The mill was acquired from Arc Euro Trade in an imperial configuration, meaning it had leadscrews of 20tpi, as I intended that most of my work would be in Imperial measurements.

The mill but with the short X table

Ball screws

There are many decisions to be taken about the mechanical changes to make for the mill to be converted to CNC and one of the most significant is whether or not to use ball screws for the lead screws.

In essence the salient arguments for using them are zero or near zero backlash and a very much higher efficiency in load transfer. The compelling argument against them is their high cost. The cost can be reduced significantly if rolled ball screws as opposed to ground ball screws, are used. However even these if bought new are expensive. A rolled ball screw without a preloaded nut will have some backlash (axial clearance) but it will be very small, but importantly it will be consistent (as far as the accuracy's modellers will generally work to) over the length of the screw. So what is "very small"? Well published figures are given for various grade of manufacture of both rolled and ground ball screws, they are international standards, so knowing the grade the accuracy's can be determined. Typically a rolled ball screw nut may have up to 0.1 mm (0.004") axial clearance for a non preloaded nut and for screws in the 14 - 28 mm diameter range. Better figures down to zero can be obtained as the quality standard increases and preloaded nuts are used. But all these improvements come at increased cost. Also the diameter of the screw is significant as may be judged by the published figure for 6-12 mm rolled screws where the axial clearance is halved, i.e. 0.05 mm (0.002").

The question of linear travel error ("lead accuracy") is not as important as the "lead accuracy" for a rolled ball screw will lie in the range of + or -50 microns/300 mm (0.00004") to = or - 210 microns/300 mm (0.0008"), again depending upon quality of manufacture. I think sufficiently small to be ignored for most purposes.

So why move away from the 20 tpi "Whitworth" form thread? The thread is seemingly well made on inspection, runs in a rudimentary preloaded (malleable cast iron?, but could be a steel) nut created by the simplicity of a slot in the nut that is adjusted in width by grub screws to force the nut to rub against both flanks of the thread. The wear rate was a ponderable and it would be a certainty with time to cause differences in both axial clearance and lead error over the length of the screw especially on the Z axis where the loading would be higher even with a counterbalancing arrangement. The loading efficiency would be about 70-80% less than a ball screw with the corresponding greater torque required from the drive motors (or less torque being available for use against the cutting tool). Had it not been for a fortuitous find of ball screws I would have been faced with the nasty decision. As it was fate came to the rescue by way of a stand at the Model Engineering Show in London (2006) that was selling some second hand rolled ball screws of 4 mm pitch and nominal 12 mm diameter (at a model engineers price!) that just happened to be the right length for adaptation to the mill. Now the grade and quality of the screws is totally unknown so only once they are assembled can the axial clearance and lead accuracy be determined.

So the decision, for me, was made to go to ball screws ...... but they are metric I hear you cry and you wanted an imperial machine. So I did, but apart from beggars not being choosers, it is possible to work to quite acceptable imperial accuracy's with metric leadscrews by using the micro stepping facilities of the stepper motor drivers and the CNC software. By way of explanation ......the typical accuracy of drawings for model loco parts will be drawn to 1/64" ths or multiples thereof. With a standard 200 steps/rev motor in micro step mode of 32 making 6400 steps/rev 1/64" accuracy can be achieved. The math's is simple:

Basic motor 200 steps/rev and 1 rev = 4 mm so 1/4 rev = 1 mm = 50 steps.

Now 1"=25.4 mm so 25.4 x 50 = 1270 steps which at the multiplier of 32 gives 40640 steps/inch.

This whole number of steps is divisible by 64 thus giving a whole number of steps for all multiples of 64.

So what is the error for a movement that is not a multiple of 64? Well taking 0.001" as the probable smallest accuracy then 1/1000 = 40640/1000 steps = 40.640 steps which of course cannot be achieved so there will be either 40 or 41 steps done.

At 40 steps this will give 40/6400 =1/160 mm which equals 0.000255" and:-

At 41 steps this will give 41/6400 = 0.00640625 mm which equals 0.0002522",

so a 1/4 thou error can be expected. Am I bothered? (to use current jargon!)

