Synergic MIG: what the machine is really doing, and why trim is not a heat knob

Where the word comes from

It’s Greek. syn means together, ergon means work. Synergia, working together.

Note what it does not mean. It isn’t “one plus one equals three”. That meaning is modern, about 1957, and the management consultants flogged it to death. The welding meaning is the old one: two things working together instead of fighting each other.

Who came up with it

Someone called Amin, writing in Metal Construction in Britain, 1981. It’s a research term, not marketing, and it’s about 45 years old.

The problem was pulsed MIG. Too many knobs for anyone to set by hand: peak current, background current, frequency, pulse width, ramp rates. So Amin made the whole lot follow one input.

Britain named it. Australia already had it running. CSIRO had a pulsed current generator working in 1979, and Welding Industries of Australia sold it here from September 1983 as the Synchro-Pulse CDT. Two knobs on the front: ARC CURRENT and ARC LENGTH. That is this whole page, in a die-cast box, forty odd years ago. The photographs above are that machine.

What it’s actually doing, and this surprises people

It isn’t controlling your arc.

All a synergic line is, is a relationship between amps and volts, worked out for three things:

• what wire
• what size
• what gas

Somebody did a heap of work to find out what the volts should be for the amps, for every combination of those three. Then they wrote it down and loaded the list into your machine.

On a normal MIG you’ve got two knobs that have to agree. Wire speed sets your amps, through the burn-off. Volts sets your arc length. Only a narrow band of pairs actually works. Get it wrong one way and the wire stubs into the plate. Wrong the other way and you’ve got a long harsh arc, spatter and undercut.

Synergic just stores the pairs that work. You set one, it looks up the other.

Now the important bit. On a plain synergic machine, the thing holding your arc steady is not the computer. It’s the same physics that held it before synergic existed: when the current goes up the wire burns off faster, so the arc pulls itself back to length. That happened on a 1960s machine and it happens on yours. The synergic bit only saves you guessing the volts.

It’s a lookup table. Not a brain. Interpolation is joining the dots with a ruler: the maker measured four settings, say 100, 150, 200 and 250 amps, the machine draws a straight line between them, and every setting you pick in between comes off that line, not off anything anybody actually welded at.

Think of it as the modern version of the settings notebook that lives in a good welder’s toolbox. Somebody else’s notebook, written on somebody else’s job.

The quarter turn

Think about a two-stroke. Chainsaw, outboard, whipper snipper, whatever you’ve got out the back.

The factory set the carburettor needles for one fuel, at one altitude, at one air temperature. At sea level on 91 it runs sweet. Take it up the hill, or change the fuel, and you give the needle a quarter turn to suit.

Nothing stops you running the factory setting anywhere you like. It just runs rich or lean. And the only thing telling you is how it sounds.

That’s a synergic machine. The stored curve is the factory setting. Trim is the quarter turn. And the arc is the sound.

Somebody in a laboratory ran that exact wire, in that exact gas, at that exact diameter, at one stick-out, and wrote down the right settings. Then they loaded the list into your machine. That list is the synergic line. Four points on a graph, usually. Straight lines in between. That’s it. It’s not clever and it isn’t watching you.

Lean and rich, in weld metal

Wind a two-stroke lean and it screams, gets hot and seizes. Wind trim down too far and the wire stubs into the plate before it melts. Harsh popping, stuttering, whiskers, and lack of fusion.

Wind it rich and it four-strokes, fouls and won’t rev. Wind trim up too far and the arc goes long and lazy, the arc root lifts off the leading edge of the pool, and you get undercut at the toes and porosity as the gas coverage lifts.

Same knob. Same two extremes. Same feedback, which is your ears. A dip transfer arc that’s right is a smooth fast crackle, like bacon. Spray is a smooth hiss. If it’s popping harshly, you’re lean.

Factory settings are right for one day, in one place, on one fuel.

So what does the primary knob do

It moves you along the list. Nothing else. On some machines that knob reads in metres per minute, on others it reads in amps, and on others it reads in millimetres of plate thickness. Same knob, three different units, and no manual tells you they’re the same knob. It’s like calling a length 25 mm, 2.5 cm or 0.025 m. One thing, three ways of saying it.

And this is the bit that costs money. When the dial says amps, you are not setting amps. You’re picking a point on the list, and the machine looks up the wire speed and the volts that go with it. The actual amps still depend on how far you’re holding the gun out, how good your earth is, and whether the tip’s shot. The machine has no idea about any of those.

