Soft Discharge on Welder Capacitor Bank.

[Ploddy]

up vote
0
down vote
favorite
   

I live in Australia where we have a 240 V, 50 Hz supply. I have upgraded a (nasty, cheap, low quality) transformer based AC output welder into an AC and DC welder.

It works great.

I used a set of heavily under driven bridge rectifiers and a bank of capacitors of ~0.01244 F total capacity.

I also have a 5 x 2 megaohm 1/4 W bleeder resistors in parallel, to discharge the bank with, simply for safety - there is a fair amount of power in the capacitors, so self discharge is both good design and mandatory.

I think the output voltage is around 45 V DC.

This set up works really well when welding. However, when the electrode shorts out on the work there is a huge fat discharge from the capacitors, which to me does not seem to be the best practice - so what what resistor type and value would be the best to soften the absolute "dead short" discharge?

Since I am working on this unit tonight, I do have a bunch of low impedance wire wound resistors 1 R () that I could simply wire up in series on the Negative line. While I do not "totally understand all the maths and electronics, is the bank were to be charged up to their maximum OCV, and then discharged as a dead short through the resistance to say 50% of their voltage in 1 second or so, that would be acceptable.

Perhaps....

What do other people think.

[Ian.M]

Schematic?   

It probably needs a beefy choke in series with the output (something like a MOT core with the welds cut, regapped with thin card and superglue and wound with high current cable)  to decrease the rate of rise of the current if the electrode shorts. It will also help stabilise the arc.  You'll also need a pulse rated diode that can carry the peak current for one half-cycle of the LC resonance, cathode positive, across the capacitor bank to prevent resonant reversal killing them.   Adding any sort of fixed resistor would compromise welding performance.

Your discharge resistors are a couple of orders of magnitude too high resistance. 500K (4x || 2Meg) with 12.5mF takes about 30 seconds to fully discharge - that's long enough to get you in trouble.   Reducing the resistors to discharge the capacitor bank in under a second is advisable.

[woodchips]

On my ancient transformer based DC welder there are no output caps, but there is a large DC choke. At 450A it is about the size of a shoe box. Apparently the core material has an abrupt hysteresis curve with minimal area so made from either silicon steel or nickel steel.

Off load voltage should be about 70V or 90V to get it to strike, on load more like 30V.

[Ploddy]

Hmmmmmm Yah... I don't have enough brains or time left to "do a super good job" of this.

The capacitor bank is there to bump up the voltage as the zero volts point is approaching, thus stabilising and leveling the DC., thus giving a better DC than plain old FULL WAVE DC.

What I want to work on is to simply make the "dead shorts" into fast discharges, and resistors are fine for that.

I could suck it and see - what works, what catches fire, what hardly works at all... or I could ask.

[Ploddy]

https://www.allaboutcircuits.com/textbook/experiments/chpt-3/capacitor-charging-and-discharging/

 Build the “charging” circuit and measure voltage across the capacitor when the switch is closed. Notice how it increases slowly over time, rather than suddenly as would be the case with a resistor. You can “reset” the capacitor back to a voltage of zero by shorting across its terminals with a piece of wire.

The “time constant” (?) of a resistor-capacitor circuit is calculated by taking the circuit resistance and multiplying it by the circuit capacitance. For a 1 k? resistor and a 1000 µF capacitor, the time constant should be 1 second. This is the amount of time it takes for the capacitor voltage to increase approximately 63.2% from its present value to its final value: the voltage of the battery.

It is educational to plot the voltage of a charging capacitor over time on a sheet of graph paper, to see how the inverse exponential curve develops. In order to plot the action of this circuit, though, we must find a way of slowing it down. A one-second time constant doesn’t provide much time to take voltmeter readings!

We can increase this circuit’s time constant two different ways: changing the total circuit resistance, and/or changing the total circuit capacitance. Given a pair of identical resistors and a pair of identical capacitors, experiment with various series and parallel combinations to obtain the slowest charging action. You should already know by now how multiple resistors need to be connected to form a greater total resistance, but what about capacitors? This circuit will demonstrate to you how capacitance changes with series and parallel capacitor connections. Just be sure that you insert the capacitor(s) in the proper direction: with the ends labeled negative (-) electrically “closest” to the battery’s negative terminal!

The discharging circuit provides the same kind of changing capacitor voltage, except this time the voltage jumps to full battery voltage when the switch closes and slowly falls when the switch is opened. Experiment once again with different combinations of resistors and capacitors, making sure as always that the capacitor’s polarity is correct.

[Ploddy]

Schematic. Transformer. Bridge Rectifier. Capacitor bank. Leveled DC output. Polarised terminals. Welding leads

[T3sl4co1l]

Cross-link: https://electronics.stackexchange.com/questions/339470/slowing-the-capacitor-bank-discharge-in-a-dc-converter

[Ploddy]

Posting a link to the same issue on different sites is your idea of a solution?

[T3sl4co1l]

No, just providing it in case anyone accidentally reaches a conclusion already made.

So, an arc has a negative resistance, which has the effect of making an oscillator when you connect a bunch of capacitors across it.  As you are probably intuiting, shorting out a bunch of caps is just going to go SPARK SPARK SPARK SPARK, and not actually weld at all.

Also, 12mF isn't really enough for welding current (if a capacitive filter were suitable in the first place).

Fortunately, as woodchips noted, inductance is what you need.  Take another huge transformer (it has to be about the size of the power transformer), remove whatever windings are there, fill the winding space with as much welding cable as can fit, then add an air gap in the core (this is easiest if it's constructed like a microwave oven transformer, with a stack of 'I' shapes welded to a stack of 'E' shapes).  The gap should be about 3mm.

It'll still buzz a lot, but it should at least maintain an arc this way.

Good luck,
Tim