[mmstw]

Hi,

I've attached a couple of pictures of my desktop mill project inspired by a project by DevilMaster. The machine was designed/sketched using Blender. The frame is made of 45x90 mm (and some 45x45 mm) aluminum profile. X, Y, and Z stages use THK and NSK linear guides and trapezoidal lead screws with 3 mm pitch. The steppers are NEMA 23 size with 180 Ncm holding torque, coupled to the lead screws with pulleys and belts.

The electronics consist of a switched-mode power supply providing power to three A3977 based stepper motor drivers. There's a solid state relay for spindle (Kress 1050FME-1) control. The whole system is controlled by a Microchip PIC18F8680 based board with a graphical touch screen lcd. The code is read off a normal usb memory stick so that no computer is needed. The microcontroller does rotation/scaling of coordinates if needed. The rotation is calculated based on user-defined horizontal/vertical reference or two reference points. This I did mainly make it easier to drill circuit boards using this thing. This way I don't have to pay that much attention to how I mount the circuit board on the table. There's also a probe for touching off a conductive surface for getting the Z axis depth and/or X/Y references right.

Here's an out-of-focus video of a test running some code and drawing with a badly mounted pen:

[Jesse B]

Very nice looking machine! What kind of precision do you figure you're getting?

[mmstw]

Very nice looking machine! What kind of precision do you figure you're getting?

Thanks. I haven't measured much yet. I guess precision depends mostly on the backlash in the polyacetal nuts and linear slide tolerances. So far it *feels* ok, but maybe measurements will show otherwise. As far as accuracy is concerned, I don't know how accurate my trapezoidal screws are. Moreover, tramming the axes is a bit nasty especially due to the way I decided to attach the X axis extrusion to the vertical extrusions. I think that getting down to 0.1 mm or so would already be very good.

[litris]

The whole system is controlled by a Microchip PIC18F8680 based board with a graphical touch screen lcd. The code is read off a normal usb memory stick so that no computer is needed

Please information schematics and Hex for pic?

Saludos.

[Jesse B]

Wow, missed that line. I'd love to see some information on that if you're willing to share it. You at least need some pictures and maybe a video

[Kipper]

Hi,

I've attached a couple of pictures of my desktop mill project inspired by a project by DevilMaster. The machine was designed/sketched using Blender. The frame is made of 45x90 mm (and some 45x45 mm) aluminum profile. X, Y, and Z stages use THK and NSK linear guides and trapezoidal lead screws with 3 mm pitch. The steppers are NEMA 23 size with 180 Ncm holding torque, coupled to the lead screws with pulleys and belts.

The electronics consist of a switched-mode power supply providing power to three A3977 based stepper motor drivers. There's a solid state relay for spindle (Kress 1050FME-1) control. The whole system is controlled by a Microchip PIC18F8680 based board with a graphical touch screen lcd. The code is read off a normal usb memory stick so that no computer is needed. The microcontroller does rotation/scaling of coordinates if needed. The rotation is calculated based on user-defined horizontal/vertical reference or two reference points. This I did mainly make it easier to drill circuit boards using this thing. This way I don't have to pay that much attention to how I mount the circuit board on the table. There's also a probe for touching off a conductive surface for getting the Z axis depth and/or X/Y references right.

Here's an out-of-focus video of a test running some code and drawing with a badly mounted pen:

The controller sounds fascinating! Do you do exports?

Machine wise....moving table and precision rails= precision machining....ballscrews and shorter belt runs next time maybe...or some idlers to prevent resonance on the long runs.....Easily better than any of the "Pro" machines on ebay!

[rowbare]

Very nice. What do you use as input to the controller? Does it understand G-Code or some other standard control language?

bob

[mmstw]

Very nice. What do you use as input to the controller? Does it understand G-Code or some other standard control language?

bob

I don't think it makes much sense to process G-Code directly with the microcontroller since I'm using a computer to make/write code anyways and I don't intend to include an editor in my controller. Thus, I've written a program that converts G-Code to a binary format which my controller readily interprets. G00/01 commands as well as pause and tool change commands are passed through almost as-is. The converter turns G02/03 commands into several linear moves. What is left for the microcontroller to do is to calculate the coordinate rotation/scaling I mentioned earlier, to execute stepping using the well-known Bresenham's line algorithm with linear acceleration ramps (if we're going fast), do A/D for touchscreen control, read limit switches and the touch probe, and drive the graphical lcd. The firmware is written in assembler, of course. The current version of the firmware buffers data from the usb stick and performs the calculations right before starting a linear interpolation. If rotation is needed, the calculations create slight delays between moves. This will be taken care of later by double buffering.

As the microcontroller I'm using cannot function as an usb host (I think it's possible to do that by bit-banging but that doesn't seem like a very fun thing to do), I'm using a VNC1L by FTDI. The display is a generic KS0108 controller based 128x64 pixel graphical lcd with a resistive touch panel added on top. It's not very big but the UI is finger friendly anyway. For the video below, however, I used a pen. There are still some superfluous readings visible on the screen for debugging purposes. There are also some artifacts coming from clocking bytes into the lcd a bit too fast.

[Jesse B]

That thing is freaking awesome!

[litris]

Very nice!!!

Please more! Schematic Pic.

Saludos.

[RomanLini]

Impressive machine mmstw! Also impressive firmware, having done the same myself I know it is no easy task doing the full motion control on a PIC 18.

When you are ready to align the machine and want to check your backlash etc I developed a "live backlash" test cut procedure you can see in this thread;
Please help me backlash gurus

My machine is a similar design to yours, moving table fixed gantry and polymer leadnuts etc, I was able to get it quite accurate.

[mmstw]

Impressive machine mmstw! Also impressive firmware, having done the same myself I know it is no easy task doing the full motion control on a PIC 18.

When you are ready to align the machine and want to check your backlash etc I developed a "live backlash" test cut procedure you can see in this thread;
Please help me backlash gurus

My machine is a similar design to yours, moving table fixed gantry and polymer leadnuts etc, I was able to get it quite accurate.

Thanks. And thanks for the tip. After a couple of measurements I'll know whether I should make better leadnuts or if backlash compensation only in software is enough. So far I have noticed similar elasticity as you mentioned in one of your posts. That I checked by pulling on the spindle while having a dial indicator on the spindle bracket. It could be the polyacetal but it could also be the less than perfectly rigid frame/aluminum parts.