Is it possible to build a CNC whose Linear motion system does not contain any timing belt(pulley) or lead screw(threaded rod). I was wondering whether I could directly control the Linear motion by securing wheels of a slider onto aluminum rails & directly connecting the wheels to a stepper motor.
The main objective of this question is to find the cheapest method for controlled Linear motion.
I would say the answer to the first question is "Yes and No".
Yes, you could hook a stepper motor directly to drive wheels on some sort of shuttle. CNC controllers normally have a setting for inches/step or some equivalent. If you set that correctly, the machine should drive the motors "correctly".
But No, this not likely to give you the output you want. Stepper-based CNC machines work on the idea that the machine never "loses steps": that for every electrical pulse, the motor moves a fixed amount; and that N forward steps followed by N backwards steps brings you back to exactly where you started. If you are driving wheels which rely on friction for motion, those two assumptions are almost certain to fail, and your output path will be stretched or compressed in strange ways.
CNC controllers, in most cases, control rotary motion and the model of how this rotary motion is tranformed, by the mechanism attached to the motor, to a translational motion is implemented in the controller.
You can use any method of transforming rotary motion to a linear motion as long as the model for it is pre-implemented in the CNC controller or you can implement it to the CNC. Moreover, it does not even has to be a decoupled translational motion. E.g. LinuxCNC can controll multiple axes to achive 1 decoupled linear motion. On a parellel mechanism, moving all axes results in a Z translation. Video of this here.
The only limit to what mechanism can you put on the end of the motor of a CNC is what are you able to model and integrate to the CNC controller. If you do not have the possibility to implement a model, then the question becomes what pre-implemented models are there and how can you adjust their parameters to make them refelct your mechanism.
Needles to say the accuracy of the model will affect the accuracy of the CNC.
If you want to build something that more or less resembles CNC machine, and use a pen as a tool and draw some pictures, then you will probably be fine (as long as you can do programming and have a lot of patience to do calibration).
Why is this not a common solution?
Usually you want CNC to be precise and to have enough power to move a tool at reasonable speed. Cheap rubber wheels are... well cheap. Nobody cares if their diameter is 10mm or maybe 10.3. Or 9.7. What's more you can buy several wheels and each one of them will have different diameter. This will lead to problems when calibrating the machine.
Additionally, threaded rods and belts can transfer more torque without slipping, so higher dynamics will be achievable.
What you are looking for is called a linear potentiometer. It is the same tech that is in your car known as a fuel sending unit, volume control knobs, etc. It is most basically a voltage divider whose resistance is increased or decreased and can be measured. Aditional details can be found on the wiki at en.wikipedia.org/wiki/Potentiometer
Sort of. This is in the possible but not probable category of things. All you need is a system with very little play and active position feedback and you can get around the usual requirement of stepper motors and lead screws. What is wrong with belts though? They are really not that expensive for what they do and how well they do it compared to alternatives. At time of writing AndyMark carries timing belts for $14. Given, however, that you want the amazingly large size of 100m×50m (!) might I suggest steel cable and a roller system on 80/20. Make a free rolling system and then use steel cables to move the positions on each of the axes. Such a large system will have both hysteresis and nonlinearity so you're going to want active position feedback and PID control scheme to make sure the machine head actually ends up where it needs to be for each line of the gcode. Can you not partition that space down to smaller quadrants though? 100mx50m is enormously demanding and de facto firmly out of the budget CNC category of projects.
Interesting question. Some degree of precision is required. There are two ways to do this:
a) use an encoder associated with a wheel which reports the rotary position. That value is then related to the wheel circumference to obtain the distance travelled. Perhaps a wifi transmitter on each carriage would be more practical than wiring.
b) use an absolute positioning measurement system. Your dimensions of 100m x 50m are very large and clearly any traditional belt drive/screw drive is impractical. As an example of absolute positioning, a laser rangefinder (+/- 1-2mm) at each corner would locate the gantry. How a commercial product would be interfaced with the PID/control system is a homework exercise.
Use a robot vacuum with mapping ability. Ramp up battery and find a friend than can code . That way you could get that sort of size for way under 4000$. Simple fix. Can be made accurate with RFID sensors and will give you limitless boundaries in theory. Just a thought
In the most literal case, the answer to your apparent question (which is not terminated by an interrogation point but is otherwise phrased much like a question) is "No"... because no such thing as a "CNC" has ever existed in the sense you mean. "CNC" is an adjective, not a noun. Therefore, it is not possible to "build a CNC" at all.
Opening up our translation a little and assuming that by "a CNC" you really mean a CNC router or laser engraver or 3D printer or sewing machine... yes, it is possible to build such a machine using a linear motion system which does not contain any timing pulley or lead screw. You never asked how to accomplish that, but there are many ways of getting it done.
In response to your stated main objective... the cheapest method would be to manually drag you axis back and forth with tiedown twine stolen from Home Depot. That probably isn't going to feel very satisfactory to you, though. I think what you're actually looking for is going to be a stick of All-Thread rod and some clanky nuts that fit it, maybe even some shopmade ones so you don't have to weld them (or glue them) to your carriage. I wouldn't expect your accuracy and repeatability to be any great shakes, but it would do the job very cheaply.
Given that you'll likely want greater accuracy than that, you'd best get used to the fact that any reasonably accurate positioning system is going to be a bit more expensive than you're ready for. That's a good thing, since it means that every dog's man isn't wasting his time & money building a nasty CNC machine in a back bedroom with baling wire and bubble gum.
Just sigh and relax and surf Amazon and eBay looking for Asian import lead screws and ballscrews. They'll get you there. The lead screws will be the most accurate, the timing belts will be fastest, the ball screws will fit somewhere in the middle between speed and accuracy. Oh, they'll also wear worse than the lead screws but not as bad as the timing belts.
