FUZZY LOGICS Key Words:

NDT, PCB, FPGA, LUT, Linguistic Hedges, MISO, Interferogram, Fuzzifiers, FLIPS, Topography. ABSTRACT:

Holography, the art of complete recording, is well suited for the field of Non Destructive Testing (NDT) during the manufacturing of printed circuit boards and the integrated circuits in the electronics industry. A precision topographical mapping system is generated using a double expose holographic interferometer for this research. This system is used to generate a three dimensional graphic image of a cantilever beam under stress, as well as Printed Circuit Boards (PCBs) and one Integrated Circuit package under stress. Various applications of human expert knowledge in a fuzzy logic controls environment are implemented using VHDL coding of a digital fuzzy controller on to a Field Programmable Gate Array (FPGA) for a rapid prototyping.
INTRODUCTION: This paper presents the work in developing a rudimentary system to non destructively evaluate the displacement of microelectronic surfaces, and then process the data using the principles of fuzzy logic in a controls environment. The basic concepts of a fuzzy logic knowledge base, i.e., a set of rules and an inference mechanism, are applied in the controls setting for the manufacturing. By defining the knowledge base and possible inputs, a Look Up table(LUT) can be formed that can convert input data into output tailored by the specified knowledge based using the Computer Aided Design (CAD) tools, the fuzzy controller is synthesized on to a FPGA. EXPERIMENTAL SETUP: In the system shown below, light exhibits the characteristics of both particles and waves. Illuminating an object with a coherent light source (plane wave), each point on the objects surface acts as a light point source. Thus the object generates the complex wave form that contains all the amplitude and

phase information introduced by the objects topography. By illuminating the object and the recording medium with the same plane wave, an interference pattern of all the optical data from the light diffracted by the object with the reference beam will be recorded on the medium (photographic film) as an intensity distribution. To calculate the information in the hologram, a reference co-ordinate system must be arbitrarily chosen. Assume the film is located in the xy plane and the reference beam propagates along the z axis. The object beam meets the film at an angle at the film z=0.

Multiple-Input Single-Output Fuzzy Controller: The following is the fuzzy controls problem. given input by the holography interferometer quantifying the relative displacement of a thermally stressed PCB, as well as quantifying the relative placement of the solder on to the pads, it is desired to determine whither or not to evaluate specific areas of PCB for solder pad integrity before shipping. Additional information is communicated in the fuzzy inference system through the use of linguistic hedges. The following are the fuzzy inference rules of inference system:

 if (very small) relative displacement and low solder error, then (Increase) probability of passing inspection. If (more or less median) relative displacement and median solder error, then (question/zero increase/decrease) probability of passing inspection. If (large) relative displacement and high solder error, then (decrease) probability of passing inspection. The assumptions made in the above rules are as follows; probability of passing has a range of [0,100] percent. relative displacement has a range of [0, 20] m. relative solder displacement has a range of [0, 40] m. inputs will be broken into 2 m and 4 m intervals, respectively. for a multi input single output (MISO) which has two inputs utilizes three rules. Resolving any two inputs into a single output control then requires that the two values be added together, and then divided by 2. The following figure is the raw intensity mapping of the interferogram created due to this mechanical stress.

By sampling the data at the intensity maximum peaks of this gray-scale image the measurement of the displacement of the cantilever is accomplished. The following is an example of the intensity mapping for a vertical slice of the grayscale image.

The calculation indicates that 29.56 bright fringes should occur for the displacement. Compared to the 30 bright fringes of the experiment, this yields 1.5% difference. This gives an error in measurement.

Here is a comparison of the raw intensity data with the smoothed cubic spline data.

The raw intensity mapping of the interferogram created due to this mechanical stress is recorded. By sampling the data at the intensity maximum peaks of this gray-scale image the measurement of the displacement of the cantilever is accomplished. The following three dimensional graph is generated for the cantilever beams plane of observation, using which shows the deformation of the due to the force applied. A similar method is applied for a resistive printed circuit board (PCB), a 7400 quad dual-input IC with biasing circuitry PCB, and the 7400 IC dip package.

MISO Fuzzy Controller Specifications:

A dual-input single-output fuzzy controller is developed as a member of the class of multiple-input single-output fuzzy controllers. Using the same methodology, the inputs to the fuzzy look-up tables where calculated off-chip using C++ program. For the MISO design improved flexibility in programming the range of operation for the controller is desired. Approaching the problem serially would satisfy the pin requirement, but the loading of the data is painfully slow, and the required logic causes a violation of on-chip resources. The solution is to implement these changes using the software. Loading generic specifications in the configuration file proves only for integers in the Synopsis synthesis tool, so all of the inputs for the fuzzifiers, defuzzifiers, and LUTs are tailored using the keyword constant in the VHDL behavioral code. Though the control environment cannot be upgraded on the fly, this design does allow the chip to be quickly re-programmed to the specifications of the control environment. For greater number of inputs, the design must be modified. As the number of antecedents increases, the design will at some point exceed the available resources of the FPGA, requiring redesign of the fuzzy controller, either to multiple FPGAs or an ASIC.

The fuzzy controller successfully synthesized and operated at a clock speed of 20E6 fuzzy logic inferences per second (FLIPS). There is a 4 clock cycle delay between the initial input and the first output for the pipeline.

CONCLUSION: Using an argon laser and a double-exposure holographic interferometric technique, differences in displacement of 0.26 microns in one direction are measured. The intensity map of the stressed object is used to generate a precise microscopic topographical mapping which can be evaluated for defects. Through the design of appropriate sensors this data can be evaluated using fuzzy logic in order to localize the search for problem areas in microelectronic chip wafers and printed circuit boards. Though the optical components require precision placement and an isolated environment in order to gather quality data, as seen in the high quality of the topographical mappings obtained. The marriage of holographic interferometry and fuzzy logic appears fruitful. REFERENCES: 1. 2. 3. Optical Methods of Engineering Analysis by GARY CLOUD. Introduction to Fourier Optics by Joseph W.GOODMAN. Holographic Nondestructive Testing by Robert K.ERF.