application example using this architecture.

Figure 4: Multi-chip Solution

 

The motion control IC shown in Figure 4 can be a one- or two-chip solution, depending on the complexity of the signals it handles and the number of axes it controls.

Let’s look at the ways we can implement the motion control IC. As we said in the “Motion Component Definitions” section, this can either be an in-house solution or 3rd party pre-programmed/ASIC solution. Creating your own motion control IC solution and writing your own algorithm can make sense for the single-chip solution.

It depends on the complexity of your motion control, i.e., motor type, precision, coordination, etc., whether you have enough volume that it make sense both from a cost point of view, and/or it fits as core competence in your company. For most companies, the better option is to buy a 3rd party solution.

As mentioned before, these come as either single-chip or two-chip solutions, depending on the number of axes and the complexity of the motor control.

The basic implementation of these 3rd party motion control IC solutions is for a 1- to 4-axis solution per chip(s). Below is an example of a minimal implementation. 

• Pulse/Direction output or ±10V analog command signal.
  • 2 pins per axis.
• Quadrature encoder feedback.
  • 2 pins per axis.
• Digital and analog sensing/safety inputs.
  • 4 to 8 pins per axis.
• Status outputs.
  • 4 to 8 pins per axis.
• Data and address bus interface.
  • 8- or 16-bit data, 6-bit address, 2-3 control pins.
• Serial communication interface.
  • 2 to 4 pins, depending on type.

If we add this up, we get 30-49 pins for a single axis solution and 66-109 pins for a 4-axis solution. This is not including any system pins, general purpose I/Os, or power/ground pins the chip might need. As you can see, it would be hard to use anything less than a 64- or 80-pin package for the single axis and 80- or 144-pin package for the four axes. If you want something more sophisticated, the only way to accommodate a two or more axes solution is to use two chips. Some of the common additional features of a more sophisticated solution are: 

• PWM commutation signals for multiple motor types, e.g., DC brush, Brushless DC and Stepper. (Depending on the amplifier used.)
  • The stepper motor requires up to 8 PWM signals per axis.
  • Brushless need up to 6 PWM signals and possibly 3 hall-effect signals per axis.
• Multiple serial communication option.
• User configurable analog inputs.

These would be additions to the aforementioned signals. Features like this allow you to have more sophisticated control. In some cases, you will be able to have different motor types on each axis to optimize your design even further. But, as always, the fancier it is, the more work it requires.

Architecture Options: FPGAs & CPLDs

Third party preprogrammed Motion control IC chip sets or ASICs provide the motion processing and the additional peripherals needed for multi-axis designs which avoid extensive design work and save time. FPGAs are a way to further augment Motion Con-trol IC – based designs if further peripherals are needed. If you are considering designing a chip-based motion solution from scratch then FPGAs are a common, if not required, compliment that should be looked at.

They provide pre/post processing of signals going into the processor and are a way to add peripherals such as timers, PWMs and I/O expansion if needed. Adding an FPGA or CPLD to a design can also decrease the design time needed and help with redesigns and bug fixes.

They can add value to your project and should not be dismissed purely from a component cost perspective (as costs are relatively low now). For example, in a single-chip solution with multiple axes, an FPGA/CPLD is nearly always needed, as most MPUs, MCUs and DSPs have limited peripherals and you can run out of resources as you reach two or more axes.

In Figure 8 of the “Architecture Examples” section you can see an application example of how the FPGA/CPLD can help in a design by adding additional I/O and feedback connections.

Card-Based Architecture

As we talked about in the “Motion Component Definitions” section, motion control cards are normally PC-bus cards, like PCI, ISA or PC/104. Some manufacturers support industrial PC-buses like VME or PXI. Also becoming common, are stand-alone cards that are effective in designs that do not require a PC-bus. They are similar, but impact system integration configurations differently. Since the PC-bus cards are more common we will start with them.

A block diagram of a system that is using a PC-bus motion control card looks similar to the multi-chip one. Figure 5 shows a generic implementation which also includes possible common expansion features.

Figure 5: PC-bus Card Solution

: PC-bus Card Solution

: PC-bus Card Solution

 

This design is scalable, but it comes with a price. As the design is scaled up, the cost will increase proportionally based on the cost of the cards and number of axes per card. Also, you must use a PC-bus based system, which means you must accept the form factor it dictates. The PC/104 card has a fairly compact form factor compared to the PCI, but compared to a chip based system, it is large.

The benefit of using a PC-based system is that you are running a common operating system and your time-to-market will be reduced. This can be easier than a chip-based design. Still, the code writing requires some expertise when you are working with these PC-bus cards, which make code development a challenging task for inexperienced designers.

