I used a dremel this time instead of a router, since the router was pretty much overkill for engraving anything. I made a mount for the dremel by cutting up the little plastic guide that comes with it. It has an approximate O.D. of 1″ and acts as a nut that you can clamp with a shaft collar, and still unscrew the dremel from the mount if you need to get it out. I welded the shaft collar to a stainless steel bracket.

I haven’t made any boards yet to see if the dremel has much z-axis play.

There’s nothing special about this machine electronics-wise. Since I have NEMA23 steppers and 16TPI lead screws I don’t need a really high voltage drive or anything special. I’m using a 24VDC linear power supply with two automation direct stepper drives and one from Schneider. I didn’t connect the home and limit switches as I don’t think they have much value on machines this small. It’s a lot of work to do proper limit switches that disable the drive and signal the controller that a limit has been reached. Also to have home and limit switches on each axis you need two parallel ports. TurboCNC, being run in DOS, can only use the onboard parallel port so you can’t use PCI parallel port cards with TurboCNC. Only Mach3 can recognize inputs on the extra card since it’s address will map outside of the legacy range.

Power Filter Boards

I made these power filters for a few reasons. One is to protect the stepper motor drives from the inductive spikes that can come off of the motors.  They can be pretty big. Another reason was that I wanted to use a switch-mode power supply to run the drives. There’s a lot of good reasons for this:

1. Switch-mode power supplies are cheaper than linear power supplies.

2. Switch-mode power supplies are smaller than linear supplies for equivalent output power.

3. Switch-mode power supplies come in a wide range of voltages, so you can run the stepper motors at the highest voltage possible. (The highest voltage your controller supports) This allows for more torque from any motor because a higher voltage will push more current through the inductance of the motor than a lower voltage. It will also do it quicker, so you should get a little more speed too.

However, I wasn’t sure that the chopper-style stepper driver would be okay with a switch mode supply, as it pretty much shorts out the supply every time the chopper turns on. So I needed a buffer.

I found this article at EETimesAsia by John Betten from TI. I modified the circuit for the voltage levels I wanted to run, and also found a suitable replacement for the FET since I couldn’t find one at the time. Here is the original schematic:

I’m using an IR IRFP9140N in place of Q1. I also replaced D2 with a 56V TVS from ON semi, 1.5KE56A. I used 56V because the LMD18245 motor driver IC I have has a continuous rating of 55V and an absolute max of 60V. I also oversized the output capacitor just to be on the safe side since I had some big ones laying around anyway. They are 22000uF 100V Panasonics. They are overkill, the voltage is rock-solid even when the motors are running at full clip. I wanted to be able to recycle the boards though if I ever upgrade to a bigger machine and have bigger motors. Here’s the layout for my circuit:

Opto-isolated Parallel Interface Board

I designed this parallel interface board after killing a parallel port with a breakout board that I bought off the internet. I think it just pulled too much current from the port. I designed this board to pull the smallest amount of current from the parallel port as possible, while also providing good drive characteristics for outputs. This board is customized to my application, so the voltages and bias might not be appropriate for all.  Check to make sure your inputs will work before using the values here. I couldn’t find the schematic, just the layout but it’s not too complicated to figure out if you have the datasheets for the TLP2631 and the SN74LS244N.

The 74244 is just a buffer; high impedance inputs and good drive characteristics. All inputs and outputs have RC filters. Adjust these values to suit your application. The 2631 is an opto-isolator that keeps the PC’s voltage potentials isolated from the CNC machine’s voltage. That way no voltage spikes hit the PC and there are no ground loops with the power supplies. Complete isolation! You just have to run power over from a spare drive connector on the PC to wherever you put this board.

The astute observer will notice a fried resistor on the board. That’s the board doing just what it was designed to do, protect my computer from dumb screw-ups. I accidentally miswired one of my limit switches during the build and the little resistor took a hike. I could have just replaced the resistor, but I’m not using all of the inputs so I just moved the wire over and remapped the limit input pin. Here’s the layout. Click to view full size.