The use of ball screws obviously requires greater modification to the mill than if its original leadscrews were retained. The most significant of these changes is the accommodation of the ball nut. The ball nuts on my second hand screws were for a particular purpose having had special fittings screwed to them. These had to be removed leaving me with a plain round nut having a sliding fit dust cap on one end and an unknown screw thread at the other, the overall length being 1 3/4" long and a diameter of 1.012". I chose to fit a dust cap to the threaded end rather than thread the nut into a fixed block and to mount the nut into a block having a counter bore of the same diameter and with the nut held in place by grub screws bearing on it circumference through the block to ensure no twisting and a cap on the end to stop the nut moving laterally in the bore. This arrangement gave the smallest possible block length so as not to interfere with table maximum movement. I had originally intended only to use the grub screws to hold the nut in place but found the nut material exceptionally hard and feared the grub screws would not be effective so the nut block bore length was a couple of thou less than the nut itself such that the end cap clamped down onto the nut. The blocks and their nuts are mounted to the Y table in the case of the X axis, and the machine base in the case of the Y axis. The blocks needed to be of different detail design due to the differing clearances involved on each table arrangement.

The ball nut housing mounting for the X table

The ball screw requires to be mounted in bearings at both ends and at one end (at least) the bearing needs to be a thrust bearing to take the axial load in both directions. This thrust bearing arrangement can be achieved with two angular contact bearings mounted back to back (the right way round of course). The other end can be a simple roller bearing. The bearing mounting arrangements I thought should be kept similar for all three axis if possible to simplify manufacture. As it turned out similarity was there but the dimensioning was different. I chose to mount the thrust bearings at the motor drive end of the shaft as my screws would also be manually operated by the supplied handles. As I chose to drive the ball screws indirectly via HDT belts the wider bearing surface provided by the thrust bearings would give some advantage to rigidity and the acceptance of the forces applied from using the manual winding of the handle which was a direct attachment to the ball screw.

Which brings up the subject of the drive arrangements to the ball screws. Direct drive or indirect via pulleys and belt?

Drive arrangements for X and Y tables

A direct drive is simpler in the main and cheaper than indirect as there are no extra pulleys and belts to accommodate and thus the whole drive arrangement is more compact. The alignment of motor and ballscrew becomes very important and a flexible coupling of some sort that allows a small amount of movement is used by some who carry out the CNC modification. I chose a belt drive as it offers the possibility, if it becomes necessary, to put a ratio pulley drive into place thus increasing the available torque at the ball screw. Here the thinking was that the Z drive may well need a 2:1 drive ratio. Of course I could simply make a different mounting arrangement for the Z axis but I felt having three similar was to be preferred. The accuracy of mounting the motor becomes a slightly less important factor too. Finally by indirect drive the stepper motor shaft does not take the loading of turning the manual handwheel.

So to the fitting of the ball screws.

Initially I drew out the arrangement on CAD to ensure all the various parts could be assembled together and there were no fouls from table movements. The net result of this was the determination of the modifications needed to the mill to accommodate the ball screw nuts. The nut blocks are illustrated in the drawings shown and the Y nut block needed 1/8" milling off the the existing face on the mill base that took the original nut. This was done first making sure all the paint was removed from the underside of the mill base feet such that it would sit flat onto the workshop mill table. The fixing screws for the new Y nut are the original 6 mm cap head screws.

The Y nut block was made and fitted the important part being to ensure the bore of the block was parallel with the base of the block.

The Y table plate that holds the stepper motor was made 3/8" thick (except the area for the motor mount which was thinned down to 1/4") and provision was made for two extra fixing holes in the table so there were four in total to ensure a solid and wobble free fitting. To provide for belt adjustment the ball screw bearing mounting plate was mounted in an eccentric having a 1/16" movement (1/8" from one side to the other). This would provide sufficient slack to fit the belt and pulleys and then subsequently tension the belt by turning the eccentric housing. However this design means that the centre line of the ball screw changes so once the belt has been adjusted the ball screw bearing plate has to be aligned with the centre line of the ball screw nut and the fixing holes for the plate are sufficiently oversize to allow the plate to be adjusted to achieve this.

