Computerized Rotary Table

A powerful feature of the Cartesian Display is the optional integrated Rotary Table. The system described here is one possibility. It is also a stand-alone system that can be used without the Cartesian Display in a manual mode or when connected to your microcontroller-based custom design.

The Table can be commanded from the Cartesian Display and/or from a TTY Terminal on a separate computer. It will also automatically move when the machinist rotates the part geometry which is useful for machining angled features on a part.

I chose the Phase II brand 8" table since it's the perfect size for a Bridgeport-scale machine and it's also fairly precise. This same table has been branded under other labels over the years.

Shown below is the full setup including the ruggedized controller box. The controller sits on the floor next to the machine and there is constant chips and coolant raining down on it, so it needs to be tight.

The table is retrofitted with a NEMA-34 stepper motor as shown below. The motor adapter is also available separately for those wanting to do the conversion themselves.

The rotary table controller core module

The controller board implements the human and computer interface and stepper motor control. The human interface is implemented as a USB connection to a Desktop computer such as a PC or Raspberry Pi or anywhere a TTY Terminal Emulator can be used to command the table.

The computer interface is an RS232 port with an RJ-45 connector. This is the interface that the Cartesian Display uses to command the table. This means that users can also develop their own computer interface to command the table. For instance, one could develop a touch screen interface for a custom controller when using the table as a stand-alone tool on your mill. The packet-based communications method is made available for anyone to use.

The control signals are meant to interface to the Digital 32-bit stepper motor driver available from Automation Technology, Inc.

https://www.automationtechnologiesinc.com/

The signal wires for PULSE, DIRECTION and ENABLE are deliberately short as shown.

A GPIO port is implemented with 4 inputs and 1 output. This port is intended for table interlock switches and a brake signal. For instance, the output could be used to actuate a pneumatic brake that acts on one or both table clamps. The inputs could be used for switches on the brake, the quill or anything that the user deems important. While I have yet to implement this equipment, the idea is to set and release the brake whenever the table needs to move. The interlock switches are intended to force the user to get the tool out of the work before rotating. Of course, there are times when you want to tool in the work during rotation and there is a way to override the interlock.

The 5V regulated input is intended to come from the output of the main power supply. The power supply units available from Automation Technology, Inc. conveniently offers a 5V output in addition to the power needed for the motor.

What does the Controller board actually do ?

Humans don't think in digital pulses, they use engineering units such as degrees. The Controller takes care of the calculations necessary to move the table with digital stepper motor pulses so the machinist can conveniently command the table using degrees (fractional or DMS) and other commands. It also handles acceleration and deceleration as well as backlash compensation. It has other nice features such as Indexing, jog mode and automatic table zeroing where the machinist can give the Controller two points along a straight edge and it will "do the math" to align that edge with the X-Axis and set ZERO.

Motor Conversion Solution

This is one possible solution for a direct drive motor adapter. Note that this design retains the original slip-ring graduated dial and the worm-screw clamp. The graduated dial is useful for reference. Only NEMA-34 motors or larger should be used. These types of tables have no bearings, thus they present significant parasitic drag, the motor needs to overcome this friction and have power to spare to move the table -while cutting or indexing.

Motor Conversion Solution

Shown here is the zero backlash universal joint. This is not a requirement; normally you would use a Beam type coupling. I implemented this special universal joint in an attempt to deal with the problem of reversing direction of small angles. I have not figured out the physics of this phenomenon yet where if you reverse the direction of the table and the delta angle is small, say less than 1 degree, the backlash compensation pulses do not seem to work properly. If such a move must be made, then I back off a few degrees and then go to the new angle; then it will be right on target. Of course the specific machining case would need to accommodate this move. The joint helped by cutting this error in half; but the problem persists. The error is very small and many users would never even notice it. Work is ongoing.