Stepper Motor Controller

I used the “Microstep” from EAS Electronics. It’s worked well, but the switching frequency is low enough to create an obnoxious whine when the motors aren’t moving. I read in the LMD18245 datasheet that you can adjust the chopping frequency, but I haven’t bothered to try. It’ s loud enough anyway when my router “spindle” is turning that it’s not too important to me.

It’s PIC based and the source is available if you want to modify it to do custom stuff. The microstep ranges are 10,8,4,half, and full step.

My next machine will use Automation Direct’s stepper drivers. They aren’t as cheap as building your own, but they are pretty close. I used two of their STP-DRV-4035 drives in a project at work along with their stepper motors and was pleased with the results. If you need really fine step rate control, they have one that is programmable but it costs more.

I found some mechanical engraving bits at Think & Tinker which is actually here in Colorado at Palmer Lake. Their URL descriptor says they offer “Instrument and PCB prototyping equipment”. They also sell carbide engraving bits, and the particular one I selected was the 60° cutter, part number #EM2E8-0625-60V. The bits came in a nice container and are labelled as having a 0.005″ tip. The shipping was also wicked fast. Granted, it’s only a few hours away from my location but I ordered them and got them in the next day.

I’ve only cut one board, but I’m really happy with the results. I successfully achieved 0.008″ isolation between the pads of the devices, and now I can finally route traces inbetween the pads of 0.1″ headers. I couldn’t do that before.

Most of the machine is made out of either aluminum or acrylic. These materials are both easy to work with when all you have available is various hand tools and a drill press with a cross slide vise. The motors are NEMA 23 high torque, the threaded rods are 1/2-10 precision ACME and the nuts are anti-backlash. This results in pretty decent X-Y movement. I made the slides from extruded aluminum profiles available at the hardware store and some strips of Teflon.

I made all of the electronics that support the machine. I made the optical home and limit switches, as well as both parallel port interface boards. I also assembled the stepper motor drivers and built power filters/regulators for the input to the stepper motor drives on all three axes. This prevents inductive feedback spikes from the motors ruining the stepper motor control ICs. I need this because I’m running the motors at 48VDC and the controller IC has an absolute maximum voltage rating of 60VDC. I also wanted to keep any switching noise and voltage dips from the switch-mode power supply out of the stepper drives. Each filter has a FET-regulated output that clamps the voltage and then sends it out to some big capacitors to prevent the voltage from dropping.

The parallel interface boards completely isolate the controlling PC from the drives and other electronics. Buffers handle all the appropriate levels and opto-isolators keep the ports safe.

I’ve run it extensively with TurboCNC, but I don’t get any of the limit and home functionality because TurboCNC only supports legacy port addresses, and most parallel port PCI cards can’t map to legacy addresses. This leaves me with only the one on-board parallel port, which is mostly used up by the 3 axis step and direction signals. It works in Mach3 CNC, but I can’t afford to buy that program right now. So I’ve used it in evaluation mode for doing simple text engraving on plastics and acrylic. Oddly enough I couldn’t get it to work at all in Mach2. The pulse train output to the stepper drives was inconsistent enough that the motors would stall. I even tried it on a few different computers, one of which was a brand new XP install.

The machine mostly cranks out PC boards thanks to Eagle CAD and PCB2GCode. I’ve made quite a few since the machine was finished. It does a pretty good job, and holds flatness to a couple thousandths. It’s enough to get routine 0.020″ isolation on traces and pads, and clean 0.012″ trace widths. It’s a bit slow because of the low-buck slides, the motors stall if I speed it up much. It’s not exactly a high precision machine, so I can live with slow.

I’ve got some better parts since I built this one, and I’m planning on building another one with much better accuracy in the near future. This machine won’t handle the fine traces necessary to make boards with the newer components.

Here’s a video of the machine in action:

You can see more videos at youtube: imsolidstate’s CNC machine