fbpx

Design Theory

 

There are many factors that are involved in engineering decisions such as pricing, target markets, technology that was readily available at the time, overall investment in development, internal politics etc.

What we would like to say about ours is that we really used this opportunity to start something from the ground up to be as optimal as we could. We weren’t trying to build a system that already adapted to an older one and we weren’t locked into any design choices that weren’t our own decisions that forced development in any one direction. We didn’t have a “that’s the way it’s always been done” mentality and really asked ourselves if there might be a better or more practical solution. 

 

Frame

 

To start with, we weren’t personally all that impressed with the classical column design found in most mills. It has its pros, but also its cons.  The big points we considered is that it’s cheap to make, easy to design/develop, and fairly lightweight. It’s cons we determined are worse. Due to the design, the X axis typically causes a large X axis footprint of the machine. 

 

Very roughly, machine width minimum becomes a little more than [x table width + 2 times X travel] and that typically results in the footprint of these machines being very large or rather very wide. Since that means the largest dimension of the machine is the width when you’re standing in front of it, it results in inefficient shop density optimization (the setting up of mills/operators/and ancillary equipment per usable unit space). Small businesses that might operate in much smaller workspaces will definitely feel the pain of trying to optimize their spaces to fit such large machines, especially if they own multiple mills. By using a gantry-column hybrid style as we have, we basically are able to build a machine at the same travels but at roughly 1/3 the width of the classical column design.    

 

So along with column mills having a large width as a con, column mills will typically have much more torsional strain.  This is especially seen on micro/cheaper mills where bottom line costs and often poor or no engineering skill is taken into account. You will have a tall narrow column (meaning a large height to cross-sectional area ratio) and attached to that you have a large cantilevered head which acts as a lever.  When loading in the x axis, that acts like a wrench that twists the column. This effect is least pronounced at Z min, which is closer to where the majority of machining jobs typically occur. Larger manufacturers of industrial grade mills have done the engineering analysis to figure out how much material they need to add to the column to prevent that from being a concern, but that is also reflected in the price (being that there are no “cheap” industrial level cnc mills). You will typically see pretty narrow columns on lower priced machines and that makes them quite susceptible to those forces. 

 

In comparison, a gantry style mill typically is more rigid, having no issues with x axis loading torsional strain, but also a generally heavier design. However, it still has one element that is typically cantilevered way out as well, and that is the Z axis. Again anything that is cantilevered way out leverages forces acting on a structure and results in excessive strain (aka flex). The cantilever aka moment arm is greatest when at z min, which is a location where the majority of machining for various jobs occurs. At Z max, cantilever is minimized and gantry frames are the most rigid. Reinforcing those parts with more frame material is still the best solution in those cases but what if you had a frame that minimized the cantilever moment on all 3 axis?  That’s what we have developed with our mill, using a frame geometry we simply call a gantry-column hybrid. 

 

We call our design a hybrid between both as we feel we have taken the strengths of both frame geometries and combined them into one, without any of the serious cons they present. Our frame design results in the table itself only moving in the Y direction, the head moving in the X direction, and a carriage that connects the head to the column moving in the Z direction. Essentially we built a column that was extremely wide, as wide as the machine itself, and moved the x axis from the table to the head. The result is the best of both worlds. Cantilever forces are minimized on all axis and the resulting footprint of the machine is extremely compact. 

 

Since our target market is small businesses and micro fabricators we realized that having a large machine comes with large space requirements. We wanted to be able to create a compact mill that could fit through a doorway with the minimum sacrifice to capability. 

 

Linear Guides

 

Now we’ve covered frame geometry, lets cover the linear motion. There are multiple forms used to constrain motion to one axis, the most common are linear guides, dovetails and box ways and all three are still relevant in this day and age. Dovetails and box ways maintain a large surface area contact between moving parts meaning forces are well distributed between them and often exhibit superior dampening capabilities. Linear guides distribute their forces through multiple point loads (aka the ball bearings) and thus have extremely high point load forces, though the rails are hardened steel vs a softer cast iron of dovetail/box w so it balances out.

