Projects Publications Resume Contact About Youtube Donate

Electric Arc Furnace (aka Carbon Arc Furnace)

Cutting Metal

I decided to make the EAF frame out of aluminum. I thought it would be a lot lighter than steel and corrosion resistant. I have used my 4.5" angle grinder in the past, but getting cuts square and even is pretty difficult. I started with a reciprocating saw. If I was careful and slow enough, I got pretty good results, but I wasn't fast or consistent.


At this point, I'd like to thank my friend Chris for helping me realize what I didn't want to buy. He had an abrasive chop saw and wasn't wild about the noise, dust, and sloppy miter vise. Then he upgraded to a horizontal band saw, which turned out to be quieter, slower, cleaner, and difficult to adjust. My search led me to some expensive wet-cut or cold-cut tungsten carbide tooth (TCT) saws and then to more reasonably priced dry-cut TCT chop saws. They seemed great because of the TCT blade, lower RPM, and no abrasive dust, but they all seemed to have the same sloppy miter vise setup. I decided that what I really wanted was a wood style miter saw that could also do metal. A quick look at motor and TCT blade RPM limits told me that the average wood miter saw spun way too fast for metal cutting blades. Luckily, I came across the

1. Check square tubes for twist after checking for straightness.
2. Don't assume your welding table is perfectly flat.
3. More clamps is almost never a bad thing.

Linear Bearing

I only found two good pages or videos that discussed a linear bearing for square tube. Both were pretty half baked, but they got the idea across. I used BSB ABEC 9 skate ball bearings, 5/16" fasteners, and 3/4" 0.0625 wall aluminum square tube to make a linear bearing for square tube. With this arrangement, the square tube must be smaller than the diameter of the skate bearings to clear the rail. I drilled two outboard through holes on one tube face and two inboard through boles on the other face. That made sure all 4 tubes were exactly the same. Access to a drill press comes in handy when being square and accurate is important. Otherwise, you'll have to do very careful layout and punch marks on all sides of the part. Red thread locker is handy for making custom bolts from threaded rod and nuts.

Actuator

I chose to use 5/16" 18TPI threaded rod as my linear actuator since I had a lot laying around after building the linear bearing. The fine pitch makes for pretty slow operation, but it should be able to handle a good amount of weight. I simply tapped threads into a piece of aluminum square tube to ride on the threaded rod and attach to the linear bearing. I found a 5mm (stepper shaft) to 8mm (threaded rod) flexible aluminum coupler on Amazon, marketed for the RepRap 3D printer.

Electrode Jaws

I chose to make water-cooled aluminum electrode jaws for holding the carbon electrode. I cut a few 1"x1"x5" pieces from 1"x5"x9" plate with my TCT saw. Luckily, my friend Chris has a mini mill, which worked great for face milling the rough cuts, drilling all of the holes, and milling the cylindrical electrode contacts. I drilled three holes to form a u-shaped cooling passage in each jaw. I plug welded the third hole to convert from a 3-opening F-passage to a 2-opening U-passage. I used 3/16" hose barbs and 3/16" hose for plumbing. After sanding one end of the barbs for precise fit, I used J-B Weld to permanently install them in the jaws. After that, some carefully drilled 1/8" pivot holes and clamp holes completed the drill press work. I hand drilled and tapped one of the clamp holes 1/4-20 and hand drilled the other for clearance. I used a 1/4-20 socket cap screw and a split lock washer as the electrode clamping mechanism. The lock washer provides sufficient tension. Carbon electrodes are brittle, so be careful to not over tighten.

Electrode Arm

I made the electrode arm mostly using 2" square tubing left over from building the frame. For the bottom two linear bearing attachment holes, I measured the threaded rod distance with dial calipers with the linear bearing installed. Then careful marking, punching, and drilling made for a pretty exact fit. The holes are a tad oversize for the threaded rod for a little wiggle room. I drilled the two gussets, installed them on the top threaded rod, and then welded them with everything in place and supported. This resulted in zero slop. Notice that the arm is connected to linear bearing bolts on the front and sides. This arrangement provides good vertical and lateral stability. With everything just finger tight, there is only a tiny amount of play in the vertical direction from the flexible aluminum shaft coupler where the threaded rod meets the stepper motor. The wood 2x4 link provides thermal and electrical insulation.

Control System

I used an Arduino Uno and an Adafruit Motor Shield V2 as the brains of the control system. Power and project enclosure came from two ATX PC power supplies. An upper limit switch came from a spare TIG torch switch. The up/down switch was the power switch in the second power supply enclosure. The analog input is fed by a voltage divider for voltage protection and a bridge rectifier for reverse polarity protection. I welded a 4"x0.25" plate to the base of the frame to mount the power supply, microcontroller enclosure, and stepper motor.

1. The Adafruit Motor Shield V2 Library.
2. A patch to the Arduino library: eaf_arduino_patch.txt
3. The Arduino sketch that ties it all together.

