Overview For creating prototype parts, I have found the CNC milling machine to be an absolutely indispensable tool. This tool, along with the proper CAD/CAM software, goes a long way towards freeing one's mechanical creativity. I started out with a MAXNC 15 CL machine, which uses stepper motors equipped with encoders for closed loop operation. This machine worked pretty well for very light jobs with easy materials, but its aluminum ways and low-force anti-backlash nuts proved to be too delicate for many jobs. After investigating the other desktop-sized mills on the market, I decided that a conversion would be a good way to go. So I purchased a Harbor Freight 47158 Micro-Mill. This mill comes ready for hand use, with hand-wheels on each axis. Even though it is not CNC, it comes with some really nice features such as all iron/steel construction and an industry-standard spindle taper, and it is relatively easy to convert. It's pretty heavy-duty for a machine its size.
Mounting the Axis Motors I used the control system from the MAXNC 15 CL for the conversion to CNC. Most of the custom parts for the conversion were made on the MAXNC before its control system was stripped. These parts could have also been made on the HF machine itself before performing the conversion. After cleaning the machine, I lapped the gibs to provide nice, smooth operation. Motor supports were machined out of 6061-T6 aluminum to provide solid support (see photos).	Custom X and Y Axis mounting hardware. Custom Z Axis mounting hardware.
Motor to Leadscrew Coupling If you are new to CNC, the connection between each motor and lead screw is more involved then you may think. First, on a milling machine, there are high axial loads on the lead screws which must be accounted for. I used thrust bearings from McMaster-Carr (part #7806K63, ~$9 each). Two sets of these bearings were used on each axis (one bearing on each side of the bearing block). I enlarged the bore on the inside of the X and Y bearing blocks to accommodate the inner thrust bearings. Second, the coupling between the motor and the lead screw must be somewhat flexible to accommodate slight mis-alignments between the two, as well as shaft wobble caused by a slightly bent leadscrew or mis-aligned nut. Yet the coupling should have no backlash and be able to withstand high load forces. To achieve this, I used a Helical Beam Coupling (Also from McMaster, ~$30 each) on each axis.	Enlarged bore on bearing block. Thrust Bearing and Helical Beam Coupler.
Y-Axis Assembly.	X-axis Assembly.
X and Y Axis Assemblies.	Z-axis Assembly.
Z-axis Supplemental Force The weight of the Z-axis spindle and motor assembly means a lot more force is needed to move it up than down. To counteract this as well as help provide zero backlash, I added a 20 lb. counter weight, braided cable, and pulleys. Some people have used gas struts instead.	Z-axis counterweight attatchment.
Home/Limit "Switches" Since the optical encoders on the stepper motors only sense incremental (not absolute) motion, a means is needed to give each axis an absolute "home" position. Then all other positions can be determined relative to that. This can be done with simple mechanical switches that are activated when an axis reaches its end. However, mechanical switches are prone to inaccuracies and breakage. So I decided to try a hall-effect system. I used small, low-cost Hall-Effect sensors (Digikey part # DN6848-ND). Each sensor has an open collector output which will sink current when a magnetic field is applied to the sensor. Each sensor was mounted in a short brass tube by potting it firmly in place with hot glue. Prior to potting the sensor, each brass tube was shaped to fit nicely on the axis it was meant for. A cable runs from each sensor to a small box that houses the simple combinatorial logic and pull-up resistors necessary to convert the signals for use by the MAXNC controller. Each sensor, with its brass tubing mounting bracket, was mounted to the machine near an axis way. Than a small 1/8" diameter rare-earth magnet was placed on the moving axis near its end. My moving the magnet, the exact end-point can be adjusted. It is also possible to place another magnet on the other end of the axis for sensing both ends. In my case this was unnecessary since my software determines the other end. I have been extremely pleased with the results of this Hall-effect system. Every time I have told the machine to "go home", it has calibrated to exactly the same place as the previous time, to within the accuracy of my software (about 0.0003"). One caution: I do not cut ferrous materials very often. It is possible that ferrous metal chips could get stuck on the magnets and affect their performance. I mounted the magnets out of the flight path of any chips, so I doubt this will be a problem.	Y-Axis Hall-Effect sensor. The sensor is mounted inside the Brass tube, at the right end. You can see the magnet just above the sensor. Z-Axis Hall-Effect sensor. The sensor is mounted inside the brass tube, at the left end. You can see the "potting compound" (hot glue) at the end. The magnet is mounted on the moving portion of the Z-axis (not shown).