Accuracy of the Electronic Rotary Table

A great deal of work has gone into determining the precision and accuracy of this system. Unfortunately, the measuring instruments necessary to do it right are out of my reach. The Phase-II table advertises Indexing accuracy of 80". This probably means that any given angle you rotate to should be within 80 arc seconds. But with a high resolution micro-stepping motor driver, I have shown it to be much more accurate than 80 arc seconds.

CNC machines allow rotational axes to be programmed to 0.001 degree resolution. My system is also programmable to the same resolution; however, whether you actually attain that resolution is debatable.

Let's explore this by establishing some parameters. It is helpful to visualize the table positioning error in terms of physical linear displacement in INCHES as opposed to units like ARC SECONDS. I consider the useful working envelope of the Phase-II 8" table to actually be 10". Thus, machined features of a part could be up to 5" from the center of rotation.

First, what would +/- 0.001 degrees look like at 5" from the center of rotation. It's useful to make mock triangles to get a sense of the physical linear displacement at a given radius and angle.

These are just approximations, but they get us in the ball park. Sin and Tangent are almost the same for very small angles. Thus the Hypotenuse and the side adjacent are nearly identical even though my drawings are deliberately exaggerated. This is not rigorous analysis, but gives a useful way to approximate the relationship between units like arc seconds to what that means in inches at a given distance from the center of rotation.

Rotary axes on the finest CNC machines "may" be this accurate. Keep in mind we are talking about milling machines. Make a right triangle with radius 5" and an angle of 0.001 degrees as shown in the above diagram. The side opposite the angle would only be 0.000087".

The point is, no one is using a milling machine to achieve such accuracy. And we can get away with a lot less accuracy from rotary axes on manual machines.

With my system, even though we can program the table to 0.001 degrees of resolution, we are likely not getting there; and most importantly, it does not matter. For all practical purposes.

In practice, after retrofitting several 8" Phase-II specimens, I find the practical resolution to be about 10 times less than 0.001 degrees. In the diagram above, our mock triangle with an angle of 0.011 degrees would produce an error of about 0.001" at a 5 inch radius. This means that a feature machined 5" from center, for instance a hole that is bored, might be off as far as 0.001" in either direction. So we can estimate the accuracy to be about +/- 41 arc seconds.

Some table specimens have shown accuracy to +/- 20 arc seconds which would produce an angular error of +/- 0.0005" at a 5" radius.

But even +/- 0.001" is pretty good for a milling operation; and still it can only be achieved with some finesse and a carefully tuned table. Lucky we are not usually holding sub 0.001" tolerances on a milling machine.

In terms of precision or repeatability, with careful setting of the backlash pulses, the repeatability is almost not measurable. It can repeat much closer than +/- 20 arc seconds. It can be dialed in extremely close. I don't have the equipment to make these measurements correctly making it difficult to separate out other sources of error. Also, because of the inaccuracies in how the ring and worm are manufactured, the backlash is not the same for all locations 360 degrees around the ring. Usually it is measured in at least 4 places and averaged. This procedure is described in another section.

There is a pile-up of error sources in this system:

1 The ring and worm gears are only so accurate -and they are not ground, they are machined.

2 The ring and worm gears are always slightly eccentric.

3 There must be some end-play on the worm shaft.

4 The motor, itself, is only so accurate. The poles are not exactly spaced.

5 The motor coupling contributes some error by storing pulses in its elasticity.

6 The viscosity of the gear oil has an effect on accuracy.

This rotary table conversion system has been in use for 5 years with great success. The controller will work on any table with a 90:1 gear ratio along with the off-the-shelf components described above.

Bare Metal Technologies

Electronics and Software Systems for Machining

Computerized Rotary Table

Retrofitted Phase-II 8" Rotary Table and Controller

The rotary table controller core module

Inside the Controller

Motor Conversion Solution

Motor Conversion Solution

Motor Conversion Solution

Motor Conversion Solution

Motor Conversion Solution

Accuracy of the Electronic Rotary Table