Laser Focused

Two years ago, TeoSys Engineering LLC (www.teo-sys.com) and Photoscribe Technologies (www.photoscribetech.com) developed an Excimer laser-based diamond etching system. The deep UV, pulsed laser effectively delivers short bursts of energy, burning microscopic pits (1 to 2 microns deep) onto a gem's surface to form the markings. But its $300,000 price tag limited its use to major diamond labs.

Once the companies determined that a machine with a $100,000 price tag could open up a substantially larger market, particularly at shopping mall jewelry shops, the challenge was on for engineers: How to reduce the cost of the machine by two-thirds, while making it easy enough to use by people other than skilled technicians and without reducing its core capabilities. "This was an exciting opportunity for us. A lot of times engineers don't have the chance to make a system better," says Engineer Dana Church, co-founder of TeoSys. She has an MS in Computer Engineering from the National Technological University in Fort Collins, CO.

Church and Managing Partner Emre Teoman developed the LMS 5X0-a desktop-size machine they claim is the smallest Excimer laser micromachining system in the world-in just five months, an impressive time span given the engineering constraints. "It's relatively easy to build a machine of this type for sophisticated users, but it's much more difficult to make a machine that someone can learn to use in a few hours," says Teoman.

Leaving no stone unturned, literally, in their push to cut cost out of the design, the two engineers altered many facets, from switching to a less powerful laser to incorporating a low-cost illumination system and an air-filtering machine that can be bought in retail stores (see sidebar next page). While these design changes made a big dent in the overall budget (the laser by far is the biggest cost component), a redesign of the machine's motion system involved some of the most interesting engineering trade-offs.

The LMS-5X0 creates marks by moving the diamond in the X and Y axes while the laser, which remains fixed, is firing. Precise motion control is a necessity for etching lines less than 5 microns deep onto the gem's "girdle," a circle that surrounds the underside of the diamond. A goal of the etching process is to make marks that aren't discernable to the naked eye, so the lines are often faint. It's not unusual for gemologists to want them a bit darker, so the inscriber makes it possible to repeat the engraving process with such fine precision that the original lines are only made deeper, not wider. Coordinated motion is mandatory, particularly in the moves that involve circular interpolation and vector positioning.

The motion control system therefore had to deliver the required precision, but be cost effective enough to meet the pricing goals. Its predecessor, the LMS 2500, had a full servo control loop, using 2-mm, high-precision, ground ball screw stages driven by a brushless dc servo motor, encoder feedback, and control card capable of handling eight axes of motion. But the machine offered far more functionality than the low-cost 5X0 family needed. Not to mention a 4-inch travel distance.

Although the maximum speed of the servo motor used is 115 mm/sec, Church points out that the laser limits the frequency to 350 Hz, which translates to 350 microns per second. "Because our vectors are generally 50 microns or less, we usually do not even go that fast in our application. In fact, with acceleration and deceleration, we never even get to a speed of 200 microns per second."

A stepper motor (really a hybrid of a variable reluctance and a servo motor) looked like a winning alternative to the engineers. Given the extra poles, stepping motors can be used in simple open-loop control systems, yet achieve a level of accuracy and repeatability that is comparable to servo. In small frame sizes, stepper motors (including motor, cables, and drive) cost about one-half to one-third less than servos, says Rich Lenzing, a product manager for stepper motors at Danaher. "The real advantage is that engineers don't need to pay for a feedback device. Moreover, if you look at the servo motor applications that fail, it's typically because of the feedback. A stepper motor will have less than half the failure rate of a servo."

TeoSys engineers selected a 0.9 degrees stepper motor (Oriental Motor) with a resolution of 800 steps/revolution for their low-cost design. An advantage of this motor type compared to the more common 1.8 degrees stepper motor, says Nick Johantgen, engineering manager of stepper motors at Oriental, is the increased stiffness. "That means that it takes more force to twist the shaft from where it wants to be, which translates into greater positioning accuracy than a 1.8 degrees stepper motor." He also notes that steppers are particularly appropriate for applications involving short moves. "Servos have all those gains and settling times, where as a stepping motor can move to the spot you want and position more quickly," he says.

By combining the motor with a custom leadscrew and precision track roller guide block, Church says that engineers were able to achieve a resolution of 1.526 microns per step. Through experimentation, they determined that the repeatability is 1.5 to 2 microns over a travel distance of several millimeters.

Though this low-cost system does not have the repeatability of the higher-cost system, Church says that a tradeoff in speed allowed the system to achieve acceptable performance. "Because the marks we make are so small, we have to contend with rapid acceleration and deceleration. The faster we go, the more error we introduce, which we cannot compensate for without encoder feedback. But that's not a limiting factor, since we are laser-limited to 30 microns per second and the application involves very short vectors (& 25 microns). Our average character line width is 10 microns, so a 1.5 to 2 micron error when we remark a string of text is not significant."

Engineers do not expect that the fact the LMS 5X0 is slower than its predecessor will be a problem. "Most users don't want to do a diamond every 30 seconds, they're happy with 3-4 minutes. This is not like making doughnuts," Teoman says. He notes that slowing down the speed makes it possible to use less powerful lasers, which also help reduce expenses.

Coordinated Motion

An economical motion controller (the DMC-1822 PCI controller by Galil) provides coordinated motion on two axes, though it can handle up to four axes. It provides both circular interpolation and vector positioning-crucial capabilities for etching precise circles, arcs, and other marks on the diamond. Configurable for either steppers or servos, the low-cost (approximately $895) PCI controller is based on a 32-bit Motorola 68331 processor. A key reason for picking the board is that it did not require servo loop control to provide the degree of repeatability that engineers were seeking.

"This PCI controller is one of our most popular control cards because of the acceptable trade-off in cost versus capabilities," says Lisa Wade, marketing vice president at Galil. "Even when a motion card is going into a million dollar machine, we're finding that design engineers are still sensitive to price. They don't want to have to pay for features and capabilities they don't need."

The motion controller works in conjunction with a proprietary board built around a PIC processor from Microchip Technology. Designed by Teosys engineers to work in a PC environment, the board controls laser triggering, user interface, motor drives, and mode select switch.

Programmability was a necessity, since identification marks are generally stored in a CAD format. These DXF vectors are then converted to the motion programming language recognized by the motion control card using a translator that Church wrote.

Though motion control was instrumental to achieving both accuracy and cost cutting, Church notes that the system uses the simplest type of motion control out there. "Keep it simple stupid was our theme during the design," she says. The only feedback comes from the video imaging system, which does the "very difficult" task of showing both visual and UV data merged. The merged images are displayed in real time so users can see the diamond on the screen while it's being marked. That's important to ensure correct etching on the diamond.

Repeatability is assured because software records the movement of the gem during inscription. Successive passes for enhancing the depth of the image are controlled by the software, so any deviation is minimal. The software is written in PICBasic, a subset of Basic designed for Microchip's embedded applications. One focus of the software development was figuring out a way to prevent users from making mistakes during the etching process. "We specifically designed the system so that it doesn't leave a lot under user control. The software has steps one through ten, and then you're done. It even prevents the operator from doing steps out of sequence," says Church.

Another plus: Like the original, the new machine uses a patented, intra-cavity variable aperture, giving operators greater beam control over a large focal depth. This builds a degree of "forgiveness" into the system to compensate for less-than-perfect focusing, while achieving the goal of perfect etch marks on the diamond.

Contributing writer Terry Costlow can be reached at[email protected].