So pick one unit and stay on it across the shop. Amps is the one worth standardising on, because amps is the number that turns up on every procedure sheet and every report you will ever have to defend, and it’s the one arc energy comes off. A shop where one welder dials amps, the next dials plate thickness and the third dials metres per minute has three mental models of the same machine, and not one of them lines up with the sheet. The makers were trying to simplify it. Mostly they’ve convinced people it’s three different things.

And trim

Trim is the quarter turn. Zero trim means you’re on the list, using the arc length the laboratory wrote down. Wind it up and the arc gets longer. Wind it down and it gets shorter.

Roughly a millimetre of arc per volt of trim. So plus 3 volts is about 3 mm more arc. Plus 5 is about 5 mm, and 5 mm is a lot of arc.

What trim is not is a heat knob. It doesn’t touch the wire speed, so your deposition doesn’t change and your amps barely move. Cigweld calls it Heat Control on the Transmig 355i and that name has done real damage, because someone chasing a fusion problem winds up the trim, gets a wider, flatter, prettier looking bead, and has not fixed the fusion.

It isn’t arc dynamics or inductance either, and those two get mixed up with it constantly. Different knob, different job. Trim moves the voltage, and voltage is arc length. That’s the lot.

People call volts heat for a reason, mind you. Lengthen the arc and you get more visible light, a more fluid pool and a wider bead, so it looks hotter. And the arc energy sum backs it up, amps times volts over travel speed. But most of the energy comes from the amps, and it has to: feed more wire and you need more energy to burn it off. Amps is wire feed by another name.

The one that costs money: your stick-out

The list was written at one stick-out. On the 1.2 mm solid wire in co2 example in the tool above, Lincoln published it at 13 to 19 mm, so call it 16.

Now reach into a joint. Everybody does it. Your stick-out goes out to 25 mm.

The wire speed hasn’t changed. The volts on the display haven’t changed. The program hasn’t changed. Nothing on that machine has moved. And the amps at the arc have gone from about 200 down to about 174.

You’ve lost 13% of your current, and every number in front of you says you haven’t. Your deposition is identical, because deposition is just wire speed times cross section. So the bead looks the same size. It just isn’t fusing the same.

That’s cold lap on a machine that “was set right”. That’s the whole mechanism, and I see it constantly.

What to do about it

• Learn your nominal stick-out and hold it. Roughly 10 to 12 mm on 0.8 and 0.9 mm dip transfer, 15 to 20 mm on 1.2 mm spray, out to 25 mm on 1.6 mm cored. It’s in the tool above for whatever you’ve selected.
• Check the program matches what’s on the machine. Not the wire you meant to load. The one that’s actually in it, and the gas that’s actually in the bottle. A synergic machine will happily weld on the wrong program. It won’t fault, it won’t warn you, and it’ll still put a bead down.
• Never run cored wire on a solid wire program. Metal cored burns off faster at the same current because the current runs in the sheath, not the core. 1.2 mm metal cored puts down 5.4 to 6.4 kg an hour against 3.6 to 4.5 for solid the same size. Run it on the solid program and you get undercut. Thats documented, not a maybe.
• If you’re winding trim to the stops, stop and look at why. Next section.

You can run any line with any wire and any gas

Nobody tells you this part.

Nothing stops you running the wrong program. The machine will not fault. It won’t warn you. It’ll weld, and it’ll put a bead down that looks like a bead.

No 0.9 mm line in your machine? Load the 1.0 mm line, run the 0.9 wire on it, and trim the volts to suit. That’s a legitimate workaround and I wouldn’t blink at a welder doing it.

What it costs you is ‘accuracy’. The curve has stopped doing the work and you’re doing it by hand.

And the bill is smaller than you’d think one size away. Running 0.8 wire on the 0.9 program wants about half a volt of trim at 100 amps, and about 2.6 volts by 250. Go two sizes, 0.9 wire on the 1.2 program, and you need nearly 3 volts at 250 and 8 volts at 380, which is off the end of the knob.

So the rule’s simple. One size away is nearly free at the bottom of the range and gets worse as you wind it up. Two sizes away runs out of knob at the top. The gap always widens with current, which is why the borrow that ran beautifully on 3 mm plate falls apart when the same welder goes up to heavy section.

Gas is the same story, and the direction surprises people. Load a spray tuned argon program, run co2 on it, and you’ll be winding trim down, several volts, and you still won’t get spray, because co2 can’t produce it at any current. Lincoln’s own SuperArc L-56 sheet has 1.2 mm in straight co2 at 19 to 21 volts and 145 to 200 amps, short circuit, and the same wire in 90 argon 10 co2 at 27 to 30 volts and 285 to 340 amps, spray. Same wire, same page, and co2 sits right down the bottom. Going the other way you wind up. Know which way you’re borrowing and by how much.

