In manufacturing, good design generally creates long product life. Every day, shops use tools, workholders, ancillary equipment and numerous other products that have undergone few, if any, fundamental changes since they were invented--way back when. It's not that shops disdain new kinds of equipment, but rather, in many cases, the original was so right it takes a long time to improve upon it.
High volume grinding operations are a case in point for the longevity of good design. It's estimated there are 10,000 hydraulic and mechanical ID grinding machines, with a basic design changed little since the 1960s. They're still in operation cranking out bearing races, valve lifters, CV-joint components, U-joint cups and numerous other applications. They are fast, accurate and very reliable. For years, these machine tools have had no productivity peers.
Machines like the Heald CF grinder and Bryant's Centalign and 2209 are workhorses of an industry where cycle times are measured in seconds and volumes in six figures. Breaking into this market with machines that can compete has been a frustrating quest for many builders.
Basically three machine design ingredients are needed for efficient high volume ID grinding: machine stiffness, accuracy and speed. Only recently have these criteria been packaged in an electromechanical design capable of surpassing these old mechanical and hydraulic machines.
To find out about the technology that's been pulled together to offer shops a significant productivity improvement to replace their bread and butter machines, we talked to Bryant Grinder Corporation (Springfield, Vermont).
Why Consider Change?
Shops have seen demands for rapid delivery and documented quality increase. At the same time, traditional cushions of overrun and inventory are all but eliminated. Line-ready workpieces are what customers want and, they want these workpieces on demand.
At one time, a customer would order a year's worth of parts and a shop would set up to run them all at once. Now, just-in-time programs break these orders up so a shop is now asked to run the same job six times a year rather than once. That's six setups versus one. And, of course, the unit price of the part is the same or less.
Shops are seeing a change in customer demand for quality too. It's no longer acceptable just to make good parts. Part "goodness" must be documented for manufacturing process and quality verification procedures. Statistical process controls (SPC) capable of 1.67 Cpk and higher are demanded by suppliers. Shops that can't demonstrate such predictable process control may find themselves out of the running for many contracts.
Meanwhile, the installed base of hydraulic and mechanical machine tools is aging. Many have been rebuilt repeatedly. While they are still capable of producing very good accuracies and short cycle times, shops are finding they are reaching the limits of performance capability improvements.
Compared to many of the grinding machines made by Bryant and other builders, the high volume ID grinder is a relatively simple machine tool. It's generally a two-axis machine that feeds a wheel into the workpiece using a Z-axis move. Once in the work, the X axis moves the wheel into contact with the surface to be ground. Cam driven mechanical or hydraulic actuation is used to move the axes.
These machines operate in a world of high volume, family of parts applications. For example, a bearing race--a typical application--is a relatively simple part geometry. What gets complicated is the number of sizes and shapes that a bearing race can be.
One area of machine performance improvement is in changeover from one job to the next--flexibility. The first technology that comes to mind for flexibility is CNC. Rather than changeover cams or mechanical stops and shim them in to get the correct stroke from the wheelhead, CNC should allow the operator to accurately set stroke electronically--saving significant setup time.
But the flexible CNC approach, as applied to other grinding machines, hasn't transferred easily to high volume ID grinding. The reason is simple. It's too slow compared to the mechanicals. Setup is fast, but processing speed for a traditional grinder CNC is generally inadequate for high volume ID work.
The advantage of mechanical machines is response of their actuation system. It's virtually instantaneous. Stroke length, dwell and acceleration/deceleration are all controlled by the shape of the cam or the position of mechanical stops controlling the axis. It works very quickly and accurately.
If a workpiece with 20 moves can run in 5 seconds and the lot size is 50,000, the job will take approximately 69 hours, straight cycle time, to run on a mechanical machine.
According to Bob Harrison, vice president of sales and marketing for Bryant, a traditional grinding machine CNC takes approximately 0.1 second processing time for each program command. For the same 5-second job, using a traditional grinding machine CNC will add 2 seconds to the cycle for processing 20 commands.
