TIG inverter project - new thread

The inductor is obviously L1 on the schematic, and I see the arc starter, too.

I see that the welder has SCR phase control of the welding current. Since the current can be controlled by the welder, your inverter need only chop up the DC into AC; you don't need to worry about PWM of the inverter to control the current.

I also see something on the schematic that makes me think the inductor must be saturating with currents typical of welding (for example, 200 amps). There is a network across the stick electrode and work terminals to carry the HF starting current. This network will also limit the voltage that will be produced by the inductor when the arc is interrupted. The main part of the network that will limit voltage consists of a 3 uF capacitor in series with a 5 ohm resistor; this series combination is directly across the arc in stick mode. The capacitor is rated at 230 VAC (I think that's what the schematic says; it's a little blurry), or 325 volts peak. If the inductor was 1.85 millihenries when carrying 200 amps, interrupting this current would result in the voltage across the 3 uF cap ringing up to about 3300 volts. In order to limit the voltage across the cap to 300 volts when interrupting 200 amps, the L1 inductance would have to be no more than about 35 uH when carrying 200 amps.

It would be good to make a measurement to determine is this is close to the truth. The only convenient way I can think of to do this is to operate the welder with, say, 200 amps, into a short if that's possible; if not, then have a 200 amp arc going. Look at the voltage across the inductor (L1) with the scope (get a photo with your digital camera) and also get a scope picture of the current in the inductor. The current can be measured across the shunt R4. The schematic shows some "output recorder jacks" that give access to the shunt terminals; the two labelled "BK" and "RD" (I think; it's a little blurry) are the ones you want. From these two waveforms we can calculate the inductance when L1 is carrying 200 amps.

Now that I've written the above, it occurs to me that it would probably be just as good to measure the AC voltage across the inductor when carrying

200 amps, and also measure the AC voltage across the shunt at the same time. These waveforms will contain the rectified 3 phase ripple, and from that we can probably infer the saturated inductance.

This is important to know, because it bears directly on the kind of snubbing you may have to do. If you have a suitable high voltage probe for your scope, you could have a look across the welder terminals when you interrupt the arc and see what the maximum voltage is. If you only have a

10x probe, you can put an auxiliary voltage divider across the welder terminals and then look at the output of that divider with the scope. For example, put a 10 megohm and 100 kohm resistor in series across the terminals, and look with your scope across the 100 k resistor. Don't use a little bitty resistor for the 10 megohm one; it will have to stand off the high voltage if there is any so use a 1 watt unit, or series up several smaller resistors to get to the needed 10 meghoms if that's all you have.

Variable frequency is no problem, but why do you want variable +/- pulse widths as percentages of period?

If you pull the torch (rod?) away from the work piece, when the arc dies, I think it will be a lot faster than .1 second. You could find out by monitoring the arc current (across the R4 shunt) with your scope.

Since there is no filter capacitor following the 3 phase SCR rectifier assembly, there will be 3 phase ripple riding on the DC. That's what we will be measuring. And, in fact, now that I think about it some more, the SCRs probably won't be full on, so scope pictures would be necessary to see what the conduction angle is, so we can accurately calculate the inductance.

As I said above, on further thought we may need at least one scope picture, the voltage across the inductor, to see what the conduction angle of the SCR array is.

No, don't do this; the DC component will saturate the clamp-on. Just measure the voltage across the shunt R4 at the "BK" and "RD" jacks.

Yes, I've seen that post, but that's a far cry from 200 amps.

you're

I looked at the spec sheet for your IGBT modules, and I see that they are rated at 1200V. If the maximum voltage the inductor can generate with the given RC network across the welder terminals is only about 300 volts, then you could use IGBTs with substantially lower voltage rating. The Toshiba's have a 3 volt drop at 200 amps, for 600 watts dissipation. A lower voltage, lower speed rated part could have a saturation voltage down around 1.3 volts as described in AN-1045.

I'm thinking that since you will be feeding your inverter through an inductor, you could use that fact to your advantage. You can modify your drive so that your IGBTs are used in overlapping conduction mode. That way, you will never be turning off an IGBT into an open circuit; you don't need to worry about getting dead time just right. And you might not need any snubbers other than the existing 3 uF, 5 ohm network across the welder terminals.

That guy (with whom I corresponded) built a fully homemade tig welder.

