When you set out to design your Capacitive Discharge welder you might be interested to know just how much current and energy you can effectively deliver to a weld junction.
A recent question from a reader (thanks Rob) has caused me to rethink both my previous measurements of weld current and estimates of delivered energy (such as those shown in Figure 13 on the Continued Development page of this site). I have concluded that while my measurements are accurate, I have not adequately appreciated what I was actually measuring and therefore my current calculations were wrong.
Hence I have revisited the issue of weld current on the basis of a simple lumped component LCR model from first principles. I have rigourously checked the mathematics and compared the model with both actual welder measurements and distributed component computer modeling. I am confident that it will provide you with a reasonable upper limit estimate of your welder design’s performance before you start melting solder.
Figure 1. Simple LCR Model of Welder Discharge Path
The mathematical derivation is rather tedious because I have deliberately formulated the solution in the time domain with sufficient detail so that you can work through it. Please download it by clicking on the following link (and try and find the obligatory deliberate mistake).
I have implemented the model in a Microsoft Excel spreadsheet that requires no mathematics whatsoever from you and produces very pretty graphs of discharge current, capacitor voltage, delivered load energy, etc. The spreadsheet calculates everything explicitly with no iterative solutions or macros to avoid security alerts from your browser security. The calculations are based on your welders design parameters and should predict just about everything you could ever want to know about your design’s performance. I encourage you to download and experiment with this.
The model has provided a number of insights into several aspects of my welder’s performance:
 With very low resistance weld junctions my welder is operating with an overdamped response, but close to critical damping.
 My previous weld current calculations were not correct but the reported results are still in the ball park (for example my previously calculated 9,300 A discharge was closer to 8,700 A).
 The peak discharge current I can reasonably expect to achieve with my welder design is about 10,000A.
 The peak current occurs within a few ms of pulse initiation. Consequently I need to increase the resolution for setting the duration of my initial pulse. This is supposed to burn off impurities but I suspect that it is almost certainly forming an initial weld.
Through playing with the model I have also come to the conclusion that any real CD Welder operating from 20 V with usable weld cables and three 1 Farad capacitors will not to deliver more than 12,000 A peak. If you need more current then you will need a higher capacitor voltage. Have a play with the model and see if you reach the same conclusion.
If you are building an SCR based welder you can still used the model to estimate performance although the forward voltage drop of your particular SCR will be based on the high current linear resistance approximation for your device (typically 5.7 milli Ohms for a 1,000 A SCR) as opposed to a true diode characteristic. Enter 1 MOSFET in the spread sheet with an RDSon of about 5.7 milli Ohms. Remember to extend the pulse width until the current drops below the holding current because your SCR will stay on until the current drops below this threshold.
Click here to navigate to the next welder page, Dusted Off.
