DIY AC TIG Welder using an e-bike controller?

I've been on a DIY how to build "absurdly expensive shop tool X, using off the shelf parts" kick lately, and while sourcing aluminum for a cnc router project, I realized that it would be much easier to build the router if I built an AC tig welder first. It then occurred to me that the PWMed output of a beefed up ebike controller would make for an ideal current throttling device in my welder. I've only recently begun the process of attempting to wrap my brain around this tech, so bare with me if for a moment if I say something that is completely off base, but my understanding of the way a controller works, is that it modulates the pulse width of a square wave so fast that it looks like a sign wave to the motor, and as this simulated sine wave passes through the windings of the motor an alternating magnetic field is created, which in turn (no pun intended) pushes against the magnets in the rotor so that they "slip" back at a speed that is slower than the rotating field, subsequently creating torque on the wheel. Now, I am told that because PWM is not a true sign wave, it is not suitable for welding aluminum, but my thinking is that if we can create an alternating magnetic field by passing the PWMed wave through the windings of a motor, then we should be able to create an alternating magnetic field by passing that same wave through the primary winding of a one to one transformer, and that as far as the tig gun is concerned, the alternating current induced in the secondary winding of the transformer should, effectively, be indistinguishable from any other source of AC.

At any rate, if you think about it, the throttle on an ebike does almost exactly what we want a welder to do, only backwards. With an ebike, you want high current at low voltage that will gradually ease into low current at high voltage, where as you want a welder to arc at high voltage and low current then ease into low voltage at high current. I'm guessing that this effect is controlled by varying the frequency of the simulated sine wave in response to a combination of input from the hall sensors and the throttle, so that by manually feeding the hall sensor input a signal, we should essentially be able to trick the controller into thinking that we are on the flat at speed when we want to establish an arc, and then convince it that we are trying to climb a hill when we want to weld. We' should be in the right voltage and amperage range though, since you usually want about 70v to establish an arc, and about 15 to 20v at 100 to 200 amps to do your weld. I imagine that some sort of active cooling for the fets would be necessary, and I'm not sure what frequency a tig welder needs to operate at it in order to penetrate the aluminum oxide layer, so its possible that we may needing fets that are faster in oder to simulate it, but if it can be done, we would have an inexpensive way to turn a charger and a controller into a portable light duty AC welding machine .

Indubitably said:

but my understanding of the way a controller works, is that it modulates the pulse width of a square wave so fast that it looks like a sign wave to the motor,

Click to expand...

Not exactly. The simple view is that the motor itself (or rather, the inductance of it's windings) creates a countering force to the current flow that makes it "look" to the motor as if it were a more or less constant flow thru it, rather than a bunch of little pulses of flow.

So you would need an equivalent inductance or capacitance or both to help smooth out that flow, if you want a constant current or voltage available at the output of the controller. At that point, you have what might as well be a regular switching power supply.

But IIUC, you are trying to get a sinewave out of it, not a constant level. To do that, you'd need a controller that can output sine--most ebike controllers use trapezoidal control for the commutation.

However, you are wanting a single phase of this, and you might not be able to modify an existing ebike controller to operate correctly as you need it to if only outputting a single phase. You'd actually want a brushed controller for that, whcih typically is only going to output standard square-wave PWM (because it isn't trying to commutate anything at all--the brushed motor does that on it's own, mechanically).


But a much simpler method was suggested to me by MattsAwesomeStuff: use a light dimmer "switch" capable of handling the current you need, and use actual wall AC voltage. Then you get the sine you're after, chopped into slices by the dimmer.

I haven't tried it yet, so I don't know how well it works myself, but there you go.