I already own a TIG welder, I bought it for $9.99:

I do not want to impair its existing functionality in any way and only want to add functionality.

It happens that it already has an inductor. It's probably needed to provide constant current. It also has an arc starter, which is placed between inductor and welding leads.

Well, in my case I have to deal with an inductor.

I am not sure what you mean here. Sorry. I have to provide alternating AC (+ and -) to weld aluminum and some other metals.

i

That's great. I am glad that these manuals are helpful.

Yes. That's a great schematic. They do not make things this way anymore, with schematic supplied. I love this old iron stuff that one can actually figure out.

Well, I do want variable frequency AS WELL AS variable +/- pulse widths as percentages of period.

I tend to agree with you, not knowing much.

Yep.

Yep.

I really, really like the way you think about it. Honestly. Note though that the above implies that interruptions of the welding arc are instantaneous. If you allow for these interruptions to occur within a perios od time (say, 0.1 sec for example), then the math no longer applies.

Still, you have a great approach of looking at limitations of the existing snubber circuit and deducing what the IGBT snubber circuit should be.

OK

I am confused, the inductor is carrying DC voltage, no?

OK, I see that now. I think that you have a great, novel, truth based approach to this that takes speculations and guesswork out of the equation.

Yes, there is AC ripple and if I can measure the AC component of the current (which I hope I can with my clamp on military surplus ammeter), as well as AC voltage across the inductor at 200A, I would know the actual inductance at 200A.

Please note that I did measure inductance at 12.5 amp AC by powering a room heater with the inductor in series (using household current with the welder off). It came to be 1.8 mH.

I do not, I sold one

I see. I really like the way you think about it. With some refinements, it may get me directly to the answer.

i

In TIG welding of Aluminum, percentage of electrode + vs. - are referred to, respectively, as "cleaning" vs. "penetration". + cleans aluminum of oxides, and - transfers heat to the welding puddle.

Don't worry about that, I already have a circuit with adjustable frequency and duty cycle, put together.

Yep... Need to brush myself on triggering. I am not sure how to capture it.

I see.

OK.

i

Yes. This is a semantic thing. It's like saying that the output of an ordinary full wave bridge rectifier has two phase ripple; I prefer to say that it has double frequency ripple. I've always thought it was strange (even though it's conventional) to refer to a "full wave" rectifier that derives its input power from a 3 phase grid connection as a "6 phase rectifier". I suppose there could be such a thing as a "half-wave" rectifier driven by 3 phase power; then the ripple might be called "3 phase". But we're talking high power here; if the power level is such as to call for 3 phase power, who would do such a thing?

In the case at hand the ripple will be sextuple frequency. By calling it "3 phase ripple", I was alluding to that fact that its ripple is coming from a rectifier powered by a 3 phase grid connection (sort of; I think Igor is using a "phase converter" he built, but it's still going to be sextuple frequency ripple, we hope).

you're

you

I agree. Until and unless you blow them up, you should certainly use the Toshibas you already have. Cross your fingers. :-)

be

getting

I didn't realize you were using that gate driver. You won't be able to do overlapping drive with it because it has shoot through prevention. Darn!

You should try to get a scope picture of the max voltage at the welder terminals when interrupting the arc. Set your scope (I think you said you have a 475; it has good triggering) so that the trigger is going to trigger at 100 to

200 volts with maybe 2 mS per division sweep speed; try both directions of trigger slope. Since it's not a storage scope, you'll have to eyeball it. :-)

I hope you can get some good measurements of the ripple and max voltage transient. I'll be looking for your results.

Minor point - this is a 6-phase rectifier, not 3-phase. Note the two Y secondaries; one of them is of opposite polarity to the other. The T-connected transformer in the return circuit to the two Y neutrals is an interphase transformer, which helps balance current between the two sides, since in normal operation two SCRs will be on at all times (one on each side) at all but light loads, where the phase angle is reduced below 60 degrees and only one at a time will be on.

you're

you

And can survive higher overvoltage for a relatively brief period of time. 2500 VAC for a minute.

Yes.

Well, I already have these Toshibas, and, in fact, they are nicely mounted on a heatsink. It was a pretty good deal, $50 for all four, with heatsink, plus $12 shipping.