Then write the trim value on the sheet. That’s the whole discipline. A borrowed program with the trim written down is a controlled process. A borrowed program with the trim left to whoever picks up the gun is not.

Trim at the stops is telling you something

If a welder consistently needs a lot more volts than the list says, the list isn’t wrong. One of three things is, and all three are ten minute checks:

1. The stick-out is long. The extra resistance has dropped the current, so the list’s volts are now too low for the arc that’s actually there.
2. The program doesn’t match the wire or the gas actually loaded. Change the gas and a trim of zero is a completely different voltage.
3. Volts are getting lost between the machine and the arc. Four volts down long leads and dodgy connections on a 31 volt arc is documented. The machine sends what it was asked for. The arc never gets it. So the trim goes up to make up the difference.

Gun angle is the fourth, and it’s the one that beats the feedback loop entirely.

So log the trim next to the current on every machine in the shop. A fleet sitting near nominal has the right programs, has been set up properly, and has welders holding their stick-out. Anything on the stops, or past about plus or minus 3 volts, is telling you which of those three you’ve got. Costs nothing but a bit of time, and it’s the closest thing to a free diagnostic the equipment offers.

What it sounds like

• Trim too low, arc too short. Harsh popping. Stubbing. The wire drives into the plate before it melts and you can feel the feed motor labouring. Whiskers. Burnback to the tip if you keep going.
• Trim too high, arc too long. Lazy, wandering, wide. Spatter comes back. Undercut at the toes. And porosity, because the arc root is sitting up high and the gas column isn’t wrapping it any more.
• Right. Dip transfer is a smooth fast crackle, like bacon. Spray is a smooth hiss. If it’s popping harshly it’s either the trim or the wire speed, and the fix is not always trim.

And the gas, which the machine cannot see at all

No synergic machine senses gas. Not what’s in the bottle, not the flow, not a leak. And composition is the big one, because there’s no such thing as one argon co2 mix. Some shops run 80 argon 20 co2, some run 82/18, and then there are the tri mixes, argon with 16 per cent co2 and 3 per cent oxygen. Under ISO 14175 the first two are both M21, which covers anything from 15 to 25 per cent co2. The tri mix is M26, and your machine may not carry a program for it at all.

Then look inside the designation. Any component above 5 per cent can be delivered within plus or minus 10 per cent of its nominal value. Not ten percentage points, ten per cent of the number. So a nominal 20 arrives anywhere between 18 and 22, a nominal 18 arrives between 16.2 and 19.8, and both are fully compliant. That’s nearly six percentage points of co2 across two gases the machine holds one program for, and it has no way of knowing which bottle it’s being fed from. Enough to change how the arc sounds. Enough to want a volt of trim.

Straight co2 earns its own warning. Argon is a single atom and ionises into a smooth stable column. co2 has to be torn apart in the arc before it will conduct at all, and it can’t produce spray at any current. Wind the volts up on co2 expecting a nicer arc and you don’t get one. You get globular: big droplets, severe spatter, runs like a pig. A co2 procedure runs low volts on purpose. And running straight co2 in a modern shop is false economy, particularly with skills where they are, because you’re handing the welder a harder process and an erratic arc, and losing it again in spatter, to save a few dollars on gas.

Flow is the other one the machine can’t see. Turning the gas up to fix porosity can cause porosity: past roughly 21 L/min through a 16 mm nozzle the exit velocity creates a vacuum at the nozzle lip and pulls atmosphere into the plume. The working range is about 12 to 14 L/min on that nozzle. Not 28.

And the wire, which it can’t see either

Two spools can both say ER70S-6 and still run differently, because the standard allows a range, not a number. AWS A5.18 lets manganese sit anywhere from 1.40 to 1.85 per cent and silicon from 0.80 to 1.15. Those two are the deoxidisers, and they set how the pool wets, how the arc forms and how much silicate island you’re left picking off the top. A wire at the bottom of both bands and a wire at the top are both compliant ER70S-6, and they don’t run the same. Add the diameter tolerance, and copper coating on some and none on others, and it’s enough to sit you off the line.

Two reasons that matters more than it sounds. Silicon raises resistivity, and resistivity is what heats the stick-out, so two compliant wires don’t burn off at the same rate at the same current. And the wire isn’t only the thing being melted: iron vapour ionises at 7.9 electron volts against argon’s 15.76, so metal coming off the wire is a large part of what carries the current in the arc column. Change the chemistry and you’ve changed the conducting medium, not just the filler.