• eaf_controller_20150427.ino is the sketch from the first arc test. Full revolutions between measurements, 28-30V.
• eaf_controller_20150525.ino is the sketch from the second arc test. Half revolutions between measurements, 40-45V.
• eaf_controller_20150725.ino is the sketch from the third arc test. Tenth revolutions between measurements, 38-42V.
• eaf_controller_20150921.ino is the sketch from the fourth arc test. Tenth revolutions between measurements, 40-45V.
• eaf_controller_20150923.ino is the sketch from the fifth arc test. Tenth revolutions between measurements, 35-40V.

Refractory

My first attempt at furnace wall refractory involved a suggestion from a friend and some internet searching. I bought alumina and kaolin clay from a local ceramic supply warehouse. I mixed up a very small batch to measure shrinkage and final density and I just about destroyed my forearms and hands in the process. That stuff is incredibly difficult to work by hand and I would like to finish the furnace body before the inevitable heat death of the universe. So, I switched gears and bought two 55-lb bags of 3000F Kast-O-Lite 30 LI castable refractory from Legend Mine. I bought some "14 inch" and "10 inch" concrete form tubes from a local supply place. Those dimensions were actually pretty rough. I ended up with about 1.5" between the forms and a section of 2x4 was perfect for tamping. I used hot glue to tack everything together and then put on a couple liberal coats of polyurethane for waterproofing. This stuff was a lot easier to work than the clay, but it was still pretty back-breaking work. It didn't exactly pour into the forms either. I filled them with a hand trowel and tamped it down little by little. I made two extra parts to use the remainder of the second bag. I made a short stand, nearly the same size as the furnace bottom, to set hot stuff on. I also made an air/fuel flow section in case I want to try some fuel-fired furnace or forging work. I realize the gap where the sections meet will not be air tight and this layout is not ideal for a fuel-fired furnace, but the furnace should be as small as possible for arc melting and the additional section adds some versatility. I followed the 100F/hour initial heat up to 500 in my oven and then I lit a charcoal fire in it.

Crucible Tongs

I have not seen crucible tongs for perfectly cylindrical crucibles before. All of the casting tongs I have seen have either been hot-dog size for jewelers' crucibles or something that takes advantage of the curved profile of something like the #10 SiC crucible I mentioned above. A crucible that gets wider at the opening seems like an easier geometry to grip securely because of its roughly wedge-shaped profile. Obviously, you can lift a heavy cylinder if you can pinch it tight enough. However, the lubricating and brittle nature of pure graphite made me want to try something other than just gripping it really tightly. I decided I would leave an inch gap in the carbon rods the crucible sits on at the base of the refractory. This left enough room for a lifting tongue to be inserted under the crucible, which greatly reduces the required grip pressure. Then it was just a matter of devising something that would hold it tight enough to keep it controllable. The following two photos show what I came up with. The circular metal band is lowered over the crucible. Then the nut is tightened on the threaded rod, which forces the lifting and retaining tongues over the crucible. When it is snug, the crucible is pulled into the vertical and lateral supports so it is restrained in every direction. My only concern with this approach is the relatively thin (1/16") band, which may deform or stretch with heat and load. This tong design requires the refractory furnace walls to be lifted off the crucible; the crucible cannot be lifted out of the furnace. The first test will certainly show if this setup is adequate.

Power and Efficiency

I hooked up my clamp-on current meter and my multimeter to get a rough recording of instantaneous current and voltage. I recorded the measurements from arc strike to long arcing. Afterward, I plugged some of the numbers into a spreadsheet to figure out a rough voltage range that should be close to maximum power output for my configuration and equipment. The optimum voltage range is around 35-40V. With the welder set to the maximum 100% duty cycle at 200A, the maximum power output range is between 7KW and 8KW. That is with the 3/8" x 12" carbon electrode. If the arc gets too long, high voltage range, the arc resistance shoots up and the current and power fall off. If the arc is too short, low voltage range, the power also falls off.


Next I sustained an arc for about 30 minutes. I was testing some soft lower limit code and it took a while for the electrode to wear down on DCEN. I recorded my power meter at the same time, which allowed me to do a rough efficiency calculation. Power consumption was around 5KWH over 30 minutes or around 10KW. The arc power was around 7.5KW, which makes for an overall efficiency of around 75%.

Preheat

My first big melt with the 7" crucible was too big. I melted about 10 lbs of steel before my exterior 240V circuit breaker overheated. It was just too much current for too much time for my stock house wiring to handle. I switched to the 5" crucible, about half as much steel, and I decided to preheat with charcoal. I filled and lit my charcoal starter and then waited for all of the charcoal to ash over. Then I used tongs to fill the area around the crucible with hot coals. I put the lid on and let that preheat, without forced air, while I worked on the mold. The charcoal and ash also provided additional insulation between the crucible and the furnace body.