# Questions about capacitor banks

#### Walks91

Joined Apr 28, 2018
20
I'm an EE undergrad working on building a prototype for a magnetic pulse welder. The picture at the bottom shows the basic circuit design. The idea is to very quickly discharge a large amount of energy from a capacitor bank through a metal coil creating a strong magnetic field that essentially welds two pieces of metal together. For our specific application this requires a large amount of energy, roughly 15kJ. The frequency of the discharged current is very important as it affects the penetration of the magnetic field it should be around 10k Hz. The speed of discharge is also important, generally the pulse is discharged within 20-100ns. So my questions are,

1. If I use a AC power supply at 10k Hz should I expect 10k Hz as the frequency for the current through the coil?

2. I know the equation for Energy stored in a capacitor is E = (1/2)*C*(V^2) but the actual energy that will be discharged in the 20-100ns will be determined by the time constant t=1/RC right? So I think I shouldn't use the total energy for my current calculation I should only use the discharged energy.

3. I plan on charging the capacitors in parallel but discharging them in series would be ideal to get a higher voltage output without using a large capacitance value. I know there are capacitor banks that do this but I can't think of any simple solutions that don't cost thousands of dollars.

4. Would it be better to just use a transformer to pump up the power supply voltage and use high voltage rated capacitors (preferably around 2-5kV) rather than using many smaller voltage capacitors (around 350V)? My school has a 1kV AC/DC power supply but I'm not sure what the AC output is.

5. We intend on using an aluminum 'bitter coil' or disc coil as the primary coil to send the current through. Will the inductance of this disc coil be much of an issue. It will probably have about 10 turns each being about 0.5 cm thick. (They need to be beefy to withstand the magnetic pressure.)

I understand you guys don't like answering questions about high power applications but this is a project supervised by EE professors, I'm only asking here and not my professors because I want to make sure I have things right before presenting our plan. I really do appreciate any help!

#### Colin55

Joined Aug 27, 2015
519
You are talking about 10kHz for 100nS. That is absurd.

If you have a 10MHz frequency, one cycle will occupy 100nS.