None of that is on the panel and none of it is in the program. It comes out as a wire that wants a different trim to the last one, on the same job.

Where the word comes from

Straight from the Greek. syn, together, plus ergon, work. Synergia meant joint work, cooperation.

It arrived in English in the 1650s as a theological term for human will cooperating with divine grace, and Jacob Henle picked it up in 1847 to describe the concourse of action between different organs. The meaning everybody now assumes, greater than the sum of the parts, is late. It dates to about 1957, it was popularised through Buckminster Fuller, and it was borrowed from pharmacology and metallurgy where two substances together did more than either alone. Then the management consultants got hold of it.

That history matters, because synergic control is not that kind of synergy. One plus one does not equal three inside a welding machine. It’s the older and truer meaning: two things working together instead of working against each other. No extra output. Just cooperation instead of argument.

Where it comes from in welding

Not marketing. It’s a research term, it’s British, and it’s traceable to one researcher.

M. Amin, Synergic pulsed MIG welding, Metal Construction 13(6) 349 to 353, 1981. Three more papers followed over the next six years: the pulse current parameters paper in 1983, microcomputer control of synergic pulsed MIG in 1986, and with Ahmed, Synergic control in MIG welding, in 1987. Metal Construction grew out of Metal Construction and British Welding Journal.

So the term is about 45 years old, and it was coined for a specific problem. Pulsed MIG had too many independent variables for a human to set at the machine: peak current, background current, pulse frequency, pulse width, ramp rates. Amin’s answer was to make all of them follow one input.

The usual line is that the Poms invented it and we improved it. Not this time. Britain named it, and Australia already had it running. Graeme Ogilvie started exploratory experiments at the CSIRO Division of Tribophysics in 1978 and had a thyristor pulsed current generator working in 1979, two years before Amin’s paper reached print. Welding Industries of Australia took it to market in September 1983 as the Synchro-Pulse CDT, Controlled Drop Transfer, with arc voltage and current sensing fed back to hold the drop transfer steady. Ogilvie took one of the three inaugural CSIRO Medals for Research Achievement in 1985. The machine welded Australia II and the flagpole on Parliament House, and it sold into shipbuilding overseas on aluminium and stainless, where the welds had to be right. Dates and the launch from CSIROpedia, Synchro-Pulse Controlled Drop Transfer Welding System.

And this is where the confusion starts, because the CDT was both. A pulsed transfer process and a synergic control, in the same box, in 1983. Load a program, turn one knob, and the machine scheduled the waveform off it. Two words for two different things, welded together in one product, and the trade has used them interchangeably ever since. It was a fabulous piece of gear, and the front panel of a CDT 450 is this whole page in a die-cast box: two knobs, ARC CURRENT and ARC LENGTH. One to move along the line, one to bias it. The photographs above are that panel, and the CIG Transmig 350 Pulse next to it does the same job with three switches for the three inputs and a knob marked FINE VOLTAGE for the trim.

They weren’t widely understood from the start either. As an apprentice I asked the tradesman whether the new synergic machine was any good, because I was keen to have a shot, and he told me the sparks don’t even burn. I found out that wasn’t the truth pretty quickly. The CDT was known for running an arc longer than most applications wanted, and you would often run out of adjustment. Which is exactly what the second knob was for, and exactly the knob nobody was taught to use. All I had to do was lengthen the earth lead and the long arc would have fixed itself.

It is not controlling the arc

This trips up nearly everybody, including a fair few of the people who sell these machines.

A synergic line is not arc control. It’s a relationship between amps and volts, built from three inputs: wire type, wire size, gas type. That’s the lot. Somebody did a heap of work to establish what the volts should be for the amps across every combination of those three, and the machine holds the answer.

Compare the two. Non-synergic gmaw gives you two knobs that must agree. Wire feed speed sets the current, through the burn-off relationship. Voltage sets the arc length. Only a narrow band of pairings works, and outside it you get stubbing one way or a long harsh arc, spatter and undercut the other. Synergic control stores the pairings that work. You set one variable and the machine reads off the matching one, and in pulsed mode every pulse parameter with it.

Now the bit that matters. On a plain synergic cv machine, nothing in there is controlling your arc. The arc is held by ordinary constant voltage self regulation, the same physics that held it before synergic control was invented: when the current rises the wire burns off faster, so the arc length restores itself. That happened on a 1960s machine and it happens on yours. The synergic layer never touches it. All it does is save the operator from guessing the voltage.

