Galil Motion Control
Mountain View, CA
Jacob Tal and his partners founded Galil Motion Control in 1984. His current research interests include digital motion control, optimization systems, and design methods for high-performance systems. Hewlett Packard Laboratories employed Tal from 1978 to 1980, and he spent his time there working on new methods of motion control. From 1970 to 1978, Tal taught at the Department of Electrical Engineering, the University of Utah, Salt Lake City, UT. During this period he worked as a consultant and did research on optimization of motion control systems, phase locked loops, and digital motion control. Tal received his Ph.D. in electrical engineering from the University of Minnesota, Minneapolis, MN, in 1970.
Size doesn't ensure success in a field like motion control. Smaller companies can succeed in technically challenging areas if they offer customers something unique, says Tal.
Design News--Where does Galil stand in the ongoing debate over proprietary versus "Open" protocols?
Tal: We certainly stand on the open side, because we believe that the industry is going modular. Today you could buy an amplifier from one company, a motor from one company, a power supply from a third, a controller from another one, put them together, and within a few minutes they work. There have to be standards that allow the flexibility of mixing and matching components.
Q: Motion control can be an amorphous concept. What is your definition of motion control?
A: On the lower end, motion control stops at the motor. I don't consider couplings, x-y tables and gears as representing motion control. They're components. On the upper end, amplifiers, motion controllers, and software programs that generate the motion are truly motion control. There would also be a lot of software programs that are not truly motion control but are necessary--like automatic tuning--and also functions for activating input devices and output devices that are related to motion. But that's not the true hard core.
Q: Why should a company purchase a motion control system instead of designing its own equipment in-house?
A: Let's start with the assumption that most OEMs use less than 1,000 motion control cards a year. Often an engineer in a company like that would come to his boss and say: "Hey, I can do that! If I count the parts on the card it would cost me $200 or $300." But what is the real cost? To design a motion controller, if you know what you're doing, takes a couple of man-years. So it's a very significant effort. There is also an element of risk, because a lot of people start motion controllers and never finish them. After two years or a year you may find out that it never materialized, and the product has missed a year or two from its cycle life in the market. And that is the real cost.
Q: What's most important: software or hardware?
A: Most people only use 20 to 30% of their system's capability, and they stop there because that's what they need. In most cases the hardware is not important, even the software is not very important. What's important--in most cases--is ease of use. Because people want to be able to get the system to work in two hours and not in two weeks. Now once the application gets to be more demanding, you need to have both hardware and software that are more sophisticated.
Q: How can your small company continue to compete against the big firms that sell motion control equipment?
A: When we started out, some larger companies like Hewlett Packard and National Semiconductor entered the field. Everyone expected that they were going to crush Galil and similar small companies. But one thing that became clear is that the motion control business requires a different type of company. Supporting position control requires a level of attention that's extremely difficult to maintain on a large-company basis. All the leaders in motion control are small companies of our size. At the present time I don't see large companies stepping into this arena, and I'm not sure that they would be successful if they did.
Q: What improvements in sensors are necessary to advance the motion control industry?
A: Obviously the most important thing is resolution. And we've recently seen some linear sensors that can resolve to microinches. At the present time we don't have a lot of good sensors for sensing load, tension, and so on. Typically we use potentiometers and dancer arms. If there were very precise sensors that could detect those parameters, that could help achieve accuracy in very demanding applications.
Q: What's necessary for an engineer to successfully found and operate a technology-based business?
A: A lot of engineers think that because they're excellent engineers they're going to succeed. And it's not enough. You have to be good or very good in complementary areas like finance, marketing, and sales. Realizing that you need a team that covers all aspects is essential. And within that team there has to be at least one area of excellence. Engineering could be mediocre if you have outstanding sales. Between sales, marketing, and engineering, you have to bring something very unique to the table. And above all, the most important thing to have is customers.