Now, you're looking at a run time of 97 hours. Even the setup savings provided by CNC over a cam or hydraulic machine (usually minutes versus hours) can't offset a run time differential of 28 hours. "This is why mechanicals have hung in there so well, and why most machine builders haven't tried to develop a traditional CNC-based replacement for these older machines," says Mr. Harrison.
Confluence Of Technologies
So, what does it take to build a machine to compete with aging mechanicals and take advantage of future technology developments? "Interestingly, in the last couple of years, our electromechanical ID grinder wish list has become a `to do' list," says Mr. Harrison.
What's made this possible is application of emerging technologies into a grinding machine that not only competes with the mechanicals but actually outperforms them. Only recently have the necessary technologies to do this been available.
Many of these new technologies are electronic, such as more responsive servodrive motors, quicker servo updates and control architecture that processes data faster. Job one in developing an electromechanical equivalent is to create a motion control system that can deliver very close to instantaneous response and a machine structure stiff enough to maintain accuracy at faster grinding speeds.
Other improvements involve "tightening" the mechanical components on the machine. Way systems, stable bed construction, independent slide design--placing workhead and wheelhead close to the bed--increase wheel-to-work stiffness by a factor of four over the mechanicals.
"It's been like a puzzle," says Mr. Harrison. "For a machine tool to operate optimally, all of the various interrelated components must work harmoniously. For example, more responsive servomotors are great but can't be fully applied if control architecture doesn't feed positioning data fast enough. High speed spindles allow the grinding wheel to turn faster but limits on feed rates, because of insufficient machine stiffness, may prevent taking full advantage of high speed wheels and spindles."
If Not CNC, What?
Bryant doesn't call their new machine control a CNC. It's a distributed control. Semantics aside, the control architecture capable of making this happen involves a direct communication line between the PLC (programmable logic controller) and the motion control board.
The key however, is the placement of a memory chip on the motion control board itself. In effect, this is a program buffer that stores the current program directly on the motion board. Bryant uses the commercially available PMAC motion control board from Delta Tau (Northridge, California).
In operation, this arrangement of components reduces the calculation time for a servo update from 0.0085 seconds (8.5 milliseconds) to 0.00044 seconds (0.44 milliseconds). This lets the control handle servo loop updates 20 times faster than previous grinder controls. But the "box" is only a part of the technology package being brought to bear on the processing speed problem.
Faster Servos Too
Servomotors are a critical component of providing speed and accuracy to a machine tool axis motion. Advances in servodrives and the encoders that keep track of their position allow full use of the increased control processing speeds.
Electronics are key to speeding up the data flow from the motion control card to the servomotor. The faster this communication loop can be processed to high response motors, the closer a grinding machine's axes come to instant motion response.
An operator with a handwheel knows to crank it fast to move a tool in position to cut (take up the air) and then to slow the approach as the tool gets close. Mechanical machines perform this routine using the slope of the cam to duplicate the handwheel motion.
Electronic duplication of this mechanical motion is not simple. It takes some powerful mathematics, in the form of algorithms, to output an electronic S-curve or parabolic acceleration/deceleration signal to the servomotor telling it when to slow down or speed up. Without these acc/dec routines the axis motion would be too abrupt. With the power and responsiveness of available servomotors, a simple on or off signal would result in very jerky motion and inaccurate grinding.
These servomotors have clock processing speeds (how fast they send position update signals) that have increased from 1MHz to 10MHz. Together with the new control architecture, built around the motion control board, a 20-time increase in servo loop update speeds has been achieved.
"That's where the technology is today," says Mr. Harrison. "The trend in electronics is faster and faster processing speed. That's why we've applied open architecture in our machine tool controls. With it, upgrading to better components when they are available is relatively simple."