The plate is fixed to the table using 3/8" diameter standoff pillars 2" long having 2 BA stud ends for the plate and one pair have M6 threads at the other end to use the existing holes in the table and the other pair use 1/4" BSF ..... they could have been 6M too but the brain cell was inefficient the day I drilled the holes.

For the Y table the ball screw had to be shortened and this was tough .... literally as the ball screw was hardened and that made turning to a diameter very hard, even with tipped tools and trying to cut a few thou just caused the tool to burnish the top rather than cut. So to cut to finish size meant taking quite a large cut to get the tool to work. Fortunately I managed to get reasonable bearing fits. To shorten the screw I used diamond saws of the sort readily available for mini drills. It took two to get right through though.

When turning the ends for the bearings I used the four-jaw chuck to ensure the ball screw was clocked to run true. I also left the nut on the ballscrew which meant there was a bit of a chuck overhang and that does not bode well when turning small diameters, as the tool pushes the work away and with hardened material this effect is worse, so the tailstock centre and fixed steady were utilised to give rigidity to the work. The alternative was to make a ball keep for the nut and take it off the screw. To prevent swarf getting into the nut the adjacent threads were covered with masking tape.

The pulley for the belt was held in position by two 4 BA grub screws bearing down onto a ground flat on the shaft. The grub screws when finally fitted were fitted with loctite to prevent the stepper motor vibrations causing them to come loose.

To mill the new keyway for the handle a solid carbide cutter was used to get into the hard surface.

The X nut block required a slot to be milled in the Y table as there was insufficient clearance between the underside of the X table and the top of Y table to accommodate the X nut block. This was done next. The nut block had to be sufficiently shallow to clear both the underside of the X table and the top of the Y nut block and at the same time provide sufficient material to hold the ball nut and take four 2 BA cap head holding down bolts counter bored into the block so the cap heads finished flush.

The slot required removes all of the material in the middle of the Y table so a hole is left. The nut block was made a very good fit into the hole on the premise that when it was bolted in it would put back some of the strength/rigidity in the table lost by making the hole. A shallow slot was also required in the table to provide clearance for the ball screw opposite to the slot already in Y table that gave clearance to the old X leadscrew bracket.

The rear bearing plate was made in a similar fashion to the Y table only the dimensioning is different as the relationship of the centre line of the ball screw to the table is different. The plate covers the T slots ( see later for their access) thus providing a good flat area over which to bolt the plate down and prevent possible misalignment.

The front plate for the X table differs from the Y plate as the motor has to point in the opposite direction so as not to foul the vertical column if mounted on the right side. Or, if mounted on the left side the Y table. I chose to mount the motor on the right side of the X table as this kept the front of the mill neat and tidy and it did not interfere with my right handed operation of the X table handle. So the motor is mounted on pillars to stand off from the plate with sufficient room for the HTD pulley.

The front plate is also stood off from the table on pillars to give a 3/4" gap. This is there for two reasons: first it gives a gap so that the T nuts can be out into their slots and secondly it gives a further 3/4" of table movement as otherwise the plate would foul the base of the machine to early in the tables last inch or so of travel. The pillars are made the same as on the Y table apart from their length.

The ball screw needed a 2"extension to the shaft to provide sufficient length for the pulley and handle. The extension was fixed by screwing it onto the ball screw shaft that was turned down and threaded 6M. The shaft was hard and the HSS die was left in a poor state once the thread had been cut. The fit of the thread into the extension was such that the male was oversize and thus the extension very tight to screw on, deliberately as no play was the objective in the extended piece. The extension was made oversize in diameter and once screwed onto the ball screw turned down to be concentric with it and to the same size.

The Z ball screw arrangement is discussed after the decision on what to do with the Z axis.

Limit switches

Looking at commercially available cnc modified Super X1 mills I could see no evidence of limit switches having been fitted to the X and Y tables. They may be there but I could not see them. I wondered about the benefit of fitting them and it did not take me long to realise that one mistake on a drawing or positioning of the work on the table could end up with an expensive repair so I decided to fit them to the X and Y tables to prevent the tables running beyond their mechanical limits.

