Automating Analog Electronic Designs for Complex Physical Systems

	ABOUT THE AUTHOR Gines Domenech-Asensi obtained his Industrial Engineer degree in 1996. Currently he works as assistant lecturer in the Department of Electronics of the Universidad Politecnica de Cartagena. He started working in the development of EDA tools, thank to a stay in the University of Southampton, under the supervision of Dr. Kazmierski. This paper is part of his Ph.D. work.

Designers use VHDL-AMS (IEEE 1076.1) for the specifying and simulating digital, analog, and mixed systems. You currently have available several commercial VHDL-AMS simulators to verify the behavior of a wide variety of physical systems—electrical, non-electrical, or mixed-domain. Many in the design-automation community believe that the driving force behind future VHDL-AMS applications will be automated synthesis of analog and mixed-signal hardware structures from high-level functional specifications. Digital systems offer automation of the design process from a high-level specification to physical implementation as a standard industrial practice. However, analog-system design is highly dependent upon the designer's experience, with only a few tools to help in the design task.

The methodology we propose is based on a migration from different description and/or simulation tools to VHDL-AMS. In the case of a complex system, you must obtain mathematical models that characterize the behavior of different subsystems, each at its own level, and move to a common representation level. Once at this common level, we can complete development of the different parts of the system, refining various pieces until the entire design meets specifications.

For systems comprising the design of analog or mixed electronic subsystems, we present a method to develop a synthesis tool that lets you, from a completely specified VHDL-AMS description, do synthesis of the electronic subsystem.

For example, if we build an analog electronic circuit to control a physical system, we can use VHDL-AMS to develop equations that describe the behavior of the physical system plus the behavior of the control circuit. We can then simulate the system and use the simulation results to translate the specifications from the electronic portion to a physical synthesis of the system. It is much more reliable to perform a joint simulation of the electronic system which we intend to build together with its environment as opposed to a simulation of only the electronic part of the system. An advantage of VHDL-AMS over its digital predecessor, VHDL, is the capability of putting into the same specification a variety of physical systems that we previously needed to simulate with different tools than those used for electronic simulation.

In this article we present the general design methodology together with a practical example of the design of an analog controller for a real system—an industrial paper manufacturing line.

Synthesis Methodology
Figure 1 is divided into three sections. The first section is the initial description of the physical system in a common specification frame such as VHDL-AMS. In this section, models of the different subsections that compose the global system (control subsystems, electronic subsystem, and so on) are described in VHDL-AMS.

Figure 1:  The synthesis methodology comprises three tasks—initial description of the physical system, system development and simulation, and electronic synthesis

Once we have the entire system described in this standard, we continue to the second section—development and simulation. This is the place in which we develop and simulate the new component—the analog subsystem. You do these tasks with a tool such as the simulator distributed by Mentor Graphics. With such tools, we can develop a model for the new subsystem, which will couple the whole design and the simulation.

In this case, we want to design an analog subsystem. Once we have defined the model for this subsystem, we go to last section—electronic synthesis. In this section, we use a specific tool to perform synthesis of the analog electronic subsystem and obtain an electronic design whose behavior is equivalent to the model developed in VHDL-AMS.

Example: Design of an Analog Controller for an Industrial Process
Industrial paper manufacturing, along with manufacturing of materials such clothes with elastic properties, involves the control of two major parameters in the process: the speed of the material on the processing line and the mechanical strength of the material. The material is usually driven through a couple of rollers whose average speed determines the speed of the material and the relative torque applied to the rollers depends on the material's mechanical strength.

Figure 2:  Diagram of the industrial process

The rollers are powered by electric motors. Let us suppose we have a couple of DC motors with constant excitation field. For Engine 1, we will have the following set of equations:

(1)

(2)

Equation 2 represents the torque generated by the current i and the resistance torque due to the friction and the inertia J. If we now add the effect caused by the elastic force of the material we obtain:

(3)

where r is the radius of the roller. For the Engine 2, we would obtain similar equations.

Control parameters are the speed and strength of the material, while the control variables are the voltages applied to both engines. We can decompose these variables into common mode and differential mode voltages to obtain speed control (common mode) and strength control (differential mode) variables. From this decomposition we get:

(4)

The Controller
The controller we want to design is a bio-inspired controller, as it represents a simplification of the human spinal cord for the control of a limb position. We have chosen this controller because, in the human body, one has to control both speed and stiffness of the limb, and in our industrial example, we need to control both speed and strength of the paper driven in the rollers.

The controller (Figure 3) is composed of four nodes: M₁, M₂, I₁, and I₂. M₁ and M₂ are the main drivers of the DC motors that run the rollers. I₁ and I₂ are feedback elements for achieving more linearity in the control process.

Figure 3:  Diagram of the industrial process

The following differential equations describe the behavior of Nodes M_i and I_i:

(5)

where A_i represents the addition of the common control signal and the differential control signal.

(6)

Automatic Synthesis of the Controller
We can use the tool described in Domenech-Asensi to automatically synthesize the controller for the industrial process. Following the steps in Figure 1, we must first describe the process in VHDL-AMS language, as shown in Figure 4.

The second step is the development (already done in previous section) of the controller and its definition in VHDL-AMS: Figure 5 shows a part of the VHDL-AMS model of the controller.

In Figure 5 we can see the behavioral description for both types of nodes, M and I, of the bio-inspired controller, along with the controller's structural description.

The last stage is the electronic synthesis of the controller. We will obtain this electronic netlist in HSpice format (Figure 6), from which we can use a layout tool to obtain the final mask for analog implementation.

Figure 6 shows a hierarchical HSpice netlist of the controller, with nodes X0-X3 representing nodes M₁, M₂, I₁, and I₂ and their respective interconnections. We also can see the electronic netlist description of each of two of these modules (nodes I₁ and M₁). Modules corresponding to nodes I₂ and M₂ have the same structure as I₁ and M₁.