With all this zooming around because motors and data can go faster, how about accuracy? According to Mr. Pat Harrington, director of engineering, "Increasing the speed range of drive motors impacts their selection. Going fast is fine but critical grinding accuracy comes from low speeds, and drive motors must be able to move precisely in very small increments."
A slotless motor design is used for this. "Slotless motors eliminate cogging," says Mr. Harrington. Cogging is caused by the magnetic characteristic of electric motors. "If you slowly turn the shaft of a servomotor, you feel slots as the rotor moves from one magnetic field to the next. At high speed, cogging doesn't show up, but at slow feeds it can cause a quality problem on the final grinding pass," says Mr. Harrington.
Slotless design places the stator windings farther from the motor magnets. This increased distance eliminates the steps from one field to the next and gives very smooth operation of the servomotor at slow feed rates.
Consistent machine tool accuracy comes from how well components maintain relationships to each other. It's a machine's stiffness. When the old mechanical ID grinders were designed, stiffness was measured by the pound. Weight meant rigidity.
Not so anymore. Finite element analysis and a better understanding of thermal dynamics has led to a strategic use of mass in place of tactical. Machine bases and sliding components weigh less because iron is put in the right places, but these beds and slides are stiffer. Reduced mass for sliding components is essential for higher acc/dec rates.
Machine stiffness and thermal stability are important for the use of superabrasive grinding wheels. More aggressive cuts and higher speeds and feeds put mechanical stress that is significantly higher on machine structures than older machines were designed to handle. Superabrasive wheels require very small dress increments. According to Mr. Harrington, "Most older machines cannot work in these small increments."
A Better Way
There are two fundamental designs for internal grinding machines: compound slides and independent slides. Compound slide machines provide greater travel ranges and generally accept a larger range of workpieces. In this design, the workhead is fixed to the bed. Motion comes from the wheelhead, which is carried by the X and Z axes stacked one atop the other. The wheel can then perform contouring moves relative to the dresser unit or the workpiece, so more surface geometries can be ground.
For high volume ID grinding, extended range and interpolated axis moves are less important. Bryant and other builders use an independent slide design for their high volume ID grinder. This design locates the moving components, wheelhead and workhead, close to the machine bed and ballscrews, improving stiffness while lowering the center of gravity.
For precision ID grinders, frictionless hydrostatic ways are generally the system of choice. A film of oil keeps moving components from direct contact, resulting in virtually no friction or wear. Critical to the efficiency of these ways is the gap between surfaces.
To create a minimum gap between the way and slide, Bryant uses a cast material injected between the two components. This material duplicates the shape of the way so the slide can be fit very tightly on it. "Reducing the distance between these surfaces reduces the flow of oil which lowers its temperature," says Mr. Harrington. "The flow of hydraulic oil also is used to maintain thermal stability of ballscrews and the machine base."
Machine As Part Of A System
It has been only recently that the technology needed to produce an electromechanical ID grinder capable of competing with the old machines has been available. To make it happen, grinding machine builders have taken the best technologies available from machine component design, servo-control, thermal dynamics, PC-based electronics, superabrasive grinding wheels, high speed spindles and sophisticated motion control algorithms to create a machine that will outperform the hydraulics and mechanicals by a sufficient margin to justify the price of a new machine.
However, simply building a faster machine tool is not sufficient. Machine cycle time is only one measure of productivity. To get a complete performance picture, it is necessary to look at a process in total. Evaluation of performance should factor in setup and teardown for a job run as well as gaging time for produced parts.
A holistic view of a grinding process gives shops a better look at ways in which total per-piece times can be reduced. New machines, like Bryant's and others, are designed to take high volume grinding operations to the next plateau of productivity.
In addition to providing cycle times competitive with the mechanicals, these new machines also incorporate ancillary functions like in-process and post-process gaging for SPC data collection, quick change tooling, and closed-loop adaptive control of the grinding process to make parts fast and with predictable quality. If high volume ID grinding is your business, a look at what's new with these machine tools might be worthwhile.