Stand-Alone Cards

Stand-alone cards allow for a similar architecture to PC-bus cards, but the bus that is used to communicate from the host to the motion control system is different and allows for a different overall system architecture. Figure 6 shows a typical system setup.

Figure 6: Stand-Alone Card Solution FeedbackPC

: Stand-Alone Card Solution FeedbackPC

: Stand-Alone Card Solution FeedbackPC

 

The stand-alone cards include a mini-host that takes care of the communications with the host and executes local programs. (Note: This feature is sometimes available on the PC cards, but does not have the same impact there as it does here.)

By executing local programs, these stand alone cards optimize the motion control system performance while having a distributed motion control system. In the PC-based configuration, the host processes the more complicated, coordinated motion tasks. It then communicates these tasks to each axis on the card.

This requires a fairly high speed communication link between the card and the host. This is especially true if you are coordinating moves that are fast. In this case, it is possible to segment the task, where the host will pre-process the move, while the actual execution of the coordinated move happens in the mini-host on the card.

If you are using a fieldbus based on Ethernet, you might have enough communication speed to keep more tasks in the host, but that would not make sense with this architecture as the goal is to distribute the motion processing tasks.

As we mentioned previously, the stand-alone card architecture has significant cost and size advantages over the PC card architecture. It does not require a standard form factor chassis, which could allow for a much smaller and lower cost installation, depending on the motion application and number of axes your design requires.

Also, you can use any host processor, including a standard PC, but the programming effort will be simpler since you don’t have a PC-bus. On the other hand, you will have to program the motion control card instead, which can add some significant work, but that depends on the utility of the tools provided by the motion card supplier.

Drive-Based Architecture

In some ways, the drive-based architecture shown Figure 7 is the same as the stand-alone card we just examined. Both are usually based on a serial bus, but for the drives, that serial bus is usually a fieldbus, which may allow synchronized behavior between the drives. The biggest difference between the two architectures is the inclusion of the amplification for each axis and the support of a single axis by each drive.

Figure 7: Drive-based Solution

 

Using a drive as a motion control component allows you to:  

• Move some of the motion control processing closer to the amplifier.
• Combine control of the current and position loop.
• Simplify the communication to the host.

The concept behind the digital motor drives as a motion control component is to move some of the motion control processing closer to the amplifier and in that way benefit from the combined control of the current and position loop. It also simplifies the communication to the host with the use of a fieldbus.

These features relate directly to the most common use of the motor drive: to allow the system builder to replace mechanical indexers and cams with direct drive motors. To accomplish this, the drive does not need full programmability, but simple parameter settings that allow the installer to set up repetitive motions.

As a result, drives have continued to be an important component in minimizing the complexity around set-ting up and programming motion. This has made the drive a perfect candidate when time to market is one of the key requirements. Furthermore, it has also helped in making the drive an excellent choice for distributed controls systems, which have been further integrated by the drivemotor combinations that are now available in the marketplace.

Choices For Reliability & Maintenance

Reliability and maintainability are two key requirements for most machines. Each of the motion control components and architectures we have discussed have pros and cons in this regard, but there is a significant difference between them because of their different levels of integration.

Chip-Based Solutions

The benefits of the chip-based solutions regarding reliability and maintenance are obvious. Since you are building the system, you decide what levels of redundancy, diagnostics, and component specifications you need from a component. But, because you are creating the design, you are assuming responsibility for the reliability for the whole motion control system.

You must ensure that you have resources for testing and quality assurance, either inside or outside of your organization, that give a level of assurance you need for the machine design to be successful in the market place. Another benefit of a chip based design is that you can minimize interconnections.

For example if you design the motor amplifiers on the same board as the controls you avoid the possible failure points that a cable and connectors would have.

Card-Based Solutions

If you choose to use motion cards, you are outsourcing the board design and its inherent reliability. The suppliers of the computer and boards guarantee that their products are reliable and safe, but you may still have to consider the reliability and serviceability of the interconnect card.

For maintenance, cards are an improvement over integrated motion, especially in a PC-card based system. There you can pull out the non-functional board and replace it with a new one. Still, this raises some issues.

If a card breaks and it controls 16 axes, it can have a negative effect on the machine process that may require process calibration and axis tuning. To reiterate the point from the “Identifying Requirements” section, having a clear picture of the requirements and their priority is essential for a successful design.

Drive-Based Solutions

Because the drive is so highly integrated having control, commutation and amplification in a single unit for a single axis, it is well designed for maintenance and allows designers to optimize their systems for reliability. This of course comes at a price. Each axis cost is relatively high, especially versus the chip design, but also relative to high number axes cards.