 

We chose to go with linear rails for our machine because of multiple reasons. The first is due to the longevity of the machine. Maintenance of linear rails is far less expensive when compared to box ways or dovetails. To replace the linear rails you simply bolt new ones on. To “replace” box ways or dovetails is impossible since they are part of the frame; you can rebuild them, but it’s expensive.

 

While the greater surface area provided by dovetails and box ways are great, there are downsides to it as well. Older mills operated manually by user controlled crank typically cut large cuts and long passes. The machines will typically last for years of daily use before needing a major overhaul. CNC mills do not operate the same, however. High Speed Milling which is used in modern CNC machines typically take very shallow cuts repeatedly at high speeds. This produces excessive wear on full face contact linear bearings resulting in “bowing.” This results in the center of the axis, where the vise or rather the most motion tracked occurs, becoming loose, while the extremities of the axis staying intolerance or becoming tight. This occurs because the center gets worn out and thus lose, while the ends still remain tight. 

 

Linear rails do suffer similar issues but nowhere near to a degree dovetails/box ways do. Linear rails are hardened and precision ground. Since the linear rails are not part of the machine, equipment to produce hardened and ground linear rails, as well as economy of scale due to standardization, has made linear rail technology in recent years extremely affordable. Precision linear motion is no longer out of the realm of possibility and is now economical enough to be a viable solution. 

 

Ballscrews

 

Transmitting linear forces is done via ball screws. Really there are only a handful of technologies that do this and are common on the market. Ballscrews, linear motors, ACME/trapezoidal, belts, as well as rack and pinion are the main choices we considered. Rack and pinion as well as belts we quickly discounted, the forces we’re dealing with and the precision we want excludes those two. Trapezoidal screws are what you will see on older manual mills. They work great but are typically best for low speed as they have a decent amount of friction. Friction, in this case, means it will wear fast and quickly develop backlash and/or require replacement quickly. Really the most obvious, the only choice, would be using ball screws as they are low friction, high force, and high precision. 

 

Ballscrew precision is a fantastic debate and again this is my own personal opinion. You basically have rolled and ground screws. Rolled screws are made from rod stock that is formed through rolling them through screw dies. Ground screws are formed by grinding rod stock into a screw shape. Rolled screws are generally considered to be less accurate than ground, but are harder (due to work hardening when rolling) and therefore stronger. Ground screws are definitely considered to be more accurate, but not as strong. It’s a common choice to use ground screws which cost more because they are more accurate, and CNC milling is all about precision. We decided to use rolled screws though. Why? Well, one reason is that it’s a lot cheaper allowing us to pass those cost savings on to our customer, but really we aren’t convinced they are more accurate. The screws themselves are definitely more accurate, but that is physical. There is a software component when dealing with CNC’s and there presents the opportunity to map the ball screws in the software and have it compensate for any inaccuracies. This results in us being able to use lower priced ballscrews that are stronger (so less flex under load) than ground ball screws while being able to offer the same accuracy as ground ballscrews. 

 

Spindle

 

The spindle is considered the heart of the machine. It is arguably the most critical component in the entire machine and thus something we took extra seriously when designing our mill. We wanted a spindle that would be able to survive crashes, offer options such as through spindle air blast for tool changes, and through spindle coolant for milling. We decided on a BT30 12k rpm spindle. We felt this spindle was perfect at it excelled in speed and robustness. The bearings in it are Japanese 7010, which is 2 sizes larger than the typical bearings used in BT30’s. The larger bearings offer far more loading ability making them more resistant to crashes and increase the overall lifespan of the unit. While larger bearings do not generally like higher RPMs, these bearings operating inside a 12k spindle are still well below their limits. We saw little need to increase the price to the consumer by offering advanced cooling solutions when the grease packed bearings themselves are well within their limits using air cooling.

 

The spindle also has a water jacket that allows for an optional spindle chiller unit which can increase part accuracy and lifespan. Spindle cooling exists because of the heat generated from friction due to the shear forces of the grease or oil lubricating the bearings. Hotter bearings reduce their lifespan, and a hotter spindle shaft causes thermal expansion. It can also pump heat into the frame itself causing more thermal expansion. That growth can be ignored in many cases, but if you want to maximize precision, it would be a recommended thing to have since it would add more repeatability. While we are not including it currently as it is not necessary and would add to the price, it is recommended for heavily used machines. 

