Customer Solutions

DOE Automates Electrospark Deposition Program with LabVIEW

Author(s):

Jeffrey A. Bailey, Pacific Northwest National Laboratory (PNNL)

Industry:

Machines/Mechanics

Product:

Data Acquisition, LabVIEW, Motion Control

The Challenge:

Developing a viable automation/control program and associated hardware for the Electrospark Deposition (ESD) Strategic Environmental Research and Development Program.

The Solution:

Applying the National Instruments LabVIEW integrated solution platform and its considerable computing power to solve the previously elusive control scheme.

Bearing Sleeve

Harnessing ESD Technology
Electrolytic hard chrome (EHC) is an industry-standard technology for depositing a protective metallic surface on a variety of metal pieces, like 1957 Chevy hubcaps, gears, turbine blades, bearing housings, motor shafts, and hydraulic cylinders. Although EHC does provide an attractive, low-cost, and reasonably durable surface, it has two flaws. First, the EHC coating is mechanically/chemically bonded, not metallurgically bonded, to the target metal surface. This is more characteristic of a painted-on coating than a welded-on surface. The coating can peel off the surface. Second, the EHC process requires many processing chemicals and involves hazardous chemical byproducts, so a number of environmental issues have to be addressed. EHC cannot reasonably be used in a small, confined workspace.

ESD is one of the more promising emerging replacement technologies. The ESD process involves producing a series of short electric pulses through a moving electrode energized by a series of capacitors as it is short-circuited momentarily with the base material. During the generation of the arc, small particles of the electrode material are melted, accelerated through the arc, impacted against the molten base metal substrate, and solidified rapidly. Buildup occurs incrementally. ESD produces a true metallurgical bond. ESD requires no processing chemicals and has no chemical byproducts. But, ESD also had a flaw. Because of the short duration and inconsistent magnitude of the application pulse, no automation team was able to electronically control the ESD probe pressure. For 20 years, various research groups around the country attempted this. Then the PNNL automation team was asked to try to solve the problem using NI LabVIEW as a format. Patents are now pending for our unique solution.

The most viable control scheme developed prior to ESD was a series of springs and weights contained in an enclosing case. The application probe had to be mounted vertically in the spring-loaded case. The operator added small weights (actually, small washers) to or removed weights from the top of the case to change the application pressure. The operator changed weights based on how the arc looked and sounded. The ESD mechanical operation was restricted to vertical mounting in a gravity environment where the operator could hear and see his work.

Smoothly Converting to an NI System
Our first automation concern was converting all of the existing equipment and controls to a format compatible with LabVIEW and hardware compatible with National Instruments. Most of the control programs were in vendor-supplied programming formats or other unsuitable programming formats. We also had a three-stage (XYZ) microcontrolled automation table. There were two limit switches on the spring-loaded case that were being programmatically monitored, but there were no other attempts at parameter/data acquisition.

Initially, we used LabVIEW, a PCI-7344 four-axis stepper/servo motor control card, and the LabVIEW motion control tools to rewrite the motion control portions of the application. Because we used NI motion control subroutines and hardware that were completely compatible with LabVIEW, we were able to sharpen and simplify our motion control algorithms in a very short time. The ease and variety of programming choices also resulted in a noticeable smoothing of all motion. The smoother motion meant that our application probe did not bounce as much during metal deposition.

We used a PCI-6035E multifunction I/O board to integrate our data I/O into our new LabVIEW application. We removed the two existing limit switches because there was just too large a gap between them to provide the mechanical control we wanted. We inserted a single optical switch and monitored it with every control loop. With the new switch installed, the mechanical position was either too high (not enough pressure) or too low (too much pressure). As such, the mechanical position was corrected in every single loop. Mechanical control was dramatically sharpened.

Capturing the Parameter
Next, we began our search for an operating parameter that changed in direct proportion to the probe pressure. It quickly became apparent why no other automation team had met with any success. Because of the short duration of the pulse and the low pulse-to-noise ratio, all RMS measurements (current, voltage, and power) were useless. The individual and average pulse magnitudes also varied wildly with surface debris, temperature, and pulse bounce. We looked at integration of signals, capacitance, and standard deviations. Nothing seemed to be able to follow the probe pressure. We were discussing monitoring the various frequencies of light emitted during the spark and looking for a “probe pressure based on light.” At this time, we noticed that the minimum pulse magnitude never seemed to repeat a value (the uncontrollable parameters always changed with time). However, the maximum pulse magnitude did seem to repeat at a specific maximum value within every dozen or so pulses. This meant that every so often, a pulse was delivered where all of the uncontrollable pulse parameters were at a minimum. The highest pulse of several consecutive pulses was the same.

We rearranged the program to record enough of the probe current to capture 10 pulses and then to select the highest value in the array. This peak followed the pulse pressure very closely. We had our parameter. Later, we found that we only needed to record five pulses to capture the parameter. The high-speed data acquisition was critical to finding this relationship. We were able to capture the data stream fast enough that we lost none of the pulse information to aliasing. In our next-generation machine, we will use a control card programmed for 2 MS/s to further focus this application.

In December of 2003, Battelle applied for a patent covering these electronic controls for ESD. Battelle and the automation team are currently moving into the next stages of research into the applications of ESD.

For more information, contact:
Jeffrey A. Bailey
Battelle Pacific Northwest Division
P.O. Box 999
Richland, WA 99352
Phone: (509) 375-6346
E-Mail: jeff.bailey@pnl.gov
Web: www.pnl.gov