Now limit switches are not quite the same thing as microswitches A limit switch has to operate with repeatability and be reliable as it is a safety device. It cannot operate at one point on one occasion and another on the next occasion due to having some random hysterisis effect. The result is that limit switches tend to be larger and more expensive and by their very nature of being a machine accessory are more robust in their build. After a lot of searching I eventually found a switch that was of a suitable size to fit to the mill and not interfere with the table movements from being too large. The switch is from Farnell and of Bernstein manufacture with a manufacturers reference number 600-8354-026 and description :- "LIMIT SWITCH, 1NC 1NO SNAP ACTION; Configuration, contact:1N/C, 1 N/O; Actuator type:Adjustable plunger; Current rating, thermal:10A; Temperature, operating max:80°C; Temperature, operating min:-30°C".

Finding a place to fit the switches is not straight forward the most awkward being the Y table but eventually I settles on one switch being mounted inside the slideway and under the table so that the front plate actuated the switch and the other at the side of the bed just clearing the stepper motor and operated by a lever attached to the rear of the table. The photos should show the arrangement sufficiently clearly. The switches require front screws to mount them so suitable brackets needed to be made from 18g sheet steel. The switch under the table only just has enough clearance when sitting on the base so its bracket was of a U configuration with the bracket fixing screw to the sides. To fit this switch in the position shown required a bit of fiddling the table has to be partly assembled loosely and then the switch manipulated into position and onto its bracket. It has to be wired first as the lid cannot be opened when the switch is in position.

Limit switch positions on the X and Y tables

The switches for the X table are mounted on the rear of the Y table. These are relatively straight forward to fit on brackets from 18g sheet steel and they are operated from levers bolted to the X table ends. They too need wiring before the table is reassembled as the lids will not open once the table is in position.

These limit switches are used by the Mach3 software as the "home" switches as well as being the safety for extreme travel, and they can be configured for the 0,0 position at either corner of the extreme table travel movement. Perhaps a natural 0,0 is the same as one would choose for drawing on paper that being the left hand front corner (or left bottom if viewed in plan view). That's the position I chose.

As limit switches are a safety device their wiring configuration is important and they should be wired as normally closed contacts opening when the switch is operated. This arrangement is safe if an open circuit should occur for any reason, this being considered a more likely event than a short circuit occurring. In Mach3 this will be an "active high" arrangement, i.e. the detection pin voltage will go high (5V) when the switch is activated (open circuit). However a breakout board might be supplied that has a 5V output for the input circuits so a normally closed contact between this and the input will put 5V on the input pin and unless the breakout board inverts the signal this will become an "active low" in Mach3 as when the switch goes open circuit the 5V is removed from the input pin and the signal will (should?) be grounded by the breakout board. This arrangement is the one used by the CNC4PC bi-directional board and I found it unreliable as the open circuit condition leaves the input "floating" as the input pin is not grounded when the 5V is removed.

Another important factor with the limit switches is their cabling. There are a lot of digital pulses around the control box and to the stepper motors and the limit switch cables will be like aerials and get all sorts of noise induced onto them from these digital pulses (even though the cables to the motors are screened) and this will upset the input circuit to the breakout board. The limit switch cables should therefore be shielded and run separately and as far from the cables to the stepper motors as possible.

Initially I did experience noise problems on the limit switch inputs as I had run unscreened cables to the switches in the same flexible trunking as the motor cables (assuming the screened motor cables was adequate) and that resulted in the switches not being reliable or just not working. My solution was to employ a good shielded cable ..... aluminium foil with a drain wire ...... and each switch pair in the cable was a twisted pair. The cable was taken separate from the motors from the control box to a junction box adjacent to the mill and in that was mounted a 470 ohm pull up resistor connected between the 5V supply and the input signal core and the limit switch connected between the input wire and earth (the drain wire) in a normally closed configuration. This arrangement worked OK.

Z Axis

The Z axis control has two options. The quill can be driven or the main head on its column can be driven. The quill only gives 30 mm of travel so it is a limitation. To drive the head on its column has difficulty due to the lead screw being behind the column and thus there is a large turning moment onto the head slides due to the weight of the head and this weight means that the torque required to drive the head upwards would be quite large. The head and drive weighs about 20 lb.

Another consideration is the rigidity of the head. Normally a mill head (or knee) would be locked when milling and the quill used to change the tool position and then locked for a cut thus giving maximum rigidity. With a CNC machine the quill or head must be free to move so a decision is needed on what to drive as there are pros and cons for each option. Maximum rigidity would be obtained by locking the head and driving the quill but the limitation of this arrangement is the 30 mm movement.The requirement to clear fixing studs or other clamps on the table caused me to opt for driving the head as the quill movement would almost certainly be insufficient to give clearance and furthermore if a tool change were needed there would be insufficient space beneath the collet chuck to make the tool change. It would be worse if the collet chuck had to be changed for a drill chuck in a machining operation. Of course by splitting operations into separate programs and resetting the machine to datum could be employed in the case of tool changes.

With the decision to drive the head came the need (so I thought) to provide some form of counterbalance to the head such that the up and down driving forces were made more equal and acceptable. The solution intended to be adopted was to use a gas strut fitted to the base of the column and to the head at a slight angle. The strut does not really interfere with the working space with this arrangement. The strut chosen was of 90 N force (which equates to just over 20 lb) with a stroke of 200 mm and a maximum length of 440 mm. However having fitted the ball screw and tried the head movement (minus the motor and drive assembly) I was pleasantly surprised how easy it moved without any counter balance arrangement. So I parked the counter balance idea until I had further experience with the completed head drive as the gas strut solution could easily be added to the completed modified mill.

Z axis ball screw arrangement

The ball screw arrangement retained the existing bracket that connects the head to the old lead screw and a new block was mounted on this bracket to accommodate the ball screw nut. At the base of the column an angled bracket was fitted to accommodate the bottom ball race for the ball screw. At the top of the column the plastic cover was removed (it's only stuck on) and an angle bracket fitted on the inside of the column pointing to the right (facing the machine) and a second angle bracket fitted to the outside on the left of the column. These two brackets were made to be substantial as they had to support rigidly the plate carrying the thrust bearings, handwheel and stepper motor. The bracket fitted to the inside of the column beds onto the rough casting and thus will not bolt down square. To overcome this the top of the bracket is machined square after fitting. Both brackets need to be machined in any case so they finished at the correct height so the main plate and the thrust bearings finish tight up against the ball screw shoulder. Any slight deviation is able to be taken up by adjusting the dimension of the thrust bearing carrier ring depth.

The brackets were made to fit so their height left a gap of approximately 1/2" between the top of the column and the underside of the plate in order to maximise the length of travel of the head on its slide and of course the ball screw length was machined to suit.

Two views of the Z drive arrangement

The motor was fitted in the same way as the Y table, being bolted directly to the plate that had been "thinned" locally to 1/4" from the overall 3/8" thickness.

The motor drive spindle needed extending by an 1" so the pulley aligned with the one on the ball screw, and the ball screw shaft needed a 1" extension to accept the handle.With the drive and pulleys fitted and the head gib adjusted a trial run of the Z axis showed the motor was able to cope extremely well in both up and down directions with no apparent strain in the upward direction. I concluded that a counter balance strut was unnecessary.

The completed conversion minus drive covers

Note the convenient spot for the control box underneath and just visible

The A drive is temporarily laying at the back just visible and not in conduit

it awaits fitting to a rotary table

The covers for the drives were made from 22g brass sheet and were three sided the "bottom" being open. The front face was soft soldered into place once the cut out for the handle had been done. The X drive cover is slightly different as there has to be a cut out for the motor. This was edged with brass angle made from the sheet to give a generous overlap (3/4") to the motor sides to close the gap.

The covers have a central 1/4" BSF threaded stud that holds them in place.

To give protection to the Y slides and Z slides as well as the two X table limit switches a flexible rubber sheet shaped in a trapezoidal form is bolted to the rear edge of the X table with a clamp bar and also to the underside of the quill head. It is sufficiently long not to impede the X table travel.

The finished modifications look like this: ..................