How Can We Deal with Interference from a Welder?
by Bill Kimmel

Question:

Bill:

I need your opinion about a problem we are experiencing here in our Model Shop. There is a room where they have computers running programs for utilizing the CNC equipment out on the Model shop floor. This room is approx. 5 feet x12 feet. Immediately on the other side of the wall is a large metal table where a TIG welder is occassionally used. When this welder is used, it causes these programs on the computers to lock up. It appears they have a conducted and radiated immunity problem. I was going to suggest using power conditioner (UPS) on their ac lines. I don't know if this will fix their problem or not since it appears the noise is also being radiated. They tried putting a large metal plate up on the wall and grounded it.

The first thing I suggested was they either move the welder or the computers. That does not seem to be an option.

I was hoping you could give me some advice on what to do.

TW

Answer: The arc welder generates copious amounts of RF interference, primarily in the 2 to 20 MHz range and also generates substantial power/ground transients when the arc is struck. The high frequency noise goes out the weld cable to the work piece. The work piece is usually grounded to facility ground, so the high frequency return path tends to be building steel, conduit, etc. The transient noise would largely follow the same path. You could be radiating energy off the weld cable or conducting it through facility ground, as you suggest, but considering the description of the problem, I consider conducted interference, especially transients, a more likely candidate. Without having a better picture of the placement of the welder relative to the work piece or where your computer equipment is located, I have to guess a bit. Here are some suggestions. On the computer side: A power conditioner will help with conducted interference provided the conditioner and the computer equipment are all referenced to the same ground, preferably a single point ground at the conditioner. Grounding to a low impedance ground grid, to keep ground impedance low is a second choice. If your problem is conducted, it is because you have lots of ground noise, and a power conditioner can't correct for ground noise distributed between your various computer elements - hence the grounding requirement. Make sure your conditioner has a transformer, not an auto-transformer - you want to make sure the output neutral of the conditioner is at ground potential. If you have an isolation transformer handy, you might try it, instead of getting a conditioner, again, grounding the neutral of the secondary, as per code. Make sure your computer power is drawn from a line separate from the welder. Radiated interference would almost certainly come into the computer via the power cords or the data cables, and is best handled by shielding. The problem frequency is probably lower than what can be handled with clamp-on ferrites, but it might be worth a try. Put clamp-on ferrites on each cable, particularly the unshielded cables and power cords. You should put on several in a row or, better yet, make several passes through the ferrite. Inspect the shielded cables, make sure they have half way decent terminations. Unshielded cables can be replaced with shielded cables. If your cables are not too long or are spread out, you might try making a temporary cable shield using aluminum foil, wrapping each cable and grounding the foil with copper tape at each end, although at this lower frequency, you can probably get by with grounding at one end. On the welder side: Conducted problems can be minimized if the work piece, the weld cable return and the welder are all silngle point grounded - preferably at the welder. This will minimize noise currents being driven into facility ground. The metal plate you mentioned could serve this purpose if it were grounded at one point - immediately at the welder. The work piece can be grounded to the plate if that meets safety requirements or with a separate heavy wire back to the welder ground. The ideal scenario would be to have the welder and work piece both grounded directly to the same plate. The plate should be a continuous sheet or, if a splice is necessary, overlap it and use plenty of screws to fasten the splices. You probably won't want to monkey with the power to the welder, especially if it runs off of 480 VAC. But do make sure your computer doesn't tap power from the welder source. Radiated problems can be minimized by routing the weld cable/return adjacent to a ground plane to minimize loop areas. The plate mentioned above would serve. If that doesn't help, your next step would be either moving one or the other of the equipment (you already rejected that possibility) or building a room shield.

EMI/EMC Archive

		Click here to get your listing up.