An update on my TIG inverter project

IMPORTANT, I am not trolling, and if I am asking stupid questions, that's due to my ignorance and malice.
A while ago I asked for suggestions regarding making a square wave
inverter to convert a DC TIG welder into an AC TIG welder.
Many things happened since that time.
1. I bought a real DC TIG welder for $9.99. See it here.
Auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&itemV1560149
It is a 3 phase welder that I run off my homemade phase converter.
My page with more pictures and my experience so far
http://igor.chudov.com/projects/Welding/00-Hobart-CyberTig-Welder/ http://igor.chudov.com/projects/Welding/
2. Spent many times its cost of $9.99 on cabling and various welding doodads and consumables.
3. I played with arc welding, trying to learn to weld (see above links).
Now, I am a little closer to the aforementioned inverter project. I bought four Toshiba IGBT, mounted on a heatsink:
Auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&itemu50212847
Data sheet: http://www.galco.com/TechDoc/TOSJ/MG200Q2YS40_DAT.pdf
My plan is to follow what this guy did:
http://www3.telus.net/public/a5a26316/TIG_Welder.html
and to use two IR 2011 gate driver chips, appropriate caps and resistors, and to use a Wavetek 171 to drive the logic inputs of these chips. Wavetek 171 can make square waves of arbitrary frequency, set digitally, and adjust pulse width.
Note that each IGBT is a complete half bridge, so I need just two, out of the four that I bought. My welder is a 200 A constant current welder, so it would not exceed the 200 A limit of the IGBT. I set welding current digitally, using a digital potentiometer on the control panel of the welder. Thusly, I could, supposedly, get away with using just two IGBTs, one for each half of the H bridge.
The welder's welding voltage is 28V and OCV is 85V. The IGBTs are rated for 1,200 V.
Since wavetek 171 can put out pulses of adjustable width and frequency, I can use it as the control of this inverter. Wavetek's output will be the input of the IR2011 chips.
For power, I will try to use a computer power supply.
I read various relevant application notes and schematics, by now.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 04 Oct 2005 16:53:57 GMT, Ignoramus29341

That should read "due to my ignorance and NOT malice".
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ignoramus29341 wrote:

Does the welder have HF? If so, that is going to fry your IGBTs. Even if it has a stabilizing inductor, that will probably produce large voltage transients whenever the arc current decreases or sputters, and they will exceed the transistor's limits.
Jon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I think I noted another pitfall. Ignor mentioned a 200-amp max current rating on his transistors, then mentioned that the welder's current could not exceed 200 amps. It's not usual practice to run power devices right up against their max ratings. You should de-rate at least 30%. Over 50% is common.
LLoyd
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 04 Oct 2005 19:15:44 GMT, Lloyd E. Sponenburgh

Another good point. Since I have 4 IGBTs, and each is a half bridge, I could parallel them. No big deal. Based on my tiny experience with welding, 200 A is a lot of amps and most of the time, I would use less than 200 A.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I may be off the mark here. I've done very little with IGBTs... mostly I've worked with conventional junction-type bipolar transistors and SMARTFETs... but here goes, based upon my power-driver experience with junction types ---
Unless the devices are specifically purchased as "gain matched", I think you need to go a bit further. Install a TINY equal amount of resistance in each emitter leg as an equalization feedback element, so one transistor doesn't try to take all the load. If their gains are not precisely the same, one will assume all the load until it goes into thermal distress.
The equalization resistor is small in resistance, but large in current capacity (as much as your max, plus a factor). I don't know what the drop across your transistors will be at 200 amps, but if you were to drop an extra half-to-one volt across the emitter resistor, it would probably work out just fine. It'd take some number crunching to get it right. One volt at 200amp == 0.005 Ohms.
That's a small resistance, but significant to the purpose at hand. It might be that a couple of equal-length battery cables (y'know... eye at each end) would serve at these currents.
Now, if these are "smart transistors", ignore everything. I LOVE those things!
LLoyd
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

I built a DC rectifier for my 225A Miller stick welder. I used 12 35A bridges in parallel. To compensate for device variations I used about 14" of #18 wire (to the best of my memory) for each of the leads to the bridge. This was equivalent to 0.01 ohms so each bridge had the equivalent of 0.04 ohms in series (4 leads) which at 20A was 0.8V drop. For my normal welding range, around 120A, the drop is 0.4V.
I also put 0.01uf caps across each diode in each bridge and a 120V varistor across the +/- leads of each bridge. I think this was overkill but you can't argue with success.
Another thing is that you have to consider the maximum current the welder can but out on a dead short (when you stick the rod down). For my 225A Miller that can be as much as 325A! You need to check the VA curve for the welder. At lower output current settings the short-circuit current will be lower.
Billh
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Lloyd, these are not smart transistors and indeed, if I parallel them, I would need to put in some smarts to prevent parasitic oscillations and unbalanced conduction. Some info on this can be found at
http://www.fujielectric.co.jp/eng/fdt/scd/pdf/Manual/IGBT-08.pdf
That's a great guide to making power inverters with IGBTs.
Also, note that at 200A current, (without paralleling) each my IGBT produced whopping 600 watts of power! That means 1,200 watts on the heatsink from 2 IGBTs. A very substantial amount. Having four IGBTs instead of 2 would not change the equation very much. It would be 4 times about 300 watts.

