Sure. I'm planning on a 3 axis mill, hoping to do light steel work eventually. I have some random chinese brushless DC motors about the size of my two fists put together.
Encoders are the AMT102V (you can grab them at digikey), they are bloody amazing for the price and they come with a ton of different shaft adapters. I've used a few before, you really need a hardware/FPGA quadrature decoder for them as you can get up to 2048*4 quadrature edges per revolution.
Servo drives are Geckodrive G320X. Gecko is basically the agilent of small stepper/servo drivers imo; in the optics lab that I work in, everything runs off zabers or geckodrives. Just don't forget to heatsink them.
Smooth stepper for the motion controller was my first choice as I have heard some rave reviews. I somewhat disagree with the linuxcnc folks; a pc is not a real-time system (nor should it be), better to leave the motion control to a hard real-time device. For $150 or so you can't really go wrong.
The most expensive part by far will be the ballscrews. I'm still trying to spec some good ones, which really just means I am trying to decide if I should get real THK/Thompson precision screws or just buy the cheap Chinese ones Also, 4 race THK linear guides would be awesome, but I don't know if I can afford them/find them. 2nd hand ones are plastered all over US ebay, but none for me out in AU; shipping long hardened steel tracks over here isn't all that cheap
I plan to make the frame out of aluminium/steel. Unfortunately I have been quite busy lately so haven't worked on it. Also been spending my little spare time with the 3d printer....
Just a quick side note: encoders on steppers are a no-no. Mach3 doesn't support closed loop steppers (servos are fine), and steppers have a non-constant torque throughout their rotation iirc which messes with many control algorithms.
having started out as a total noob, and bought a TAIG 3 axis mill, with a 75$ 3 axis driver, and upgrading to G240 drivers, but limiting the steppers to 2/3rds of what they are capable..
i would say:
you need to understand a few things first:
absolute accuracy is second to machine stiffness for many but not all machine operations.
it is better to repeat a machine operation a few times to get that last .001 inches of steel removed than attempt to remove it the first time with stiffer end mills, and better accuracy.
one example, as i've recently modified my TAIG mill significantly.
I can drop a 3/8th 4 flute end mill at 1 inch per minute into an aluminium block with a 1/4th hp belt driven spindle.
note that: that is nothing compared to what is possible with perfectly sharp HHS or carbide, at three cubic inches per minute per horsepower of metal removed.
I cannot exceed that unless i play games with the RPM to avoid resonance, which is specific to the diameter of the end mill, and the number of flutes, and the material.
cast iron is easier to machine than aluminium without lubrication in my experience, but this depends on a number of things.
my milling machine cannot generate enough force to snap a .25 inch end mill. (which is great, because i can't break most of my tooling.
there were many times where a 3/8th inch end mill stalled the X/Y axis without stalling the spindle. ( this means you can save the work by pressing the stop button quick enough.
in any case, unless you're trying to machine a custom scroll compressor...
you're better investing the $$$ in a stiffer and bigger machine than in ballscrews.