Cost & Time-To-market Considerations

These are subjects that are very dear to management and marketing. Keeping costs down, both from a design and production point of view, contribute to the gross margin and commercial success of the machine.

Because time-to-market and design cost (including product cost) can often have an inverse relationship, it is important to have a clear vision at the beginning of the project as to the priority of needs. If time to market is the critical success factor then the designer should proceed with that in mind.

In choosing motion control components for end product cost optimization, the chip-based design can be optimized the most, especially if there is significant volume expected for the end product. But, inversely, chip-based motion system designs have a relatively longer design cycle compared to cards and drives. The cost for the cards and drives is usually similar on a per axis basis.

For the cards, the cost is much related on how many cards and the number of axes per card there is, but that of course relates to our discussion on reliability and maintenance in the previous section. Regarding time to market, the cards lie in the middle. Be-cause of their normally PC-based architecture, they need a lot of low level programming that can take some time, but there are little or no hardware design requirements.

The drives, on the other hand, can be optimized for time to market, because of their relative easy-to-use requirement from the beginning, but their cost and sometimes their size can be prohibitive.

Type Of Application

Theoretically, different motion control components can be used in any machine, but, over the years, a division has evolved and each of them is being used dominantly in certain industries. Below is list of applications and industries for each component category.

Chip-Based Solutions

As expected, the industries and applications that use chip-based motion control solutions are size limited and specialized, like medical/biotech analyzers (table-top versions). There are also cost-effective applications, but they normally do not use preprogrammed/ASIC motion control solutions. These are mostly do-it-yourself specialty designs.

Card-Based Solutions

Here, the PC-based host is the dominant solution for the host process, even for the stand-alone card architecture. The reason is mainly that there is a feeling that PC-OS based systems, like Windows or Linux, is the most fitting solution. It often has to do with what other processes are running on the machine. Some of them have complex programs running on them that are easier to develop and maintain in a PC-OS environment. You usually find this implementation in specialized semiconductor machines and large scale blood analyzers, or other machines with similar software process requirements.

Drive-Based Solutions

The origin of the drive-based solution is in industrial automation, but for the last 10 years their use has in-creased outside that market as they have become more advanced and sophisticated. Key factors in determining where and when to use drives are related to their integration of an amplifier and the distributed architecture based on the fieldbus. The benefit of the integrated amplifier comes when the power rating is higher (i.e. near 100W and above).

For the fieldbus, the benefit is the size of the machine/installation, such that it makes sense to have a distributed system. For these reasons, the use of a drive is most often related to the size and complexity (e.g. number of axes) of the system. They are used mainly in industrial equipment and larger scale semiconductor manufacturing systems, for example.

Architecture Examples

To illustrate how these components are used in applications, we provide three examples of machines that have features that make them a good fit for each of the motion control components.

Chip-Based Example – Medical/Biotech Equipment

This is a medical machine that has the following functional requirements:

• 1 linear axis – point-to-point motion control.
• 2 rotational axes – velocity control, i.e., pumps.
• <3 ft3 of space.
• Machine needs to test and measure the fluid being processed.
  • Calculations need to happen on the fly, i.e., as the fluid is flowing - calculation speed is important.

Based in these requirements, it makes sense to split the processing between a host processor and a pre-programmed motion controller, which leads us to use a multi-chip architecture. We use an FPGA because there are a lot of digital signals to be sensed or output by the overall system control.

Also, the FPGA will handle additional feedback signals for the motion control IC. For the test and measurement functionality, we will connect a 16-bit ADC to the host processor through the system bus. Figure 8 shows a simple over-view of the possible PCB implementation. Estimated PCB size is 6"×8".

Figure 8: Medical Equipment Example – Simple PCB Diagram

 

The example in Figure 8 uses a preprogrammed 3-axis motion control IC chip set, and simple amplifier circuits to minimize board size. Each amplifier circuit is a 3-phase driver circuit for a brushless motor. For current control, we will use an analog feedback circuit. The motion control IC is generating the 6 PWMs for commutation for each motor.

The optimization of the design (in the form of a single circuit board) makes a lot of sense, especially since the machine is small and is doing a lot of specialized processing, which is the actual value the machine maker is creating. Therefore, there is no need to spend time and money on developing a specialized motion control algorithm.

Card-based Example – 3D Printer

• This is a 3D printer that has the following functional requirements:
• 3 axes for the 3D motion – all linear point-to-point position control.
• 2+ axes for other possible motion – all linear point-to-point position control.
• System has to be Linux based to match previous product versions.
• Some axes have repetitive motions that should not be handled by host.
• Size is not an issue because of the process envelope being rather large.
• Speed is important to allow for higher throughput and making of larger parts.

