Taig Lathe with Semiautomatic Traverse Feed

Lathe Front with Stepper Drive

For as long as I can remember, I wanted a lathe for my workshop. While I got pretty adept at adapting various hardware from the local hardware store, lack of the capability to produce custom bushings, axles and the like was clearly a major hindrance in building various projects. Plus, I confess that I just plain love to build and own tools.

So, I bought a Taig lathe from Nick Carter (great source!) and scrounged a motor from a local neighbor’s trash (I am, as a non-apologetic trash reuser, an embarrassment to my family). I really love the machine…it is an elegant design; the engineers put the money and precision components where it counts.

I am very new to metal working but really enjoying learning. And, of course, the Taig just begs to be added on to and souped up.

I have included a number of links to some sources of information that I used at the end of this article.

My Favorite Projects…

After using the lathe for a few months, a few thoughts and interests came together in my mind. I noticed that the quality of the part finish was dependent on how skilled I was at turning the traverse crank handle. I decided to motorize the traverse feed to address this issue but mostly because it sounded like a fun project. Fun because it could involve a wide range of disciplines and offered a chance to learn a few new things.

Now before I describe the results, here’s what this project is not about. It uses the existing rack and pinion traverse drive so is not as precise as a lead screw arrangement. It is only one axis, so it can’t do profiles. While it automates the traverse feed, it is not a general CNC design. It is a stepper drive for simplicity. There is currently no provision for backlash compensation (but it could be incorporated). It might not even be worth it in the sense of providing a practical feature (though I think so!). It was great fun and was accomplished on a modest budget.

The basic plan was to use a stepper motor mounted on the lathe carriage driving the pinion gear/axle. I figured I could fashion some sort of bracket arrangement to the mount the motor and couple into the pinion axle.

The resulting bracket can be seen in the accompanying photos. It picks up the left top of the carriage and the side skirt of the carriage. Because of the draft on the side skirt, I used a couple of set screws to push the bracket off of the skirt while pulling the bracket against the skirt with a button head screw.

After searching around for surplus (read cheap) stepper motors, I found one that seemed to fit the bill. 200 steps/revolution with 12 volt unipolar drive, new surplus (about $20). 12 volt was attractive because I could use a surplus computer (PC) power supply (also a "reuse" item).

So, using a couple of cogged pulleys and a toothed timing belt (polyurethane with Kevlar fiber) with a 4.5:1 ratio purchased from Stock Drive Products, I was able to achieve about 0.0003 inches per half-step (I decided to half-step the motor), a reasonable speed range and more than adequate torque. A future design (redesign the software) could go to a more sophisticated micro-step approach but I am not sure it would be worth the effort.

One of the design issues I puzzled about for some time was how to allow both manual control as well as automatic traverse control. I came up with a design that was fairly easy to implement and works quite well.

I bought a new pinion axle from Taig and machined (on my lathe, of course!) a brass hub that is held onto the axle with small set screws. I threaded the center of the hub (beyond the end of the axle) to accept a knurled knob/threaded stud. I machined a new control wheel that fits onto the brass hub. I machined the large cogged pulley to also fit (rotate on) the hub.

The basic idea was to tighten the knurled knob thereby squeezing the cogged pulley between the control wheel and a flange on lathe side of the hub. Once tightened, the motor would drive the pinion axle via the cogged pulley.

To provide coupling to the pinion axle for manual operation, I drilled a hole at the boundary of the hub OD and the control wheel ID and inserted a short length of dowel rod to serve as a sliding key on the hub. The control wheel can slide over the key so that it can squeeze down on the cogged pulley when the knurled knob is tightened. When the knob is loosened, the cogged pulley can remain stationary while the control wheel is manually rotated, coupling via the key to the hub and therefore, the pinion axle.

One minor tweak I made after constructing the wheel/cogged pulley mechanism is to add three spring loaded hardened balls between the control wheel/cogged pulley interface so that when the knob is loosened, the control wheel and the pulley are slightly pushed apart. This minimizes any coupling friction/drag and results in a smoother manual operation.

The arrangement works fine and I can switch from manual to automatic operation in a half-turn of the knob.

I am a electrical engineer by training (I hack mech E stuff for fun). So it was no big deal to design a simple stepper motor driver circuit based on N-channel power FETs and some NPN transistor pre-drivers. The driver provides the power control so that logic level signals (0 to 5 volts) can drive the stepper windings. Zener diode flyback networks on the output stages are incorporated to maximize stepper torque.

I laid out a one-sided PC board using some CAD software. I had fun fabricating the board using a LaserJet printout on glossy paper for the resist (iron-on to the copper clad board…the toner melts and transfers to the copper and makes an excellent resist). Do a web search on home fabrication of PC boards to find info on this resist technique.

I mounted the driver board and keys/switch/pot controls to the back of the front control panel and cabled the controller module to the driver board with ribbon cable. All of the electronics save for 12 volt DC supply are housed in a plastic outdoor receptacle box mounted on a piece of iron pipe/flange base. This control head is mounted on the lathe base at the far right end of the lathe.

For the "brains" or controller, I decided to go for a small self-contained controller rather than a software implementation on a PC. I decided to use a great little module that I purchased from Technological Arts. It is at its heart, a Motorola 68HC11 microcomputer with flash RAM, boot ROM, an 8 channel A/D converter, I/O registers, power regulator, reset circuit, crystal oscillator, and RS-232 interface. Basically, it is a full microcomputer on a board about two inches on a side. A pin header on one edge of the board brings out the various I/O pins and control lines.

The 68HC11 is particularly well suited to the implementation I had in mind.

One nice feature is an on-board set of five programmable timers that can be easily programmed to generate stepper motor pulse sequences. Four of these timers are used to drive the four unipolar phases of the stepper. These timers have hardware support that virtually eliminates software latency during interrupt servicing.

