CNC Control in Electrically Noisy Environments

I have a CNC mill, built from a kit. It uses servos and single-ended optical encoders. When it was originally built, the encoder data streamed over 25 feet of telephone cord -- all 28-gauge wire. My problem was a combination of attenuation, noise from an industrial environment, and signal quality. Quite a challenge. At first, the mill could not even cut simple circles, much less more complex patterns. Electrical noise in the room was inducing false signals on the line.

If noise were the issue, I would block it. For the 3-axis mill I had three long antennas going to each motor -- a bundle consisting of the motor power cables and the encoder data lines. First, I separated the power from the data lines. I changed the encoder line cables to a heavier gauge, shielded, twisted pair wire. I put test leads onto one of the mill's longest motor cables. Immediately, I could see a 60Hz waveform, which could only be the AC supply line. I moved the mill away from all AC lines and a higher frequency wave showed up. That ended up being the compact florescent lights, so I switched all the lights to incandescent and shortened the motor leads. At each stage, I was conducting test cuts with the mill. Failed every time. Now, when the mill's spindle motor was on, chaotic waveforms blasted the room. I tried everything, but I could not suppress them. I was at my wits' end. Instead of working with the retrofit kit, I was ready to sell it all on eBay and get a turnkey system.

I decided to try one more option. I considered conditioning the signal from both the motor driver and motor ends. In other words, filtering out the noise digitally. While embarking on my latest effort in data communication in a high-noise environment, I came across US Digital's Differential Cable Drivers & Receivers. In particular, I found the EA-D-L-10. It is a differential RS-422 cable driver that converts the single-ended A/B/I output from USD's single-ended incremental encoders (or any three TTL level digital signals) to three pairs of differential signals. They claim this allows the encoder to drive long cables (six feet to 1,000 feet) and reduces false switching in noisy environments. At only $12 dollars each, I bought some EA-D series modules and gave it a try. I really wanted to save my CNC mill from eBay.

I installed them as shown in the figure. It was a simple installation, almost plug-and-play. Both the transmitter and receiver required voltage between 4.5V and 5.5V. Piggybacking on the motor's optical encoder 5V supply for the differential pair, the circuit was complete. Like a miracle, my first test circular cut was perfect. After several more perfect cuts, the mill was finally in operation. For $30 dollars, I was able to save thousands. The turnkey mill I was going to buy -- one to match the specs of my retrofit -- would have cost me an additional $2,000.

Tell us your experience in solving a knotty engineering problem. Send stories to Rob Spiegel for Sherlock Ohms.

Related posts: