I completely agree, I would never think of doing that.
Absolutely. There is a good reason why both torch and work terminals are insulated from the welding machine.
Someone in this newsgroup also made that mistake, a while ago.
Just to keep you posted of my progress, I have my timing circuit almost completely done. It makes nice adjustable square waves, with duty cycle adjustable 8%-92% or so, on two lines, onw is the logical negation of another.. I only need to finish and wrap up the 11 position capacitance selection switch.
Yes, all bridge rectifiers are full wave regardless of the number of phases. "Full wave bridge" is redundant.
Yes, unless you phase shift one side by 1/12 cycle, which would give you dodecal ripple, same as with a 12-phase rectifier.
I would be inclined say that a full wave rectifier draws current on the positive and negative polarity of the *rectifier input*, not the line input to its transformer. By this definition the 6-phase design is half wave, all of the transformer secondary windings see current in one direction only even though the primary windings see current in both directions. But this is inconsistent with the terminology used in single phase CT xfmr 2-diode config, so perhaps you are right here and both the 6-phase and 3-phase bridge rectifiers should be referred to as full wave.
The 6-phase configuration has 6 different phases present on the rectifier input, measured WRT the transformer secondary neutral. The
3-phase bridge has only 3 phases on the rectifier input measured WRT the transformer neutral, regardless of having the same ripple as the
6-phase rectifier. I happen to like this terminology, having accepted it a long time ago. You have some good points, but I think not good enough to change naming conventions which have been in use for almost
100 years, even if they are inconsistent.
Could be, I am not familiar with such really big installations, but in addition to the drawback of inefficient transformer utilization, the more phases the worse the power factor, and really big installations are charged extra by the utilities for low power factor.
I serviced a new 100 kW DC power supply a few years ago (failed during initial startup due to a cracked 1/4 watt resistor), and it used a
3-phase bridge with 3 SCRs and 3 diodes, which I think is now typical in medium size DC supplies.
AFIK they all do it this way. The "work" and "torch" terminal connections are interchangable with an internal switch, why would it matter which one of these "flails around" WRT earth ground? I think the parasitic capacitive load is small compared to the normal welding load, but these parasitic capicitances are probably part of the reason fast IGBTs are not recommended for this application.
I see. Are my Toshibas fast IGBTs? I did try to check that a few weeks ago, and my conclusion was that they were not fast, but I would love to hear clarification from someone knowledgeable.
I don't know that it does matter which *one*; it's just that they're both flailing in Igor's proposed arrangement. One more than is absolutely needful.
In an AC *only* welder, there would be no need for an internal switch since it's AC. If you interchange the "torch" and "work" with an internal switch, it's still the same old AC. So you could have a "work" terminal not driven with AC like the "torch" terminal. I'm not sure if this would be particularly advantageous or not. Of course, these comments would only apply to an AC *only* welder. I don't know if there is such a thing.
Do you mean the capacitive part of the normal welding load? The capacitance to ground of a workpiece (even sitting on a concrete floor) is probably comparable to the capacitance to ground through the power transformer of the welder.
But the internal parasitic capacitances are on one side of Igor's inverter, and the welding load is on the other side. Also, he is in the position of having semiconductor switches between two inductors, L1 internally and cable inductance externally. This is a very bad thing and therein lies a problem.
By the way, I was once concerned about the effect of even small inductances when multi-hundred amp ripple currents are present (this was battery current feeding a 4000 watt inverter from a 24 volt battery). So I measured the inductance of the battery cables; the number was 3 uH per 10 feet of 4/0 cable pair, with the cables tie-wrapped together. If they aren't tied together, it's more.
Page 8 of AN-1045 discusses the reasons; the main one seems to be the lower Vce(on) of the slower devices. Also, more favorable tempco.
Um, so can't you add a (commutating/snubber) capacitor across the transistors, to slow dI/dt of the overall circuit? You still get a nasty dI/dt right at the transistors, but at least with capacitors *right at* the transistors, you can keep that pulse short.
Or just switch the transistor slower. Put a 200 ohm resistor in series with the gate anyone maybe? I fail to see the difference between a slow-switching transistor and switching a fast transistor slowly...
Tim
-- Deep Fryer: a very philosophical monk. Website:
I think they are not all that fast, but they are still faster than the IGBTs that AN-1045 recommends. Comparing typical rise and fall times on your datasheet with the GA100TS60SQ that IR recommends, they turn on at about the same speed but the recommended IGBTs turn off 5 times slower for 1/5 the dI/dt to deal with.
Pardon my ignorance of the details of TIG welding, but why does the torch lead need to be thin?
The inductance of a relatively straight wire goes up as the wire gets thinner, and hanging wire in loops also increases the inductance a lot. I bet you have easily 10 uH and probably quite a bit more with the loops.
Thanks Glen. As I said in my response to another poster, I plan on having a snubber circuit and also on having a resistor to slow down shutdown of the gates.
Neither work nor torch terminals in my welder have a connection to equipment ground. I verified that.
My cables are: 40 feet of work lead 1 gauge, and 25 feet of TIG torch, perhaps 8 gauge or so. (it's conductor is water cooled and thusly can be made very thin). Not tied together. The unused part of work lead is hanging off a hanger that I welded together, in loops. (the neighbors prolly get a nice wave of HF from that loop)
You are fully right, but, I think, I should try to do with what I have.
Are you talking about a capacitor that you have added, or are you referring to something that was already in the welder? The schematic shows a 3 uF in series with 5 ohms, plus some smaller components. Is there another cap?
It is thin so that I could do very fine work with the tig torch. The way the torch is manipulated is very much like writing with a pen or pencil. Thin, flexible leads allow me to do it. Contrast it with stick welding, where I hold the stick holder like a baseball bat, with my whole palm.
thinner,
I see.
There is a capacitor in that loop between work and torch terminals, it ought to be doing its job.
I wouldn't worry about the 'flailing' of the work terminal with respect to earth ground, except make sure that the zero volts line (or you could call it 'ground' but that would be confusing) of all of the internal circuitry that connects to the welding circuit should float with the work terminal as should the shielding boxes around those bits of circuitry. The whole lot would then reside inside the outer casing of the welder, but it would be insulated from it. The outer casing of the welder would be connected to real earth ground for safety reasons. The only circuits which can be connected to real earth ground are ones which are isolated from the welding circuit by optoisolators, transformers etc. This is my suggestion.
While I plan on adding a snubber capacitor across DC input into IGBT (collectors), I was referring to the existing capacitor that was already in the welder.
Ripple with a 360 Hz fundamental. Once you have the ripple the circuit topology which produced no longer matters.
Presumably because of smaller conduction angles - more phases, less ripple amplitude, less conduction at lower voltages. Somewhere recently I saw a graph of the power factor vs. load for 3, 6 and 12 phase rectifiers but I can't seem to find it now. IIRC 6 was less than twice as bad as 3 and 12 was much less than twice as bad as 6; something like .9 for 3 phase, .84 for 6 and .8 for 12, at some unrecalled load, better at full load and worse at partial loads but retaining the same proportions.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.