I'm an electronics newbie trying to learn about the subject through tinkering, and my current project is a capacitive discharge welder for welding on battery terminals. I thought I'd better have someone take a look at my schematic before I fry a bunch of expensive components.


Basically, I have a foot pedal which tells an Arduino when to (briefly) turn on an H-bridge made of beefy n-channel MOSFETs hooked up across a big 33mF 100V capacitor that I plan to charge to 30V. The alternating current produced by the H-bridge energizes a transformer which greatly steps down the voltage and steps up the current for welding through some copper electrodes.

The things I'm most concerned about are:

  1. I have a separate 5V supply for the Arduino because I have a vague sense that it would be a bad idea for it to share a common ground with the big power capacitor. I'm not sure how electrically stable an electrolytic capacitor is when you dump all of its energy all at once. Is this a valid concern? Will separating the Arduino from the rest of the circuit with the optocouplers actually help anything?
  2. Related to the previous question, I'm concerned about having to tie GNDA directly to the negative terminal of the capacitor (-BATT). Is there a better way to do this? Could my current design result in damage to the rather expensive MOSFET driver?
  3. Are there any flyback diodes or bypass/decoupling capacitors or anything that I should add? I have no sense at all for where such things would be advisable. I added the 0.1 microfarad capacitor between +5VA and GNDA because I saw it on the datasheet for one of the ICs, but I don't know if it's really necessary. Is there anything I can do to better protect the $20 (each!) MOSFETs I'm using?
  • \$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. Any conclusions reached should be edited back into the question and/or any answer(s). \$\endgroup\$ – Dave Tweed Apr 13 at 14:37

As shown with just a schematic, No this is not enough.

There is a bit of Physics involved in the fusion of dissimilar metals using a 3rd electrode metal (titanium?) on the surfaces that will not melt during the welded of the wire.

Let me try to explain this without a lot of theory and formulae using analogy of a small fuse ( and why your task is more challenging)

We know the properties of a fuse from datasheets. The time to heat up the thin wire to liquid then separate is non-linear with A-s and A²-s of the thermal and electrical effects. The electrical resistance is expected to rise linearly with temperature as the power increases by I² until it opens.

The fuse always has a Positive Temperature Coefficient (PTC) for R vs Temp.

Fusing a copper wire to a stainless steel battery electrode wire surface also has an initial resistance from surface area and pressure but the resistance drops when the metals fuse together so you have PTC medium of metal with (hopefully) a lower resistance when completed. So you have a negative incremental resistance over a short time interval. So it behaves like an NTC resistance yet somewhat abrupt like an arc ( which also has an abrupt negative incremental resistance) but over a much smaller dynamic range of R before and after.

Designing the resistances to make this a constant R is tricky as the heat dissipated is always proportional to resistance Pd=I²R thru each metal.

This means your electrode must have very low resistance contact and also a very high melting temperature like Tungsten to clamp the copper to the stainless-steel so it does not fuse to the surface so it does not stick to the surface after the pulse current is discharged.

You also have to consider the resistance of your coils to be less than the fused resistance of your wire-bond for efficiency. This means much thicker coil wires than your wirebond.

What if we apply the fuse characteristics of I vs t but reduce the energy level so that the interface bonds instead of fusing open circuit?

However, the fusing formula is not a simple function of current or energy (A²s) or constant Amp-seconds. enter image description here

  • But, what you want is the fuse the interface of the wires and SS electrode and be mechanically strong.

This is why I said it is complicated to find the right electrical energy for diffusion bonding thermal energy for an electrical time interval with an impulse of some shape for a piece of copper wire to SS foil battery terminal..

So what will work?

( from others on the web)

  • Buy a spot welder or DIY with the power of a car battery instead a cap bank.
    • make a one shot flipflip with relays, one being a 1kA car starter solenoid
  • adjust the contact pressure to control the inital interface surface resistance which is essential to start the heat rise process.

    • fusion interface must be higher R than electrode and battery ESR resistance)

    enter image description here ref

  • \$\begingroup\$ Thank you for your detailed answer. What you were saying about designing the circuit for the specific current profile of the melting process makes a lot more sense now. I guess the only real remaining unknown is to what extent this modeling is in the domain of expensive industrial welders, and to what extent a naive H-bridge circuit like mine will basically work in practice for small welds given some amount of software tweaking. \$\endgroup\$ – Brian Gordon Apr 14 at 8:31
  • \$\begingroup\$ For example, I've already had some success just discharging a supercapacitor directly (through an SCR), using copper probes, and got it working pretty well by sharpening the probes down to a very small contact area. I didn't notice any issues with sticking. My motivation behind the more complex design was that the welds were physically a bit weak and I was struggling to deliver any more current given the low voltage rating of the supercap and the resistance of the wires. So I thought a capacitor bank and a transformer would help me dial that in to increase the strength of the welds slightly. \$\endgroup\$ – Brian Gordon Apr 14 at 8:35
  • \$\begingroup\$ Yes that's not a very deep joint. Did you compute any discharge times and energy? Remember its some exponential product of these variables that reaches the temperature and time needed to spread for a strong bond. Did optimize pressure? You can use a 1 mOhm MOSFET Pulse direct off a Li-Ion Battery with low ESR cap too then control the gate time. ( maybe using the battery itself that U are bonding) ha \$\endgroup\$ – Sunnyskyguy EE75 Apr 14 at 9:58
  • \$\begingroup\$ .. avoid inductance due to flyback , unless clamped with similar low Ron \$\endgroup\$ – Sunnyskyguy EE75 Apr 14 at 10:03
  • \$\begingroup\$ No, I didn't record anything. I was just using my hands to pull away the probes to stop the welding. Could I prevail on you to elaborate on inductance/flyback/clamping? I don't know how to manage that, especially in an AC circuit. \$\endgroup\$ – Brian Gordon Apr 17 at 0:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.