 

Homing

 

Homing is used to zero all machine axis so the machine knows exactly where all its moving parts are at currently. It’s is important for soft limits too, which is a software imposed limit to prevent the machine from crashing an axis through overtravel. Typically homing is achieved through two stages for us. Typically for smaller mills you will have only single stage homing. Most typical homing switches used are inductive, mechanical, and torque/current sensing. That is done via “crashing” the travel into a hard physical limit and picking up where that limit is, from the axis motor. Homing precision isn’t all that important if you’re planning to set up a job and run it in a single sitting. However, what about when the machine crashes, estops, or you shut it down for the day? What if you are running batch work and want to be able to come in the next morning and not have to re-zero all your fixtures? Our homing is two stage because we use the encoder index pulse on our servos after the traditional limit switch trigger to increase the repeatability of homing. After a moving axis trips the first limit switch, the axis continues to travel until it hits the index pulse on the axis drive. 

 

Controller

 

The controller is the brains of the CNC. Its main job is to plan paths and use that data to control the drivers that handle sending power to the motors. To do a simple circle motion, it not only has to coordinate between two axis simultaneously, but also send the steps and direction signals at exactly the right times. For a computer this isn’t all that demanding until you ask it to plan 3 (or more) dimensional paths at sometimes over 1000 lines of g-code per second, to send out 200,000-400,000 pulse signals to multiple drivers per second, to look hundreds of lines ahead and figure out how to smooth the future motion, etc. While gcode can say the path takes a 90 degree turn while at 300ipm feed, a physical machine throwing with the inertia of hundreds of pounds of cast iron around cannot. It’s up to the controller to look ahead, see a turn coming up, and slow the machine down on time so that it doesn’t overshoot. Therefore it’s very important that the controller has a computer completely dedicated to that sole task. It shouldn’t be running other processes that might interrupt or supersede the path planning algorithms while it’s running, and this includes the user interface. 

 

We tested multiple controllers before we settled on Acorn from Centroid CNC.  There were many constraints that influenced the decision such as human factors (ergonomics), price, reputation, capabilities/features and performance. Acorn offers many of the features of higher end controllers such as feed rate override, smoothing, ball screw mapping, lube pump control, MDI, unlimited program storage, cutter comp, part counter, conversational programming, coordinate rotation and scaling, coordinate system rotation, rigid tapping and many more. All these features are included while many other controllers or companies might not offer it at all or charge additional money to unlock them for you. Centroid CNC has been around for decades and have been using the same software (improved over the years since conception) as what will be using. Learning centroid’s cnc software is fairly easy and intuitive. There are dozens of videos online explaining its operation and there is a complete manual that covers its operation in depth. 

 

Acorn while by itself is a computer, requires an additional computer to handle the graphical user interface.  This is handy as it leaves the Acorn to be a dedicated path planner without other processes interfering with its sole job. The main computer that handles the user interface runs in Windows 10 and can be supplied by the customer or by us. Centroid CNC also offers a wireless MPG which we highly recommend from our own personal testing and a physical (vs. touchscreen) control panel that we can install if our customers desire it.

 

Axis Motors

 

Linear forces are created by electric motors attached to the end of each ball screw. Typically there are two categories of motors, steppers, and servos. Without going into depth, stepper motors are typically cheaper, offer no feedback, have impressive torque but operate at low rpms. Servo motors have feedback via an encoder, more expensive, lower but broader torque and much higher rpms, a magnitude more in some cases.

 

The axis motors we chose are all servo motors. Each of the servos has an encoder attached to the tail end of it that registers position and sends that feedback to the drives. The drives use that positional data to determine how much power should be sent to the servos, thus completing the loop. In the event of a crash or missed step, the drivers will know immediately and throw a fault trigger. This fault is wired into the controller to essentially E-stop the machine stopping all motion including the spindle. This feature we hope will not only increase user safety but also reduce potential damage to the machine and a trashed part.