That they have 600W power dissipation during 200 A welding does not bother me terribly too much, for these reasons:

1) I am rarely going to be welding with TIG at 200 amps. It just is not necessary most of the time. It's not like I do automated welding of long pipes or some such. 2) My actual use duty cycle (% f time when I am actually welding) will be quite small. I am just a guy working in a garage. 3) this welding machine has a very big cooling fan and great airflows. It could easily pick up 1.5 kW of extra heat and suck them out. All I need is provide a little fan for cooling the heatsink.

Remember, this inverter will be mounted inside the welding machine.

be

Yes, I will try to accomplish that. I think that with my IR22141SS gate driver, I can have resistors on the desaturation pins that could slow down turnoff. Not sure by how much. I just have not looked closely. I could survive with a brief open condition, with a cap across DC powe rails. i

you're

you

your

only

Thanks. I will keep my fingers crossed.

inductor,

never be

getting

At least I can reduce the circuit open interval to the minimum with the soft turn off.

to

Yes, I will try that with the 475. I have not used triggering much before.

I will try to do that indeed. Hopefully this weekend.

i

though

the

That's why I said "...we hope".

It definitely seems to be a semantic disagreement to me. See below.

As I said, at the power levels calling for 3-phase power, why would anyone ever use half wave rectification? I wouldn't call what is going on in the welder "half wave rectification". If it were, we would have triple frequency ripple rather than sextuple frequency ripple.

If you had a transformer energized from the wall socket in your house and it had a center tapped secondary with two rectifiers feeding a load, would you call that a 2-phase transformer? I wouldn't; maybe you would. Likewise, I wouldn't call what's in the welder a "6-phase transformer". It's a degenerate case and what you call it is a semantic issue. :-)

If you had a rectifier transformer energized by standard single phase on the primary, with a secondary which was center tapped feeding the load with two rectifiers from the ends of the center tapped winding, would you call that "half wave rectification"?

Two ways of achieving what I would call "full wave rectification" (in a single phase environment) are to use a center tapped winding with 2 diodes, or a non center tapped winding with 4 diodes. The trade off is less efficient use of the transformer with fewer diodes vs. the most efficient use of the transformer with more diodes (in this single phase case). A center tapped topology is what is in the welder, and I wouldn't call it a "half wave rectifier topology". (The presence of the interphase transformer doesn't materially affect the "center tapped" nature of this topology.) If it were half wave rectification, we would only need 3 SCRs and a single Y connected secondary and we would get DC in the windings and triple frequency ripple

One advantage I see for the topology in the welder is that the cathodes of the SCRs are all connected together, which means that floating trigger circuitry need not be provided.

The transformerless motor based type of phase converter will also introduce 60 Hz ripple due to unequal phase voltages. Well designed rotary converters always include a transformer to balance the phases.

But I cannot agree that the difference between a 3-phase full wave rectifier and a 6-phase half wave rectifier is just semantics even if they do have the same ripple frequency. Distinctly different circuit topologies have different names for good reason, IMO. And the term "3-phase ripple" does not distinguish between the ripple obtained with full wave or half wave rectification of 3-phase power.

I also initially took a look at the 6 SCRs and thought 3-phase bridge, before noticing the 6-phase transformer and half wave rectifier topology :-).

The term 2-phase power is normally reserved for the case of 90 degree phase difference, not 180. I try to use the common meaning of terminology as used by others, makes communications easier sometimes :-).

I see your point esp the comparison to the single phase centertapped full wave bridge, however I have never seen it referred to as other than a 6-phase rectifier (or, less frequently, a double-Y rectifier) in the literature. It is or was a very common configuration in high power rectifiers, and AFIK you are the first to call it a 3-phase full wave rectifier, a term usually reserved for another configuration lacking the common cathode advantage. It seem to me that there is a need to differentiate these topologies with different names.

There was some controversy over the "6-phase" name quite a while ago, you are not the first to complain that 3 of the phases are merely the inverse of the other 3, see ABSE for 6-Phase_Rect.PDF for a bit of history.

That would be the configuration where the three terminals from a single Wye or Delta are each connected to 2 diodes, an anode and cathode each, right? Then all the anodes on the other ends of the diodes are connected together and similarly the cathodes. This would be completly analogous to an ordinary single phase bridge. Doesn't this arrangement get called a "6-phase bridge"?

I think the way to characterize a "full wave" recitifier is that it draws pulses of current from both the positive and negative polarity of the grid voltage, be it single or 3-phase. Thus, the arrangement in the welder which has the similarity to a single phase center-tapped topology is a "full wave" rectifier in this sense, and so is the 6-phase bridge arrangement.

