Convert an AC to a DC Welder

Yes and Yes... but diode capacity must be calculated base on PEAK current, not the average current.

This might help explain as well. Each diode is carrying the entire current, not half the current.

Peter Wieck Wyncote, PA

The average current is less than the peak current, but when talking current ratings of semiconductors, it's the peak current that matters. 100A rectifier needs 100A diodes, because they still have to carry 100A, regardless of how much time they spend actually doing it. It isn't just a matter of heating.

FOR HALF THE TIME.

I have said nothing of peak current. A diode has an average rating, a peak rating, a voltage rating, etc. EVERYTHING must be selected as needed. There should also be a plot of current vs time allowed for peak current. I can't find that loking at a 1N4004. A 30 amp non repetative peak and 1 amp average, but average IS average.

greg

When you say you want a 3 amp bridge. You get a 3 amp bridge. Nothing was said about peak current. Very rarely do I worry about peak current. In most circuits the current is limited by other resistances or Z's. OK the 3 amp bride has a 200 amp peak.

greg

Look at some diode datasheets. When they say a one amp diode that is one amp AVERAGE current - or having one amp of DC flow 24/7. If you have a power supply that puts out one amp of DC at the output terminals - the diodes are seeing peaks of 6 X that depending on type of filtering - a capacitive input filter sucks a lot of current only when the diode voltage equals the charge voltage on the capacitor and then the cap looks more like a short circuit.

Now, how does that relate to welders? I don't know. My feeling is that the current output on the knob on the front is only a guide to give you some relative reference.

Realistically - My arc welder puts out about 50 volts on the highest setting (225 amps) The primary circuit is fused at the recommended 30 amps at 240 volts or 7,200 watts. 50 volts at 225 amps is more like

My feeling is a 4 diode bridge with 100 amp diodes will likely work well enough. Welding 1/8" steel I've never gone over the 120 amp setting and 75 is more normal.

Welders have a high leakage transformer in them, it works like a ballast for a discharge lamp. When you load down the output, the voltage drops dramatically and the current remains at more or less the output on the dial. In a nutshell, it's a constant-current source, not the constant-voltage source you get from the wall.

Then I guess the next question would be how sharp is the "knee" when it reaches current limiting? Its been a long time since I had the welder apart but the transformer looked like a pretty normal IE lamination but with more empty space than most transformers and taller than wider.

I don't remember seeing a magnetic shunt - but it has been awhile. I don't remember seeing a gap either. Primary on the bottom in one winding and secondary also wound close to the core, above it, with about an inch space between the two. I would think it would limit current some, but wouldn't be a very sharp limit.

It doesn't limit in the sense that it has a knee. The leakage inductance has a reactance at the line frequency that limits the current. Yes, all kinds of transient stuff happen when the arc strikes, but you get the idea.

There IS a different kind of KNEE where the core saturates. Depending on the construction of the inductor, this shouldn't happen, and would blow the line fuse if it did.

But when you start modifying the welder, you can no longer count on the safety features built into the original design. You kinda have to know what you're doing and have a fire extinguisher handy.

One more thing. The AC characteristics of the core demand that the current be symmetrical about zero. If you try to use a half-wave rectifier, you wind the core off the end of the B-H curve and (probably) blow the line fuse...i suspect there are many different core designs, but an AC welder engineer doesn't need to worry about core saturation with a half-wave rectifier 'cause he ain't using one.

When you calculate the $avings of making your own DC welder instead of buying one, you should consider that you'll probably blow up at least one set of diodes in your initial experiments. People who know how to do it right the first time, don't have to ask about half-wave rectifiers on a repair newsgroup. ;-) The oscilloscope I'd use to verify the diode voltage/current costs 50X the cost of the project. mike

Its been a long time since I had the

OK Greg, I see what you're saying, but I think that really, you are playing with words and math here, rather than exploring a concept that's relevant to speccing the diodes. Average diode current is not a good way to be choosing the device, because it relies on the limiting factor being power related, rather than a possible physical limitation of the construction. Any diode that's going to be correctly rated for this job, must be capable of handling the peak instantaneous current. How long it's passing that current for, followed by how long it's then got as a 'rest' period, is neither here nor there. So, if we were discussing a power rating, rather than a current rating, then your reasoning would have some validity.

If, in your christmas tree light example, the LEDs were specced for a max forward current of say 20mA, then by your reasoning, with an average current through the string of just 15mA, we should be OK. I would venture to suggest that this would not in fact be the case, as the peak current of 30mA through the string would over-run the diodes by some 50% at the peak of the cycle. If the reason for the 20mA max figure was one of power dissipation in the LED die, you might get away with a15mA average current figure, but if the limitation is on say the maximum current handling capability of the connections to the chip, then the likely outcome would be a blown LED. I think we're probably talking the difference between a guaranteed reliable setup, and a 'getting away with it' setup.

Arfa

Average current is what heats up the diode package. Peak current is what melts the silicon in the junction. You need to consider both. So quit arguing about it.

The welder diode specs don't care how many Xmas lights you have. There is no need for speculation...it's written in the spec.

Just look up the specs and design with both numbers in mind. mike

Isn't that what I just said ... ? It's not me that's arguing the point.

Arfa