So are you saying that the back emf sort of fills in the gaps between the pulses, and that this more or less makes the wave look smooth? I suppose I can see how that would work, provided that the back EMF was out of phase by an integer multiple of pi, otherwise you'd more or less just get something that looked like resistance in the circuit, right? Basically, my thinking was that the magnetic field would look more or less like the simulated wave, so that the current in the seccondary would look like the simulated wave. The way I had envisioned it, this would be like shooting a machine gun at a target mounted to a flywheel normal to the plane of rotation (first from the front, and then from the back). The firing frequency would then be modulated in bursts so that the flywheel would feel more or less the same force profile of being pushed by a piston. The fly wheel would be rotating like any other flywheel though, and if you used it to drive a real piston, that piston would move like any other piston. The reasoning being that the power stored in the magnetic field would carry with it a kind of EMF momentum which would smooth out the power, the same way that the mass of the fly wheel would smooth out the power transfered to it by the machine gun bullets.

As for the light dimmer, I'm not sure that I follow what you mean by "chopping" up the sine wave. How does this increase my frequency? At any rate, it may not really matter whether it is specifically a sine wave or not, since the effect we want is simply the result of rapidly reversing the direction that the current flows. Basically, pushing the current from the metal to the electrode breaks off the oxide, but pushing it from the electrode to the metal carries the heat into the metal. Actually, from what I am told, square wave is sort of the defacto standard at this point. I am also told that the frequency range we are looking for is about 100 to 400hz, and my guess is that the reason pwm is not enough by itself is because you really need it to look like that particular frequency range in order to controll the "softness" of the arc so you can get the bead size you want.

I'll see what I can find on the dimmer though, if its cheaper and easier, I suppose you might just as well go with that. Do you know if the dimmer will also provide the variable constant current limiting effect we need?

I'm also wondering why the phase issue is a problem. Don't most controllers have an option to set the phase angle between each output? Shouldn't we be able to paralell the outputs if we set the phase angle to zero (or n*360 for that matter)?

Indubitably said:

I'm also wondering why the phase issue is a problem. Don't most controllers have an option to set the phase angle between each output? Shouldn't we be able to paralell the outputs if we set the phase angle to zero (or n*360 for that matter)?

Click to expand...

Not that I'm aware of. Some controllers have a jumper for 60 or 120 degrees, and that's about it that I know of, for anything inexpensive. Stuff like Sevcons or other expensive high-end controllers may well be able to change that to any increment desired, but remember that this is ONLY the hall-sensor placement angle, because of the number/spacing of magnets in the rotor. It has nothing to do with the actual angle of the output phases to each other, AFAIK. (I might be wrong)

The only sine-output controller I am aware of that could definitely be "easily" altered to do what you want is the one by Lebowski, simply because he is programming it from scratch, himself. So if you could convince (pay, probably) him to rewrite the code to allow this type of operation, you could build your hardware output stages to work like you want.

It might be possible to fool a controller into outputting on only one pair of phases, 180 degrees from each other, but I am not sure if this is practical, even if possible.

Indubitably said:

So are you saying that the back emf sort of fills in the gaps between the pulses, and that this more or less makes the wave look smooth?

Click to expand...

Sort of. As a magnetic field collapses in a coil, it induces current in the same direction that it was already flowing, so it keeps the current more contiguous. If the pulses are short enough and long enough delay between them, then it becomes more difficult to maintain flow between them, without much larger inductors.

So you can think of coils (inductors) as storing current. So these are wired in series with a circuit.

Capacitors do the same thing, but with an electric field instead of a magnetic one, so they store voltage. Thus they are wired in parallel with a circuit.

Both essentially do the same thing, of averaging the power flow they are connected to.

The reasoning being that the power stored in the magnetic field would carry with it a kind of EMF momentum which would smooth out the power, the same way that the mass of the fly wheel would smooth out the power transfered to it by the machine gun bullets.

Click to expand...

That's a passable analogy I suppose. I don't pretend to truly understand how this stuff all works, just that I can visualize some of it happening from my early electronics classes and the explanations by the guy that started me into ham radio before that.

You can also visualize it as a river flowing from a dam, and you are opening and closing the floodgates of the dam, for either shorter or longer times. That's the PWM duty cycle. The frequency is exactly that--how long between the tiems you open the gates. The dam always has about the same pressure (voltage) behind it, so the pulses are always about the same "height", just that they are not always as long.

