• 

• 

What? Why?

I started building a CNC machine in 2008 specifically for machining switch panels for Flight Sim Ver 4. I had already built the fairly passable flight Flight Sim Versions 1-3, and I wanted to step it up. My idea was to make a ton of custom switch panels with whatever I needed for both serious and arcade simulators – enough to fly without touching the mouse or keyboard. All of them would be designed in 2D CAD, then routed and engraved on the homemade CNC. That was the reason for this project.

I learned a lot along the way and the machine ended up working better than I expected. All the panels for Flight Sim Ver 4 and Flight Sim Ver 5 were cut on this machine. Even with the insanity of bolts in tapped MDF holes, it was still alive and kicking when I decommissioned it in 2020. That was to make room for a bigger machine.

Some Examples

• 

• 

• 

• 

• 

• 

• 

• 

Rough cuts with 1/8″ end mill. Finer stuff with 1/16″ end mill. Super pointy 60 degree V bits used for engraving. All low cost bits off Amazon. Spindle motor is a Dremel with an 1/8″ collet so I was restricted to bits with 1/8″ shank maximum.

Overall Design / Structure

This was before Shapeoko‘s and X-Carve‘s so there wasn’t much of anything on the market outside of professional machines way beyond my means. I only wanted to cut plastic, wood, and maybe engrave some PCBs.

• 

• 

I checked out a number of online plans and built something inspired by a number of them. Rockcliff CNC (website no longer found) was the general approach I took, though my electronics, motors, linear slides and rails were all a hodge podge of custom work and ebay scavenging.

Structure was all 3/4″ MDF. Sounds crazy. More temperature stable than plywood, so sounds like a win. Aluminum was too pricey, and more trouble than I wanted. This was before I had a decent shop, and all fab was being done on a friend’s tablesaw and bandsaw. 3/4″ MDF was plenty stiff for the materials, cut depths and feedrates I ended up using. Milling the spoil board flat on occasion was enough to deal with any moisture related shift which I don’t think was really that severe. Check out some of the results above.

• 

• 

Motors

Steppers motors and drivers weren’t as easy to find since this was pre-home-3D printing. I picked up a 4 piece lot of steppers of eBay that were pulled out of old equipment. Like 1980’s old. From the pics on eBay, they looked like decent sized bipolar NEMA23’s. Based on that alone, I very scientifically determined they would be fine.

• 

• 

Drivers

Stepper drivers were more of an expensive specialty item back them. No standard Pololu-ish drivers yet. To keep costs down – my overall machine target was $300 – I found a site where a hobbyist sold bare PCBs. His page is unfortunately gone now. I believe it was something along the lines of PMinMo.

His boards had L297 drivers with L298 power electronics. I bought a couple of his PCBs, tracked down the components I didn’t have in my junkbox, and populated the PCBs. Ohmed out the motors to figure out their pinout, dialed in a reasonable current limit, and with a little extra heatsinking and a fan, they worked years without issue. Microstepping was only half stepping but with the high reduction ratio of leadscrews, I had plenty of linear resolution.

• 

• 

Linear Rails

New linear bearings and rails from somewhere like McMaster-Carr were too expensive. The cottage industry selling CNC parts to hobbyists hadn’t sprung up yet, or at least the few that were around weren’t cheap enough for me.

I scoured eBay looking for linear rails pulled out of equipment that were under my budget. As a result, each of my axes was hacked together with a different type of bearing and rail system. All were plenty stiff and worked out fine.

The Y-axis has the original concept from the Rockcliff design – just oiled bronze bushings and precision rod. The X and Z axes were hacked together with the eBay finds. Fortunately those each had ball bearings, which were much smoother than my Y-axis.

No anti-backlash nuts. Just a piece of nylon dowel, center drilled on a friend’s lathe and tapped. Then pressed into an MDF bracket. Nylon gave low friction and a tight fit without any backlash even after years. Leadscrews were standard threaded rod. No Acme or anything fancy.

• 

• 

Software

Control was step and direction signals coming directly from the PC printer port via Mach 3. Mach 3 was actually a good chunk of the budget, but it was the best choice at the time. No UGS or grbl yet. Mach 3 was the only real option for parsing g-code and controlling the motors. Have to hand it to Art Fenerty at Mach 3 for his software successfully generating all the step pulses real-time despite running on a Windows PC. Worked great.

For CAM/g-code generation I used Cut 2D from Vectric. It wasn’t as full featured as their higher end products, but it met my needs and was a lot cheaper. What’s available for CAM now has totally changed. Check out this reddit thread, for lots of options.

Next Steps

In 2020 I decommissioned it to make room for a new machine. Homebrew as well, but mostly aluminum extrusion and belts instead of threaded rods and MDF. Barely begun as of end of 2020, but with access to Fusion 360, 3D printers, and tons of affordable components, it’ll be an easier build.

The original machine on this page had a capacity of about 12″ x 12″ x 2″. Shooting for at least 18″ x 24″ x 5″ with the new machine. Palm router instead of a Dremel for the spindle and a stiffer structure should allow for faster feed rates, deeper cuts per pass, and access to 1/4″ collet bits. It’s barely begun, so only a few photos up on CNC Rev 2.