Coolant System and Cage When cutting certain materials (especially metal), the cutting bit can get hot enough to self destruct rather quickly. Coolant can be manually sprayed or brushed onto the tool to alleviate this for short jobs. However, part of the beauty of CNC is that the machine can run for hours straight to produce complex pieces. So it is desirable to have a coolant system that can keep the tool and workpiece cool the entire time, thus allowing continuous operation. Essentially, the cooling system consists of a reservoir full of coolant, and a small pond pump that pumps the coolant through vinyl tubing to a Loc-line (a plastic hose that will stay in any position you place it in) and nozzle. The nozzle can be positioned just right to blast the coolant onto the end mill. The Loc-Line is mounted on the moving Z-axis, so it moves with the axis and always stays aimed at the bit. The coolant then dribbles down and all over the place. The cage the machine is in keeps the coolant contained so that it all drains to the bottom and though a 1" drainage hole and tube, which dumps straight into a paint strainer (which is located inside the coolant reservoir a.k.a bucket). Here are the key materials used for the coolant system: -Harbor Freight mini pump Item # 41287 (120V pond pump) -Relay from Radio Shack to operate the pump from CNC control -5 Gallon Bucket (Home Depot) -Two Paint Strainers (Home Depot) plus spares. One wrapped around the pump and the other around the drain that dumps into the bucket. This solution works really well- I tried other types of filters and they always clogged up or let through too much debris. -13" Loc-Line Adjustable coolant hose (part #40413). This is the blue plastic that you kink into position and then it stays there, and has an orange tip nozzle. I cut mostly plastic and 6061 aluminum. Some of my aluminum jobs go 6+ hours straight, and I have never had this thing clog or stop working! The 4 walls of the cage are 1/8" plexiglass, and the ceiling is pegboard. This by itself would be way too flimsy, so I have a wood frame around all the outside edges to hold everything together. It is all permanently mounted to a plywood base. For the floor, I have a layer of 1/8" thick blue flexible plastic material (plexiglass would crack under the weight of the machine). Then the machine is bolted to a thick wooden base of its own and just sits on top of the plastic flooring. The machine stays put, and having the wood base movable makes it relatively easy for servicing.	Loc-Line mounting. You can also see the small magnet used for the Z-axis home sensing. Another view of the Loc-Line. Complete setup. You can see the top of the 5-Gallon bucket, and the cage. The doors to the cage are open in this photo.
Results The conversion process went relatively smoothly, and I am pleased with the final results. The max feed speed is about 40 IPM. The only backlash in the system is at the nuts. But the X and Y axis nuts are adjustable to remove backlash. Typically I do not adjust them tight enough to remove all backlash because this causes too much strain on the stepper motors. I have found that the lead screws are not perfectly straight, which causes annoying binding and typically requires some play in the nut mounting and/or the bearing blocks. The spindle has worked pretty good for a machine this size. The primary disadvantage has been it's low maximum speed. This is a hindrance when using very small diameter cutting tools. For now I compensate by running low feed speeds, but eventually it would be nice to upgrade to a faster spindle.
Future Enhancements The next major upgrade I have planned is to remove the entire control system (including the stepper motors, control box, and PC) and build my own servo system. This will yield higher performance, and more importantly, get rid of annoying "Servo errors" that I get with the MAXNC system. Even though this current system has encoders on the motors, it is still not able to deal with many circumstances and so it halts operation. This often requires the part to be started over. Obviously, this can be very frustrating when you get an error like this 3 hours into a 6 hour job. Another possible future upgrade is to add an extended Y-axis kit for greater travel.
Links: CNC conversion Part II Covers upgrade to DC servo system and custom modifications to run spindle at higher speed. Yahoo newsgroup specifically focused on the conversion of this machine. Littlemachineshop.com carries a nice selection of parts and accessories for the Harbor Freight micro mill. -JS-