Since the riverbanks are either narrow and high or wide and low, the pulses may stay as big waves down the river for a good distance, or they may fan out into much more level ripples--this is the equivalent of inductance or capacitance.

As for the light dimmer, I'm not sure that I follow what you mean by "chopping" up the sine wave. How does this increase my frequency?

Click to expand...

It doesn't. It only allows you to use PWM to control voltage (and ultimately, current, based on the resistance of your load). If you have to change the frequency, you'll need an active controller of some type to generate the sine signals at whatever desired frequency range you wish.

Keep in mind I don't know how TIGs work, haven't used one or had one apart either. It was just a suggestion someone else gave me regarding how I might start building one of my own. So the info you gave in your post I'm replying to is a lot more than I had tos tart wtih.

I'll see what I can find on the dimmer though, if its cheaper and easier, I suppose you might just as well go with that. Do you know if the dimmer will also provide the variable constant current limiting effect we need?

Click to expand...

That's what PWM (in anything) can do, by varying the voltage across the resistance of your load. You just need a circuit added to it that senses how much current is flowing, and feeds that info back to a circuit that alters the duty cycle of the PWM to control that current. It's what ebike controllers do, and chargers, etc.

I was thinking about the phase issue, and it occurred to me that induction should cause a time delay proportional to the amount of power stored in the field, which does seem to be the case after all, but apparently it is more complicated than that. To be honest, I don't really get what all is happening exactly, but it would seem that there is some play there in the effect inductance has on current and voltage, so that you need a circuit that accounts for both effects if you want to get single phase out of a poly phase input. It looks like something called an "all-pass" filter is used to accomplish this task in signal processing applications, so I imagine it should be possible to do something similar by scaling the components up to our power level. The "lattice" version in particular, seemed most appealing to me, as it appears to be little more than a simple network of capacitors and inductors.

I'm still not 100% on why exactly we need single phase, but it does seem that all the welders I can find produce single phase, so there must be some sort of undesirable complication at the arc involved with poly phase (perhaps the potential difference between the phases would cause interaction between the electrodes), or perhaps single phase simply has favorable weld characteristics. Welding is certainly a resistive load, and the general consensus seems to be that "resistive loads" don't care whether your power is poly phasic or not, but I'm guessing there is more to it than that.

Youv'e only got two wires, so you can't have more than one phase thru them at a time.

The reason you can't just hook up the three phases of an ebike controller to two wires is that you are going to hook up one wire to each phase, and then to use the leftover phase you must hook it to one of the others. If you do that directly, you short it out and then the controller blows up.

If you do it indirectly, then whatever you are using between them is going to dissipate the power, and it is wasted, not being used in the weld anyway. So there isn't much of a point to doing it at all.


Additionally, you then have to build something to fool the controller into thinking it is spinning a wheel, somethign that feeds fake hall signals back to it. Perhaps a modified ebike tester like the one Lyen sells might work, if you don't wnat to do it from scratch.

That would allow you to change the commutation frequency, but not the PWM frequency, which seems to be something you are wanting to be able to alter. For that, you'd have to change the MCU programming, and that means writing totally new code for the controller from scratch, as the original is not available (or readable from the MCU) to hack.

Well, I wasn't thinking you would feed them into the same two wires, I was thinking you'd have multiple electrodes. My thinking was that the problem with multiple electrodes would be that the power might want to arc between the electrodes rather than the weld, but then I suppose that produces more or less the same problem as you would get with feeding into the same wire. I am however guessing that this won't be problem if I can marry the outputs with an analog filter so that I get a single phase.

The pwm's switching frequency isn't really an issue provided that it is sufficiently fast, so that I can use the output with analog components like a transformer and get more or less the same response that I would get out of AC. The frequencies that I'd need to simulate are on the order of hundreds of hertz though, where as the pwm frequency is on the order of tens of thousands of hertz if I'm not mistaken, so I don't think that there should be any reason to fool with it. The one thing that I imagine programming the pwm might give me is the ability to adjust the weld "balance", which is to say how heavy the current is on the up swing as opposed to the down swing. With a completely balanced wave you get 50% pull and 50% push, and I think you want something closer to 30% pull and 70% push, because it keeps the tungsten cooler when pulling and helps to improve penetration when pushing.

