First time building a XY gantry, want to use Raspberry pi?

I'd use grbl running on a separate Arduino, then use the Pi to send g-code over serial, maintaining lots of free time to do experimental work and onboard data.

I'd also use "real" stepper drivers, to keep things simple and reduce room for error, but that's up to you. (I like these guys: https://smile.amazon.com/gp/product/B07B9ZQF5D)

Those limits should be fine, especially if you don't need them for really precise absolute machine positioning (usually not a good idea on switches under $100)

NEMA 17s love 24V, they're super quick. Just keep the current levels reasonable (no higher than 1.2A, most likely).

 

Technically one has the power capacity to do both y axis motors, but I wouldn't do that; you can run into current-sharing issues due to minor deviations from the motors' nominal specifications. Typically that shows up as one having a slightly lower resistance or impedance, "stealing" more of the current from the driver (which can't simultaneously adjust to both motors), and the other motor stuttering and the gantry racking over. Best, to minimize troubleshooting possibilities, to just have one motor per driver.

I use a MeanWell NES-350-24 for 24V. It's an older, more expensive (~$50), model, but appears to be more reliable. There's a cheaper one for around $33 too, that doesn't get reviewed quite as well. That's the model in the OpenBuilds PSU, which is nice if you don't want to deal with mains wiring, but otherwise quite expensive for what you get. Depends on your level of comfortability.

The Arduino can be powered from the Pi over USB quite happily, and you probably wouldn't even need the full 2.5A supply to do so. But you will need to think about powering the Pi in general. A 24-9V buck converter could do it (tons of overhead on a 350W supply), but since the Pi is technically still a computer, you'd want a nice one with good filtration and regulation. Probably easier to just get a decent second 9V PSU. I like the little MeanWell units, I have 12V and 5V ones, no idea if they make a 9V one but it's worth a look.

Limits are per axis, the motors are irrelevant there. One per if you just need homing "the machine is about here, give or take", two per if you want hard travel limits to keep from crashing the machine. Grbl is good with soft limits though (won't even motion-plan a line of G-code if it takes the machine out of bounds).

Homing's fairly approximate, just gives the machine a rough idea of where it is to within the repeatability tolerance of the limit switches. You'd want to then zero G54 work coordinates based on the actual work being done and the equipment being worked on (like a pallet or vice in a traditional mill, maybe a corner of your tray here).

 

If it's that simple, I'm sure you can. It would be pretty similar to my SliderCam build, here: DSLR Video Slider

From that, imagine this single-axis control is made in Python instead of C, and you'd get the gist:

Code:

int motionControl() {
  totalMotorSteps = currentDistanceInt * 5; //calculate total steps (0.2mm = 20-tooth gear on 2mm pitch belt; 40mm per rev, 200 steps per rev, ergo 1/5th mm per step)
  pulseDelay = (1000L * (currentDurationInt - (currentStepsInt * shutterDuration))) / totalMotorSteps; //how long to pause in ms between STP pulses to the motor driver
  intervalDistance = totalMotorSteps / currentStepsInt;

  //once per overall run
  if (travelDir == 0) digitalWrite(dir, LOW);
  else if (travelDir == 1) digitalWrite(dir, HIGH);
  //Serial.begin(9600);
  //Serial.println(pulseDelay);

  //step loop
  do {
    digitalWrite(stp, HIGH); //fire motor driver step
    delay(pulseDelay);
    digitalWrite(stp, LOW); //reset driver
    //btnVal = readLcdButtons(); //check there's no stoppage - this takes too long and significantly slows motor; use reset for stop!
    currentStep++;

    //at end of each step
    if (currentStep % intervalDistance == 0) {    //if current number of motor steps is divisible by the number of motor steps in a camera step, fire the camera
      digitalWrite(trig, HIGH); //trigger camera shutter
      delay(80);
      digitalWrite(trig, LOW);    //reset trigger pin
      delay((shutterDuration * 1000)-80); //delay needs changing to timer so stop button can be polled
    }

  }
  while (currentStep < totalMotorSteps);

} //end motion control

From RobTC/SliderCam (Obviously ignore the //commented out bits that are for bugchecking or features that didn't work quite right).

So it's not complex, just a bit of programming and adjusting as you go. Especially if it's on a grid, since then you can model the actual layout as a matrix, and just move around it in software, similar to the way I did the GUI elements (most of the program, realistically) in SliderCam.

It's faster and easier to build it as a 2-axis CNC build with Arduino/grbl, because it's really easier to calculate G-Code and send it over serial than implement a motion controller. But if you have the time to play with it and want more of a challenge to learn the hardware/software, then yeah, rolling your own is super fun.

 

i use a raspberry pi with a protoneer cnc shield to run my workbee. it's pretty cool i think...

 

Oh, yeah, then definitely make the most of the budget you have, no need to worry about driver prices. And I like those! They seem very nice. And you can get the cheapest one for this use case as well, which according to the table is $99, not $75, but whatever.

The 24V supply is like, 15A. You probably won't see more than a 2A power draw from all three motors simultaneously under load!

 

Can't power it off a 24V supply, though. I mean, maybe you could, I dunno, but it seems like overkill since it's another motion operation to implement in code. Could a 24V solenoid plunger work? They're a little power-hungry, but you have ample to spare and they're easy to set up physically and in code.

Weird that the ACRO wiring kits don't come with drag chain but the other machines do, I'm not sure where you're supposed to put the flex conduit during operation without it dragging around your parts. Other than that, which you could probably discuss with them at their main part store email or find on their Youtube, it seems like a solid option. For my laser, I bought all the wiring and connectors and drag chain and everything forever separately, and it's not worth it if you're not anticipating a bunch of future motion control builds. Just grab the $100 kit.