History of CNC Machining: How the CNC Concept Was Born

Posted on January 14, 2013 by Marty by Marty

Computerized Numerical Control, or CNC as we all now it, came about shortly after WWII as a result of the
aircraft industry’s desire to produce more accurate and complex parts. Below is a partial reprint of an article from
the August, 1996 issue of American Machinist that explains the history of CNC much better than I ever could:

“Numerical control as a concept developed in the mind of John Parsons as a way to produce integrally stiffened
skins for aircraft, and this led to a series of Air Force research projects at the Massachusetts Institute of
Technology, beginning in 1949.

“The initial planning-and-study phase was followed by the construction of an experimental milling machine at the
Servomechanisms Laboratory at MIT. Prof J.F. Reintjes, director of the lab, James O. McDonough, Richard W.
Lawrie, A.K. Susskind, and H.P. Grossimon were the people involved in the research.

“A 28-in. Cincinnati Hydro-Tel verticle-spindle contour milling machine was the starting point. It was extensively
modified: all of the table, cross-slide, and head drives and controls were removed, and three variable-speed
hydraulic transmissions were installed and connected to leadscrews. Each transmission would produce, through
gearing and leadscrew, a 0.0005-in. motion of the table, head, or cross-slide for each electrical pulse received
from the director. A feedback system was provided to make sure the machine was doing what it was told. A
synchronous motor geared to each motion generated a voltage response to movement; this was sent back to the
director and compared with the original command voltage.

“By 1951, the system had been assembled, and application studies were begun. By 1953, enough data had been
assembled to indicate practical possibilities that could be developed. A detailed 24-page report on the process
that appeared in American Machinist on Oct 25, 1954, started a flurry of further development. [...] But it was the
initially more awkward, less accurate prototype at MIT, which employed a Flexowriter and its eight-column paper
tape, a tape reader, and a vacuum-tube electronic control system that was to become the prototype for the
developments that followed.”
The article included a photograph of the prototype machine with the heading “Pioneering setup at
Servomechanisms Lab at MIT had control system that surrounded modified milling machine it controlled.” In the
photo, you can see the control systems, which consisted of metal cabinets the size of school lockers, extends 12-
15 feet (my guess) and is filled with electronics. The controls were larger than the mill itself! A sign on top of
these “lockers” is also visible, reading “PILOT SERVO OUTPUT UNITS.” Underneath there are the signs “Table,”
“Head,” and “Slide,” each corresponding to a “locker” that the respective sign was above.

The article continues:

“Parsons Corp. [founded by John T. Parsons, who had been introduced earlier in the article] had already
developed a system for producing helicopter-blade templates by calculating airfoil coordinates on an IBM 602A
multiplier and feeding these data points into a Swiss jig borer—rather than laying out the job manually […] Late in
1948, the Air Force sent a team to the Parsons plant in Traverse City, Mich. There they saw the technique being
used in producing templates from data on punched cards.

“For such significant technical contributions to the field of numerical control, John T. Parsons in 1968 became the
first recipient of the Numerical Control Society’s Joseph Marie Jacquard Memorial Award. And in 1975, the
Society of Manufacturing Engineers awarded him a plaque naming him ‘The Father of the Second Industrial
Revolution.’”

What Does a CNC Need to Make it Work? The Theory from a Different
Point of View
This was considered the first “true” CNC, but before this there were many incarnations of motion control. Many
were considered NC machines and used punched Mylar tape or hydraulic valves and cables and summing levers
or cams. Before coming to the CNC world I was—and still am—a licensed aircraft mechanic. I worked for the
now-defunct (not my fault) TWA, specializing in the Boeing 747 aircraft. The reason I bring this up is, although
the aircraft industry is the motivating factor in new manufacturing technology, they are slow to adapt it into their
aircrafts. An example of this is the landing flap control system on the 747. I’ll explain how this works, then explain
its function in early motion control systems.

Landing flaps are the large extensions on the rear of the wings. During flight, they are tucked in to make a
smooth and aerodynamic wing surface. During takeoff and landing, they are extended to provide more lift at
slower speeds. At takeoff, they are extended about halfway to offer lift but without so much drag as to retard the
acceleration of the plane. During landing, they are extended all the way to allow lift at slower speeds and to
provide the slowest touch down speed possible. This is how they work:

There is a lever in the cockpit connected to a drum, or wheel. On this wheel is a steel cable; this steel cable
meanders its way through the fuselage over pulleys, pressure fittings, and guides and finally arrives at the
hydraulic control power pack in the aircraft’s wheel well. Here, the cable is connected to a small summing lever.
When this lever is moved, the hydraulic fluid will flow at 3000 PSI to a hydraulic motor rotating it forward or
reverse depending on the way the lever was moved. As the motor rotates, it turns a ball screw connected to the
flaps’ carriage. On the flaps’ carriage there is another cable that comes back to the power packs summing lever,
called the null lever. When the flap handle is moved, the hydraulic fluid flows and the flaps move. When the flaps
reach the desired position, the null lever chases down the summing lever and stops the fluid flowing—the flaps
then stop. So, we have a command lever to start motion and a feedback system to stop motion. This is exactly
what happens in a modern CNC machine.

The way it was used in early NC machines was, instead of a pilot moving a lever, the lever would ride on a cam
that would move it back and forth, causing the motor to rotate and move the slide of the machine. A special note
on this type of system is that, like early CNC systems, if the feedback cable would break, the null lever would not
stop the fluid flow and the system would “run away.” The same as if an encoder would fail on an electrical CNC
system. Modern CNC systems are too smart for this to happen, but in the early days it was always interesting to
see what you would find after a run away!

I’ve been away from the aircraft maintenance industry for nearly 20 years, but the use of electrical motion control
was just beginning in the aircraft world back then. The Boeing 767 was fairly new toward the end of my time, and
it was using electrical controls for flap and flight controls. This was the so-called “fly by wire” system that got a lot
of press at the time. Now it is commonplace in all aircraft. Another note on this system: every control system had
two complete and separate motion control systems—if one failed, it had a complete, 100% backup.

CNC: The Essential Systems

What all motion control systems have in common is a command function of some sort, a drive or motion system,
and although not necessary, a feedback system. The command function could be digital, analog, a flap handle, a
cam follower, or a host of other devices that say “Go.” The motion or drive system could be an electric motor,
hydraulic motor, a brake, a clutch, a cylinder, a valve, or any combination thereof. It is the device that makes
something move. Lastly, we have (or have not) a feedback device. We are primarily involved with high precision
machine tools, so we will always have a feedback device, most commonly an encoder of some sort.

Motion control systems are used in many industries. For example: a rock quarry. At the quarry, an operator
pushes a button to start a conveyor. The conveyor is full and has a heavy load. The operator might have a dial to
control the speed of the belt. He or she might start the belt very slowly, then increase the speed once the load is
moving. This is the command element of the system. The operator then watches the belt to be sure it’s operating
at the correct speed—this would be the feedback system. After it’s running, there could also be a sensor that
detects that the belt is moving or even how fast it’s going to relieve the operator of the duty of watching the belt.
This is also a complete motion control system.