At any rate, it does sound like it may be more worth my time and effort to see if I can't simply get my hands on a legit driver board for an AC tig welder, but even so, the technology is so closely related that it seems like you might as well do both into the same place, and welding aluminum is something that I'm guessing more than one ES member has contemplated getting into, so I'm going to try to keep the board updated on my progress.

http://www.ebay.com/itm/JASIC-Inverter-PCB-welder-Repair-WSE-200P-TIG200P-AC-DC-/170658597127

How does this look? It seems to be the primary inverter module from one of those cheap Chinese welding machines. They also sell a board with a bunch of pots and resisters all over it that I'm guessing is some sort of input interface.

I tig aluminum every day. You're going to waste a lot of money building a welder that won't work.
Even if it did work you'd still need a torch, argon, electrodes, filler metal, flow meter, helmet, ground clamp, cables, gloves, steel table, leather welding jacket (spatter is brutal when you dip accidentally).
Plus the machine needs preflow and postflow of argon which would be hard with a home made welder.
Plus you'll spend an entire $100+ tank argon just practicing to get good.
The AC frequency and AC balance needs to be adjusted. I use high frequency and mostly electrode negative.

iamsofunny said:

Plus you'll spend an entire $100+ tank argon just practicing to get good.

Click to expand...

Why not CO2 instead of argon? It is cheaper.

Because CO2 causes the weld area to explode and sends liquid black aluminum everwhere and blows holes in the material.
Welding aluminum is a fine process that goes like this>

The electrode carries mostly electrode negative (down) and some electrode positive (up). Clearly, more down produces heat into the workpiece rather than melting your electrode into a ball. The rapid a/c changes cause the the oxide to break up and flow outward. When the torch is held mostly vertical at the correct distance with correct argon flow (1 litre per minute per 1/16" of cup size, example 5 lmp for #5 cup) a puddle forms. The puddle is perfectly clear of impurities. It is liquid aluminum. To weld, you wait until this puddle forms by adjusting the foot pedal to set the heat. Too much heat and it will make a mess. Too little heat and your filler metal will blob into an oxidized mess.
Then, when you see the perfect puddle, you touch the edge of the puddle with the filler rod. This grows the puddle. Remember, this isn't oxy-acetylene where you stick the filler into the angled flame. Tig is vertical electrode and growing the puddle.
Heat control and strategy need to be exact. Small pieces heat up quick so you need to do the ends first then tie in otherwise it just gets so hot by the end that it doens't build up and just melts into a blob.
The AC balance adjusts how fine the puddle is. I use high frequency which causes a focused puddle that I move around as I please.

A home made welder will not have the hi-frequency (not hi voltage) start. You cannot really weld scratch start because it breaks your electrode tip off and a little bit of tungsten gets on the material causing a black mess. The black mess is your enemy as you will find out every time you accidentally dip the electrode into the material and have to regrind the tungsten and re sand it to a fine point.

In short, buy a welder, whether Everlast 185 Micro or a Longevity 250 AC DC etc. And buy an auto-darken helmet too. If money is a huge issue, don't do it because you're going to spend a lot of money on gear, metal, filler rod, argon and electrodes just to get started.

and I am under the assumption this thread is for aluminum only
CO2 works for steel

Ok.

Yeah, money is something of an issue, but then money is always going to be an issue, and I was figuring on consumables cutting into my budget either way. AC tig is where I would prefer to invest my practice time, because aluminum is the material I really want to work with, but perhaps a used oxy acetylene setup and brass brazed 4130 chromoly would be a better place to wet my feet. You just don't see used AC tig machines around here for less than $1000, (which is pretty tough to justify when I could get a brand new Chinese machine with a built in plasma cutter for the same price), but every now and again you do see an oxy torch and a couple of bottles with new regulators hit Craigslist for something like $300. I suppose that would be sort of like learning to play the piano when what you really want to so is play the electric guitar, but once I get good enough with the oxy that its making me more money than its costing, I can justify dropping the cash on a tig machine without worrying that it will wind up sitting on a shelf because I can't afford the argon.