Hm, does that mean that each wire produced about 48 watts of power?

Sounds great!

Mine is a CC welder. I did experiment by shorting it. (will try again). It basically continues to produce required amps, does not go over all that much. Remember, mine is a bad ass, top of the line super expensive welder from yesteryear. It is not a home improvement store item.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ignoramus29341 wrote:

It might be a good time to explore the concepts of "Safe Operating Area".
Kevin Gallimore
-
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
<snip>

No, at 20A per bridge the power is ) 20*20*0.01=4W per wire.

I wouldn't exactly call a Miller welder with a copper transformer a piece of junk even if it is 30 yrs old. I checked the manual and on the high current range it can produce around 285A not the 325 I thought. On the low current range it is tighter.
billh
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

of
varistor
can't
the
be
If anyone wants to make a DC out of AC welder, I've got a rectifier assembly out of a Lincoln DC 400 for sale cheap. Brand new in the box. I got it at an auction with some other stuff and have no use for it. It'll go on Ebay someday if it gathers much more dust. Tom
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
How cheap? I might be interested ...
Andy

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
@NOSPAM.29341.invalid says...

The only common use for AC TIG is welding aluminum, for which you will almost certainly want HF and high currents. 200A will limit you to about 3/16" thick aluminum.
Ned Simmons
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Thanks for your note on AC TIG for aluminum.
PARALLELING
I have read the paralleling section of the IGBT textbook by Fuji. Then I checked the datasheet for my Toshiba IGBT.
Basically, I think that my Toshiba IGBTs have their collector/emitter voltage increase for high currents as temperature rises to 125C. That means that they would be more resistive than the ones that conduct less or are cooler.
So, while at lower amperages (say 100A) it could happen that one IGBT does the bulk of work, for higher amperages and temps that would not be the case.
(note that in this case, higher amperages mean close to 200 AMP).
Since they share the common heatsink, they will have their temperatures not too far from one another.
Another paralleling issue is avoiding oscillations between parallel gates, which can be done with simple resistors.
SAFE OPERATING AREA
All I read about their safe operating area (in the datasheet), in the context of a CC welder that is rated at 200 amps and does not exceed 200 amps, and for 200 amp IGBTs, suggests that a single IGBT working at 200 amps would be well within this safe operating area. Hence, I am not sure if parallleling is really needed here.
For instance, if gate to emitter voltage is 16 volts, then collector emitter voltage is about 3 volts. 3 volts and 200 amps are well within the safe operating area, according to the last page of the datasheet. I posted a link to it in my first post in this thread, so you can verify whether I am correct.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ignoramus29341 wrote:

You have to add collector equalizing resistors to give a voltage roughly equal to the expected voltage drop across the transistors, because IGBTs do not share current well at all. This is due to their positive temperature coefficient. Also, you can't use emitter ballasting, as is commonly done with bipolar power transistors, as getting an IGBT even near the linear region will cause it to self destruct. They must be hard saturated when on, or the hottest part of the transistor takes all the current, and burns up. You also need to supply enough gate current to turn it on and off VERY quickly, or it will pass through the linear region when switching, same result.
Jon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Jon Elson wrote: This is due to their positive

Umm.. The Vce goes _down_ with increasing temperature. Thus causing current hogging and thermal runaway.
http://www.st.com/stonline/products/literature/ds/6207/stgb10nb37lz.pdf
Kevin Gallimore
-
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

That's 100% right. I have not yet started reading the chapter on parallelization, so I cannot make any meaningful comment.
As for driving the gates hard, that's why I mentioned using IR2110 chips. That's what they are for.
The IGBT's also need "snubber circuits" to protect them from effects of rapidly lowering current (and increasing voltage due to main circuit inductance), which in my case (chopper) means a capacitor and a resistor between collector and emitter.
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Good point. I can turn off HF if necessary, and use a little copper plate to strike the arc (technique described in some welding books, I know relatively little about tig welding).
I could also, possibly, rewire the welder so that HF is added after the voltage is inverted. I would rather not do it though.

The IGBT is rated for 1,200 volts (collector/emitter). Do you think that transient voltages would exceed that?
thanks
i
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Ignoramus29341 wrote:

Have you ever SEEN the HF on a big TIG welder? I have a Lincoln Square Wave TIG 300, and the HF on that thing looks like those plasma globes that used to be popular 15 years ago! And, this is at a FULL atmosphere of pressure. I have no idea how much voltage it is putting out, the Argon seems to ionize a lot easier than air, but I can get 1 to 1.5" sparks from the electrode tip to the work after the argon has completely covered the area! That sounds to me like WAY over 1200 V!
The HF generator on most older, large TIG machines is basically equivalent to the high school project Tesla coils, and probably develops 10 - 25,000 volts when the main arc is not shorting it out.
Jon
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Take a look at app note AN1045 - on International Rectifiers web page - www.irf.com Uses IGBT's in a TIG design - with clamps.
Martin Martin Eastburn @ home at Lions' Lair with our computer lionslair at consolidated dot net NRA LOH, NRA Life NRA Second Amendment Task Force Charter Founder
Jon Elson wrote:

-
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.