Since this is a large machine that has to be Linux based, it makes sense to use a PC-based architecture and PCI-bus for speed/performance. To optimize the design and improve reliability and maintenance we are going to partition the axes into two groups. 3D motions will be on one motion control card while the others will be on another. One possible partitioning is to have the 3D axes on one card, possibly a 4-axis or 3-axis card (not all manufacturers have 3-axis cards), and the other axes on a 2+ axes card. Figure 9 shows a possible motion control architecture for this printer.

Figure 9: 3D Printer Example – Computer Chassis Block Diagram

 

Also in Figure 9, we can see there are two interconnect cards and an I/O card. Some motion control cards have general purpose digital and analog I/Os, but that is not always the most convenient implementation for a machine design.

As we have discussed previously, the benefit in using a PC-based architecture is the capability to expand the system and use off-the-shelf card solutions for all the various control functions one might need, e.g., I/Os, communication, imaging, etc.

Another benefit is that you can easily get keyboards and displays to function within your systems, since the PC is based on industrial standards and supported by the OS.

The optimization we did in this design made sense from the requirements we had, both business and engineering, but if there were pressures on the cost, we might want to consider getting a single card with eight axes and enough I/Os to meet the overall system requirements.

This system reflects the makers focus on the integration of electromechanical components to create 3D objects. The system input is coming from a CAD program and it contains a large amount of data, which means that it makes sense to use a computer to translate it into the controls of the electromechanical system. The core value in this design again is not the motion control algorithms themselves, but rather the translation of the incoming data to the physical 3D object.

Drive-Based Example – Paper Cutting Machine

This is the example we were using in the beginning of this paper and it is a classical industrial machine. Let us reiterate the functional requirements: 

• 2 axes for cutting – linear motion with torque control & limit switches.
• 2 axes for material handling – rotational with point-to-point control.
• 4 axes for paper holding – linear motion with torque control.
• Stand alone system with minimal factory interaction.
  • Minimal data collected and simple HMI needed.
• A lot of repetitive motion.

The functionality of the machine is elementary, and, for a machine like this, reliability and maintenance is important. The motors are relatively big compared to our previous two examples. All of this points to a drive-based architecture. Figure 10 shows a simplified diagram of the system, showing only the major motion axes.

Figure 10: Paper Cutting Machine Example – Main Motion Axes Only

 

To meet the functional requirements, the cutting axes and holding axes should have programmability to offload the repetitive motions, e.g., to cut should be a fixed motion only needing torque control with an encoder limit at top and bottom, with backup mechanical limit switches for safety.

As for the paper holding axes, the most likely implementation is torque control with the host reading the position of each axis to check the size of the paper stack and make sure the axes are not stuck or over compressing the paper stack.

The machine maker is mainly delivering their system integration expertise as the added value and that makes the drive-based solution the best fit. It is also much easier for them to add and remove features from the system using this architecture. As mentioned earlier, the reliability and maintenance capability of the drive-based architecture is also important for an industrial machine like this.

Conclusion

Now that we have reviewed the different motion controls architectures based on the different motion control components, we see that there is a lot of commonality among them, but each has its unique strengths and weaknesses. The key to selecting the best fitting architecture and components is to have clear requirements up front. To make this as simple as possible, a step-by-step process, like the one we describe in the “Identifying Requirements” section, helps keep the design on track with meeting requirements and gives those involved the opportunity to offer their input before the design phase starts rather than after the de-sign is underway when changes become much more costly.

Also, there might be more than one good solution to a design challenge and, if you have followed this process, you will be able to identify the one that best meets the requirements, especially the most critical requirements. The solution could include all the different architecture options, but keep in mind that each of them has different resource and skill requirements.

To summarize each component type: 

• Chips allow you to optimize for size, functionality and per unit cost.
  • Make sure you have the time and resources for design, testing and quality assurance if you are making your own motion control boards.
• Cards allow you to work in a PC-OS environment
  • Consider the real-time behavior of the OS.
• Drives allow you to get to market quickly and provide a serviceable solution:
  • Analyze the cost and number of axes requirements carefully as each axis is relatively ex-pensive.

For more information visit www.pmdcorp.com

FIRST Update from Team 71, Team Hammond: And so it Begins…
Randy Weber of Team Hammond

Risky Business: Funding Medical Device Innovation
Rahul Sathe, Principal Mechanical Engineer, Surgical and Interventional Products, Cambridge Consultants

Extracting Nuggets from the Invention Mine
Tom Tuytschaevers, a member of our Patent Practice Group