Microwave oven spot welder

Microwave oven transformer spot welder. Later I have replaced the welding pulse activation switch with a foot switch.

I made a spot welder out of a microwave oven transformer and a starter cable. Those are quite popular actually. They use a solid state relay to switch the transformer on and off. I use a Fotek SSR-40DA DC-controlled 40 A AC-zero-crossing-switching solid state relay. It was only 4 € and free shipping on eBay. First I needed to extract the transformer from a microwave oven and to replace the secondary coil with a starter cable.

Microwave oven (was about 1050 W consumed power) gutted. In addition to the transformer (left), some other goodies were found from the inside that I saved for later use.


The secondary coil sawed off. There was also another secondary coil (red) that I removed later. After sawing from one side, the secondary coil could be hammered out using a chisel.

Some initial trials using a 700 W rated 12 V automotive starter cable. Didn't produce satisfactory welds. The cable was the most expensive part in the project, about 20 €. There wasn't that much copper in the cable originally. Most of the thickness came from the rubbery insulation. Perhaps I should have been more patient to obtain the copper from somewhere else, as I ended up stripping the cable anyhow to double the conductive copper area. Minimization of resistive losses in the secondary is key.

After making my own cable for the secondary, with twice as much copper, I was still able to fit in a few more rounds. This increased the voltage and reduced resistive losses further.

At this point I got a nice 4.6 V AC out of the secondary.

One end of the secondary cable is routed "under the floor", where it will be grounded to the body of the welder, and then goes to the immobile jaw. The upper jaw is mobile. To reduce wear of the secondary cable here, it is shaped like a spiral. A transparent plastic block electrically isolates the upper jaw from the body.

In the bottom, the cable is crimped inside a piece of copper pipe, hammered flat.

Another view of the jaws. I wish I'd found perfectly fitting copper pipes for the sockets. Now they are a bit loose. The copper cable goes pretty much all the way to the end of the jaws to minimize resistance.

A sturdy 230 V AC fan will take care of cooling. It pushes air out through the front grid at great speed, so that sparks will have less chance of flying into the welder through the grid.

I wanted to automate the welding timing so I made this simple adjustable-duration pulse generator circuit out of some components that I had around:


Welding pulse generator circuit.

Spot welder timer circuit

CD4093BC pinout in the same orientation (C) Fairchild Semiconductor

Parts list:

  • Resistor 1 kohm (R4)
  • Resistor 9 kohm (R1)
  • Resistor 40 kohm (R2)
  • Multi-turn or log taper pot 10 kohm (VR1)
  • 3x Capacitor 10 V 10 uF (you can put the extra capacitor in parallel with C2 to make the pulse longer if needed) (C1, C2)
  • CD4093BC Quad two-input NAND Schmitt trigger logic gate

The variable resistor VR1 controls the timing. I used a helical potentiometer but a log pot would probably work just as well. The higher resolution is needed for long welding pulse times for which the whisker of VR1 is close to R2. R1 and R2 adjust the range of the voltage divider. Capacitor C2 is also charged through R1. The voltage divider range should be such that at one extreme, continuous welding is just barely enabled (by keeping the button pressed), and the other extreme just barely gives the shortest welding time that may be needed. If the obtainable shortest welding times are unnecessarily long, the capacitor C2 can be made larger. This will make it easier to get also very long pulse times. Doubling the capacitance should double the timing.

A rough approximation on how the voltage across C2 rises after the button is depressed at time $t = 0$ is given by:

$V_{\text{C2}}= \frac{5\text{V}\cdot t\cdot(R_{\text{R2}} + R_{\text{VR1}})}{C_{\text{C2}}\cdot R_{\text{R1}}\cdot(R_{\text{R2}} + R_{\text{VR1}}) + t\cdot(R_{\text{R1}} + R_{\text{R2}} + R_{\text{VR1}})}$

where $R$ is resistance and $C$ is capacitance across a component. $R_{\text{VR1}}$ is the nominal resistance of the variable resistor, not the current set value. The first half of the variable resistor VR1 and the second half of the variable resistor VR1 + resistor R2 act as a voltage divider that feeds a divided $V_{\text{C2}}$ to a Schmitt trigger which will trigger at nominally 3.3 V, ending the pulse.


Calculated pulse duration vs. potentiometer position. Drain by the IC, and filtering by C1 is not taken into account. In practice the longest duration achievable is about 2 s. Turning the pot further makes the pulse last indefinitely, or until the button is released. The shortest time is in reality also shorter than the 0.27 s shown. The function plotted is y = 2970/(x + 1060).

A logarithmic pot would give almost linear control of the pulse duration.

An initial version of the circuit took about a second to recover before it was able to produce another pulse after releasing the button. Resistor R4 was added to remedy this. It quickly drains the capacitors empty when the button disconnects the circuit from the power supply.

The circuit is fail-safe in the sense that it only gets power through the button. If the button is released, the output soon goes down, even if the IC or anything else in the circuit malfunctions for some reason. I use this circuit to control (the LED + series resistor of) an optically isolating solid state relay directly.

Noise can be a problem at longer pulse durations, where a sudden spike may end the welding pulse prematurely.

The timer is not synchronized to AC, so timing will also vary if the solid state relay is of the zero-crossing switching type.

Cabling added. As an afterthought, I might have used more grounding cable to ground all enclosure surfaces...

Solid state relay (right) and a 5 V power supply added

Enclosure almost fully assembled, only the top is missing. I put in rather hefty fuses into the 230 V AC input connector. The connector at the bottom right corner is for the foot switch.


Showing what a welding pulse does to a bicycle spoke.

Action! Just for the sparks.

Welding results are mediocre at best. But strong enough for some purposes. One problem is that when the workpiece starts to melt, the current increases dramatically, causing a small "explosion". I am not sure how this could be improved upon to give a steady current.