I upgraded the 5 resistor plate to a actively cooled 12 resistor plate to limit the current draw and therefore the power and heat created in the components (resistors, cables, printhead, transformer).
With the resistor plate and the continuous current operation I could measure the current draw for the first time - my amperemeter couldn't measure the 250ms pulses before.
When idle, the system draws 1,240A and if the transformer secondary is shorted (toolhead on buildplate, copper on copper) it draws around 3A.
I also measured the temperature while testing with a infrared thermometer.
When idle the resistors and the transformer stay below 50°C. The system can stay idle the whole time without overheating.
I tested measuring the temperature when shorted, too but unfortunately the area of the cable above the toolhead got too hot, so that I must abort testing. The toolhead will definitely need some cooling. The nozzle got very hot, too and changed its color to a more brown color.
I tried to measure the current while melting wire and it looks like it even drops under the measurement form idle mode, but it's hard to say whether this was a correct measurement because I could not hold the hight and feedrate steady per hand. Will continue working on X, Y, Z and E axis for testing it again.
A thought about cooling:
I think I could use a powerful water cooler for the hotend if air cooling would not be enough and if the cables heat up too much I could use 35mm² or 50mm² cables. For the resistors it seems like air cooling would be good enough, but will see that in longer test runs. Maybe I could add heat sinks underneath the fans for better cooling if needed.
I tried to weld a wire per hand with continuous current. It's not easy to feed the wire fast enough per hand, therefore the sparks. The wire melted in droplets, which welded good to the surface for which I used this time a sheet of metal instead of the fly screen. With stepper motors the wire should melt in a connected line. The energy from the power supply is enough and could maybe be less, but the resistor block gets too hot to touch after a few seconds and would likely fail after minutes. I think the printing speed will later depend on the power used and the power could be regulated through on and off cycles.
Will order more resistors to divide the load between them.
Tested to melt serveral "layers" of wire on top of each other.
The "printed object" welded very good to the surface - could almost lift the buildplate on it until it broke on the weakest spot - I tested to break it in two halves and it was solid and very hard to break - like wire.
I did a quick test of the new concept and first it works both welding and softening luckily without welding the head to the surface. I tested both with a 250ms pulse and ca. 2645A which works good. But then I burned the fly screen by a softening test and the steel from it get stuck to the nozzle. Since then the wire get stuck in the nozzle like @Bharbour said.
So I will try to weld or melt the wire with pulses from the spot welder power supply when it touches the workpiece with clearence between the nozzle and the workpiece. It seems like the machine is turning more and more from a 3D spot welder to a 3D resistance welding machine by excluding ways which not work.
Will also test whether the power supply can handle short pulses without upgrading it like eg. 25ms on, 250ms off, 25ms on and so on. Doing so it should be possible to regulate the intensity of the heating by the duty cycle of the power supply like the temperature of the extruder on a 3D printer.
I had an idea how I can change the working method to be simpler and more reliable.
I want to change it from bending, hopefully hitting the wire and welding to hopping from spot weld to spot weld.
I ordered welding contact tips which should act as spot welding electrode and wire nozzle. Doing so will hopefully make the whole concept a lot simpler and more reliable by removing the 4th axis and the risk of missing the wire. This would also simplify the electrical setup, firmware and software. Maybe with this method it could also be possible to do things like cutting the wire at some point and laying a new wire at another for travel moves.
I will test the new concept as soon as the parts arrive.
Here is a prototype of the wire bending & spotwelding toolhead.
The toolhead consists of a copper electrode in the center and a wire feeder which can rotate around the electrode to bend the wire in the moving direction.
There are still a few things to figure out:
- Using this kind of toolhead the machine will later have 5 axis (X, Y, Z, E and rotation) which requires software and firmware which supports it.
- The rotation axis needs an endstop, too. Maybe I could use a hall-effect sensor on the top plate and a magnet on the gear. Will test that the next days.
- There could be a problem with the bowden tube colliding with the gantry system.
- The wire is directed at the electrode, but if the electrode misses the wire the print can fail. Maybe a greater diameter electrode could help, but it could also increase the distance of two wires next to each other.
I addad a hall sensor to the toolhead to act as an endstop.
- Ordered a ∅6mm copper rod for using it as electrode, so that it's more likely that it will hit the wire.
The next step will be building the frame and finding a way in which the PTFE tube can move without colliding with the rest of the spot welder.
Until now, I used tie wire / floral wire for testing but today I tested welding wire. I was concerned that it could be too conductive because of the thin copper layer on it but it welded very good to the bottom layer and to itself.
I think welding wire could be the perfect building material because it is available in spools, it is high quality material and it's very cheap - sometimes even cheaper than e.g. PLA filament.
- Flux core wire and stainless steel wire works, too.
- Will start testing with 0.6mm wire rater than 0.8mm wire because it is easier to bend.
"Metal 3D printing filament"
5kg spool of 0.6mm welding wire next to 1kg spool of 1.75mm Black PLA filament.
The last few days I was searching for the best material on which the first layer can be welded. I tried some sorts of fences, metal meshes and fly screens made of different materials and came to the conglusion, that galvanized steel fly screen could be the best bottom layer material. I also tried stainless steel mesh and I could weld the wire on it, but because of the higher resistance of stainless steel vs galvanized steel I wasn't able to weld more layers on top of each other. The greatest amount of heat was created in the stainless steel mesh instead between the wire layers. With aluminum fly screen I had the opposite problem because of the low resistance of aluminum almost no heat was created in the fly screen, only in the galvanized steel wire.
So I stayed with galvanized steel fly screen.
The buildplate is made of an 250*250*5mm aluminum plate with a M4*40mm countersunk screws in each corner and a copper plate taped on it with double sided tape. There is also the 25mm² cable with a cable lug and a M6*20mm screw and twelve 41mm paper clips to hold the fly screen on it.
The buildplate could be mounted like in the Hypercube Evolution 3D printer and move only in the Z direction. This printer design could also be good to create a closed enclosure, so that the fumes can be kept inside and the air could be filtered before returning into the room.
I have built a Microwave Oven Transformer Spot Welder from an old microwave tansformer, a jump start car cable and some parts I had laying around.
It works so far but the circut breaker tripped randomly all 5 welds or so and the connection
was not really stable.
Another test with different wire and one electrode clamped to the surface. This time the weld quality was better and the connection was very strong. But until now the wire does not welt to the surface.
I tested again with the first wire type and it worked like the other, it seems like the one electrode placed on the surface and the other placed at the material makes the difference.
Instead of welding the wire to a piece of sheet metal I tried to weld it to a piece of wire fence and it worked. The fence is also a lot cheaper than sheet metal and can be easily cut off after the object is finished. This fence has a 13*13mm mesh but I think for automatic welding it should be denser.
I actually wanted to build a capacitor based welder but I have read that if the capacitor gets shorted it will not last for long...
So I solved the circuit breaker tripping issue by connecting five 4 ohm 100W resistors in series with the transformer primary winding. Doing so it should limit the current to around 11.5A. I did a short test and after 10 welds the resistors get barely warm. They will be mounted on a heatsink for a long duration test to see whether they can work in continuous operation.
I let the machine run in 50ms welding, 500ms pause, 100ms welding, 1000ms pause intervals for about half an hour and the resistors stayed cold enough that I could touch them afterwards without burning my skin. They can operate up to 200 °C, so that should be good enough for continuous operation. After a while the 25mm² cable got warm, too. The transformer stayed cold the whole time.
The spot welder should be ready for working with it so the next thing could be working on the buildplate and bottom layer.