Look up tinmantech videos on Youtube and his oxy acetylene welding video
He uses flux on a 5356 rod to "weld" (more like braze) the 5000 series aluminum. and I''m sure the 5356 rod works on 6061 too.
It's slow but effective.
The 4043 rod doesn't oxidize as easily, melts at lower temp, flows better and doesn't crack as easily when cooling though. The 5356 is stronger, slightly..
Don't buy a tig, it's too expensive.
Don't use alumiweld or any other zinc based solder.

Also for battery cases you can use 1/8 ply and 2 layers of thin fiberglass cloth
And you can weld steel pretty easily with a stick electrode at a lot of angle for cheap but obviously heavy material. Strong though.

I built fishing boats in the Caribbean for 3 years. We used 1/8" thick of fiberglass. I suspect you could build a frame out of 1/4" thick layers around wood and it would be strong. Not the best idea but fiberglass has its place for certain applications such as side panels, battery mounting and fastening torque arms permanenlty

Well, its not so much about building battery cases as wanting to learn how to do tig on aluminum. Granted, it would definitely prove usefull on my bike projects, and that is one of my major motivators for choosing aluminum tig in particular, but welding is a valuable skill in its own right. At first, it will just be a hobby, and like any hobby I am expecting it to eat into my bottom line, but if I'm smart about it, and practice every day, I figure I should eventually get to where my "hobby" makes me more money than it costs. If the practice was a chore, I probably wouldn't bother, but I dig working on this kind of stuff. After a day of cramming theory into your head for 8 hours straight sometimes you just want to grab a dremmel or something and start monotonously grinding out widgets untill you hit just the right shape. I'm not looking to create a masterpiece, or make a fortune, I'm just looking to make widget #47 a little bit prettier than widget #46.

PS:

Fibreglass is some pretty groovy stuff, and I've been toying with the idea of layering it with aluminum flashing to produce a very light yet sturdy fairing, in much the same way that they do with an airplane fusilage, but to design and build a decent carbon fibre frame is a lot more challenging than it is with aluminum. Aluminum might be a difficult and expensive material to master, but from a design standpoint it is actually fairly forgiving. Its light enough that when you inevitably screw up your design, you can pretty much just keep adding volume and or increasing diameter untill its strong enough to hold, and its rigid enough that it will probably more or less move how you expect it to move. Hell, at this point, I'd probably preffer to focus on frame construction even if I was just working with mild steel. I'll worry about building a cart with fancy materials and groovy excessories once I've actually got myself a good horse or two to put in front of it.

I have been building aluminum frames for about 4 months now. I don't really recommend trying unless you have lathe or small cnc or at least a drill press. But to make the head tube you'd need to either lathe it or drill it. The bottom bracket needs to be milled after welding because it goes oval so you can't buy a premade bb shell for that purpose. After welding it needs to be baked at 400F for 1 hour so the heat affected zone (HAZ) rehardens so you'd need a large enough oven. I didn't do this for my frame but I'm not selling them so it's not a problem.

I think you're in a similar position of when I was in college. I previously had done woodworking all day every day and then had to go sit in a classroom to be indoctrinated with radical leftist ideology (Business Administration program). I went home every weekend to work in my shop.

Maybe you need a part time job at a fabrication shop or something. A lot of students to unpaid internships doing administration work but I'm sure there are fab shops that would take on a low wage apprentice on the weekend

Well, I'm an idea man, always have been really, so theory is a good place for me to be, but sometimes when you've got a dozen different ideas all pulling you in different directions, there is a very real danger of either stalling out somewhere in the middle, or just plain biting off more than you can chew and subsequently choking on the whole lot of it. I actually went back to school after being self employed as a PC / home LAN tech for about 8 years, so it took me a while to figure out that sometimes a lack of structure and consistency can, in and of itself, create a pretty serious bottleneck, but it is finally starting to sink in. Never much cared to be tied down to any one way of doing things, but the more I work on improving those skills that demand a regular and consistent dedication to technique, the more I realize that its something that has been sorely lacking in my life. It probably wouldn't kill me to pick up some work in a shop to be honest, but this is really more a "Zen" sort of a thing for me, and I'd rather use it to destress in a solitary down-time environment at the moment.