To get "half wave" action, you would have to use a single rectifying component with each terminal of a single Wye (instead of the double Wye in the welder). I don't think anybody would do that. So, maybe that's why nobody uses the term "full wave" in connection with these various rectifer circuits powered from a 3-phase grid connection. Because only full wave topologies are ever used at power levels that call for 3-phase grid power, it goes without saying. Makes sense to me. :-)

It's been a long time since I looked at 3-phase and after drawing some 3 and 6-phase sinewaves, I (consciously, once again; I knew it consciously years ago-it just faded a little) realized that in a 6-phase system 3 of the phases are, in fact, *just* the inverses of the other 3. The same is true of a 2-phase (90 degree system); expand it to 4 phases and two of them are just the inverses of the other two. So the 6 voltages applied to the anodes in the welder are a true 6-phase system, and it's fair to call it a

6-phase rectifier.

The principle degenerates for a single phase system in the sense that with a true 2 or 3 phase system, you can add appropriate quantities of the existing phases and get a new phase not present originally, but you can't do that with a single phase system.

And with respect to the ripple, if it's true that rectifiers powered from a 3-phase grid connection will always be "full wave", the ripple will always be sextuple frequency (360 Hz; unless it's really high powered and the designer has used multiple windings to get 12 or more phases). I knew that the ripple was sextuple frequency when I said "3-phase" ripple; I meant the kind of ripple that a rectifier powered from a 3-phase grid connection will produce, as opposed to that from a single phase system.

[Snip]

have

to

If you can't get scope pictures, I think a simple drawing of what you see on the scope will suffice. Just include the scale info.

What I've done when I didn't have a storage scope was to put some Saran wrap or similar plastic film over the face of the scope and then you can draw on it with a grease pencil or felt tip pen, on top of a waveform repeated if necessary to get a good tracing.

I find this somewhat confusing. This kind of bridge connection is a "full wave" rectifier in the sense of my paragraph just below (typo and all); would you agree? And this means that the ripple on the output is sextuple frequency, or 6-phase ripple. You notice in a few places I used the phrase "3-phase grid connection" so that I wouldn't be describing the power actually applied to the rectifiers. But with the bridge, even though only 3-phase is applied to the bridge, the ripple is 6-phase (so to speak). In fact, if you had 12 diodes and applied 6-phase power to two bridges (outputs in parallel) with a double wye connection, you would still only get sextuple frequency ripple, wouldn't you?

So do people just not do 12-phase rectification any more? Maybe only the

*really* big installations, such as an aluminum plant.

I had thought about this when I was reading some of Igor's postings earlier. Igor acknowledges in another posting that the torch and work terminals are both insulated from the welding machine. When he gets this inverter going and connects it to a workpiece which is itself grounded, the entire welding machine guts are going to be flailing around with a square wave of voltage at the inverter frequency and the capacitance to ground through the main power transformer, etc., will have to be driven by his inverter.

Do commercial AC welders do it this way, or do they avoid flailing the work terminal by using a different topology?

have

100 to

:-)

Thanks... Hopefully I will have time today for that...

i

Perhaps by some, but I have only seen it referred to as a 3-phase bridge.

Agreed. Perhaps worth noting that, like the single phase center-tapped transformer 2-diode full wave rectifier, the old "6-phase rectifier" circuit is essentially obsolete because it makes inefficient use of transformer iron and copper, and the cost of the high side drivers is less than the cost of the larger transformer for high power apps today. Perhaps this is why the 6-phase and 12-phase designs are not mentioned in recent EE textbooks.

-----------

On another note for i:

Do not connect the welding return, often referred to as the "ground" connection, to the safety ground or machine frame in the welder. When this is done and the welding "ground" connection falls off a workpiece which is connected to earth (safety) ground (as all welding workbenches should be) while welding, then full welding current will return to the welder through the safety ground wire, overheating it or melting it depending on welding current.

This means that your H-bridge should not be connected to ground in any way except when the welder return "ground clamp" is connected to a grounded workpiece. You need to survive the situation where the "hot" lead only is connected to earth ground.

This may seem obvious, but an engineer at one of the welder mfgrs told me it was a fairly common error. (When you are buying 100+ high end welding machines the engineering department answers all your questions and some they think you may have forgotten to ask.)