While I am partial to the 68HC11, other microcontrollers could probably be used such as the PIC/Stamp series.

The idea behind the control strategy was to implement simple controls that would automate the basic traverse activities of a lathe i.e. go left to a stop position, right to a stop position and control of traverse speed independent of travel direction.

The control structure is primarily based on three momentary contact, single-pole, double throw switches. These spring loaded switches have a center off position and can be asserted either to the left or to the right.

The basic concept is to set a left-most stop position and a right-most stop position that are remembered by the controller. A switch is provided (GOTO left, GOTO right) that commands the stepper to traverse to the left stop or to traverse to the right stop).

Another switch (JOG/RUN left, JOG/RUN right) is provided to position the carriage anywhere along the traverse axis. The software is designed such that if the JOG/RUN switch is asserted then released in less than about one second, the stepper steps one-half step then stops (jogs). If the switch is held on for more than one second, the stepper starts to run continuously in the prescribed direction.

Once the desired stop position is reached using the JOG/RUN switch, a tap on the third switch (SET STOP left/right) causes the controller to remember that position as the given stop position.

An additional switch is provided that has two static positions, ONE WAY TRAVERSE or TWO WAY TRAVERSE. If the switch is in the one way position and the GOTO switch is tapped, the stepper traverses to the given stop position and then stops and remains there. If the switch is in the two way position and the GOTO switch is tapped, the stepper traverses to the given stop position and then immediately reverses direction and traverses back to the opposite stop position. This is much like a typical traverse motion manually done when cutting a shoulder. Or run one way, stop, back out the tool, traverse back, set depth and repeat.

By use of the above described switches, the left-most and right-most stop positions can be set and then automated one or two way traverses can be initiated.

Four pots (potentiometers) are provided as speed controls (values read by four A/D channels in the 68HC11). The pots to the left and right of the GOTO switches provide independent control of the left and right GOTO traverse speeds. For example, one might want to slowly cut into a shoulder going to left and then fairly quickly return to the right away from the end of the piece to set the next cross-feed depth.

The two pots to the left and right of the JOG/RUN switch provide independent control of the left and right JOG/RUN speeds. Two pots are probably overkill (who really needs independent control of left/right run adjust speed?) but I had a free A/D input and the symmetry had a certain appeal.

I wrote the software in assembly language on a laptop, assembled it and downloaded it into the 68HC11 module for execution. There is lots of development software on the web for the 68HC11. One particularly nice piece of freeware is called the Wookie simulator. It simulates the execution of 68HC11 code on a PC and provides breakpoint and register/memory observation features.

The basic software architecture is interrupt driven with a background idle loop. There are two primary sources of interrupts: a so-called real-time interrupt programmed to interrupt every 4.1 milliseconds and a timer interrupt used to generate the stepper pulse sequence (one timer interval for each half step).

The keys, control switch and the speed potentiometers are read every 4.1 milliseconds as part of the real-time interrupt servicing. There are debounce registers implemented in case the switch closures bounce…eight consecutive switch assertions (32.8 milliseconds) of the same sense are considered a legit assertion. Fortunately there are enough input lines available that I did not have to implement a keyboard scanner; I simply assigned one input line per switch.

The timer interrupt interval is used to set the pulse rate and therefore the stepper speed. The timer interrupt interval is set as a function of the A/Ded values of the speed potentiometers.

Currently I have not implemented serial communications to an external device such as a laptop as part of the operating code but should be relatively straightforward. This might be used, for example, to display relative position or some form of more sophisticated control.

Interestingly, the total size of the software is tiny…a little over 3KBytes.

It works surprisingly well! Based on my lathe mounted Starrett indicator, the carriage returns to well within 0.001" for a two way traverse. Single JOGs measure pretty consistently to the predicted 0.0003"/JOG on average. While there is a slight decrease in torque at the highest traverse pulse rates, it is more than adequate for fast traverse positioning.

The resulting part finish is certainly better than I was able to achieve by hand! (not surprising given my limited experience).

The only apparent downside is that at slow speeds, the drive is somewhat noisy (growls) but the vibrations are not noticeable on the carriage itself. I suspect the belt "twangs" on each pulse.

If I had to do it again, probably one of the things I would change is going with a shorter drive belt, more from aesthetics than for function. The current belt is 12" long. Also, the current bracket design is pretty close to the locking bolt on the tailstock and required building a custom off-set Allen wrench to tighten the tailstock position (oops!).

In all, it was a great project and certainly fun.

Comments and questions…mbonfire@hotmail.com

Let the chips fly!

Make your own PC boards using LaserJet resist:

Source for 68HC11 modules:

Source for Wookie simulator and other 68HC11 development software:

C&H Sales – Surplus electromechanical components (steppers, etc.):

Stock Drive Products – Source for mechanical components (cogged pulleys, belts, etc..):

Tutorials and info on stepper motor drives and circuitry (used for telescope drives but generally applicable):

Another little thing to share. I spent a bunch of time trying to achieve chatter-free parting. I tweaked the tool height, used high feed/slow speed, lots of lubricant (helped a lot). I also designed a pretty effective and inexpensive parting tool (see photo).

Basically, I cut a section out of a carbide woodworking blade with one tooth and mounted in onto a footed block with two screws. I used a emory wheel on a rotary hand tool to do the cutting. The parting tool has a limit of about 1.5" diameter work but is virtually chatter free due to its rigidity and practically, it is apparently hard to part stuff much bigger than that on a small lathe.

I tuned the one carbide tooth with a diamond hone to remove the hook and it works great. The blade was on sale for $3 new and I probably can cut three or four parting tools out of one blade. I realize it is probably a sub-optimal carbide formulation for metal but it seems fine for aluminum and brass.