All the same though, I'm starting to think you're right about getting into TIG right now. I picked up a thing or two from my brother about building kilns, so I could put an oven together that could hit 400 degrees if I needed to, but that means seriously getting into oven building, and heat treatment of aluminum, and TIG, in addition to what I'm already doing with electric conversions. With the skills I have right now, I probably will get a much stronger joint out of a can of hysol 9430 and a good lug than a TIG weld. Even if I could run a pretty solid bead, and I could bring the areas that have been annealed back up to something like a t3 or t4 with age hardening (which I'm guessing is a little optimistic, considering that it won't have any sort of proper solution treatment), its still probably not going to out perform a proper aerospace grade adhesive with good sheer strength in most applications, without at least a good ten or 15 years of wear.

I'm thinking that for now, maybe I'll do some "screwed and glued" aluminum frames and focus on getting the router finished, then maybe take a crack at building carbon fiber frames like you were suggesting, only maybe with prefabbed tubes and aluminum lugs instead of wrapping or vacuum bagging something myself, since I will have a little experience with a similar design, and the ability to run off custom lugs on the router at that point.

I didn't heat treat my bike. It's good enough.
You can't really glue aluminum though. Maybe braze and fiber wrap after.

iamsofunny said:

I didn't heat treat my bike. It's good enough.
You can't really glue aluminum though. Maybe braze and fiber wrap after.

Click to expand...

I don't know, it seems to me like people have glued aluminum to great effect in the past with bikes, and it is certainly common practice among the DIY aircraft folks (not to mention the aircraft industry for that mater). Either way, it won't cost me much to try, so I'll let you know if it turns out to be a disaster, but I think it probably will be the most workable solution for me at this particular juncture. I do like the idea of wrapping light aluminum with fiber though. We really don't need to be nearly as weight conscious as the pure HPV folks, and a carbon-aluminum composite could be a convenient way to keep the bike affordable but still be very light and strong. Maybe I'll even throw a titanium ferule in the mix here and there to introduce a little spring in the frame where it makes the most impact, provided they can be sourced at a reasonable price.

So, apparently an ebike controller wouldn't really make sense as the inverter drive for a welding machine, but what about an induction heater? It seems to me that if you could target the ideal frequency for heating the filler metal instead of the base metal you could braze something like aluminum pretty easily (or at least much more easily than something that forces all kinds of oxidizing agents into the puddle just to get the thing hot in the first place).

I mean, yeah, obviously running off a bunch of frames by hand and then having them all solution hardened is going to be more versatile and cost effective if you are doing small production runs and know how to weld, but it might be easier and more effecive for some of us to do something like induction brazing for certain joints and then use a combination of bolting and or glueing or fiber rap for the rest.

At any rate, I found this website that goes into way more detail than I would have expected to find on the subject, and I figured it couldn't hurt to have a well documented example of a DIY guy building a fairly sophisticated induction heating system, so...

http://www.mindchallenger.com/inductionheater/

Too off topic, so I started a new thread. Sorry

6061 t6 is solution heat treated, precipitation treated and 'artificially aged' (400F for 1 hour baking).
Post weld treatment is the same thing, just bake at 400F for 1 hour.
The weld filler metal cannot gain any strength from this treatment because it's 4043 which only work hardens, sort of. In practice 4043 just kind of cracks when you bend it. But for filler metal it flows awesome at lower temperatures, does not get oxygen mixed in as easily as the 5356 and the strength is basically the same from my tests.

Any aluminum welding, except for the oxyfuel stuff, requires argon.

If you have enough money then I recommend you get the everlast tig 185, autodarkening helmet, get some argon and go for it.

You remind me of myself a few years ago trying to build bamboo boats in the Caribbean. It can't really be done because of bugs that eat the bamboo, but I tried many times.