Where real arc control does exist, it’s a separate layer bolted on top and it is not the curve. In synergic pulse a waveform controller estimates arc length from the voltage and current on the pulse ramps and corrects it inside individual pulse periods. In an adaptive machine there’s a stick-out compensation loop with a declared envelope, and Lincoln’s R450 states its own as 12.7 to 31.8 mm. Both are closed loop control. Neither of them is the synergic line.

So the curve is a lookup table with interpolation. Not intelligence, and not control.

That last one has a tail, and the rest of this page comes out of it. You select three things. The curve assumes two more and never asks you about either: the mode of metal transfer, and the stick-out it was measured at. Neither is on the panel.

Factory settings are right for one day, in one place, on one fuel.

What the curve actually is

Given all that, the definition worth quoting is Norrish’s, because it’s the only one in print that names the thing being held:

Three things in that sentence that nobody else says. It admits open or closed loop. It’s framed on ‘equilibrium’, not on operator convenience. And it started as a pulsed transfer definition and got generalised later, which is why twi’s own faq still describes synergic as a variant of pulsed mig and is now wrong.

Note also what the term is not. There is no definition of “synergic” in any edition of IEC 60974-1. IEC 60974-14:2018 uses the word without defining it: Annex C is titled “Slope, pulse and synergic controls” and C.4 is “Pulsed MIG and synergic controls”. The term with an iso definition behind it is ‘waveform controlled welding’, from ISO/TR 18491. AS/NZS 1554.1:2014 clause 1.1 Note 2 uses both, and that note is the only place in the whole standard where synergic control is named.

Three things get called synergic, and only two of them are

1. Synergic pulsed gmaw. The original meaning. A stored parameter set, peak current, peak time, background current, background time or frequency and the ramp rates, scheduled off wire feed speed to hold one drop per pulse across the range. The parameter list is confirmed at patent level: Lincoln US20110204034A1 names pulse frequency, peak current, background current, current ramp rates, pulse width and pulse curvature, with interpolation between table entries.
2. Synergic cv. Wire feed speed is the single knob and the target voltage is looked up from a stored curve. The Power Wave R450 manual says it plainly: “for each wire feed speed, a corresponding voltage is preprogrammed into the machine through special software at the factory”. The regulation underneath is still ordinary cv self regulation. Synergic only removes the operator’s need to guess the voltage.
3. Controlled short circuit and waveform processes: stt, rmd, cmt, ColdArc, lsc. These are not synergic lines. They’re event driven closed loop current controllers, triggered on short circuit detection and on the dv/dt necking signature, resolving events of the order of 0.75 ms. They’re wrapped in a synergic interface, which is exactly why everybody conflates them. cmt is mechanically closed loop, reversing the wire. The mechanism that produces the low spatter is not the curve.

Confusing the third with the first two is the most common error in trade literature, and it’s the reason people credit “synergic” with things it has nothing to do with.

The method: what a synergic line is, and what trim does to it

The curve itself

A synergic line is a one dimensional look-up table or piecewise function, indexed on wire feed speed, returning a matched parameter vector valid for one combination of wire chemistry and class, wire diameter, shielding gas composition and intended mode of transfer. For synergic cv the returned vector is one number, the target arc voltage. For pulsed synergic it’s the whole waveform descriptor.

The internal structure is documented, and it’s smaller than people imagine. A Migatronic COMMANDER BDH holds three curves per synergic program, one for wire speed against current level, one for welding voltage, one for arc adjust. Pulsed mig adds three more for the pulse parameters. Each curve is defined by four calibration points creating three intervals, with linear interpolation between them. Try to define a breakpoint without a higher current level than the one before it and the machine throws Err A- and refuses. That monotonicity constraint is what makes single valued inversion possible.

Four points. That’s a synergic curve. The tool above plots the published breakpoints as dots precisely so that’s visible, because it’s the least appreciated fact in the whole subject.

Why the curve is single valued, and why that forces trim to exist

The curve is a function of one variable. It returns the voltage that produces the intended arc length and current only at the stick-out it was fitted at. It cannot know the actual ctwd, the joint geometry, the position, the plate temperature, the gas flow, or whether the operator wants a tighter or a softer arc. Trim is the operator’s single degree of freedom against a fixed table. It is a bias on the table’s output, never a replacement for it.

In pulse, voltage cannot be a set point at all, and Lincoln says so in the Power Wave 455M manual: “when pulse welding, the arc voltage is highly dependent upon the waveform. The peak current, background current, rise time, fall time and pulse frequency all affect the voltage.” A preset voltage is therefore impractical, and the maker hands you trim instead. That’s the whole answer to “why is there no volts knob in pulse”, and almost nobody teaches it.

What trim physically acts on, which is not the same in the two modes

• Synergic cv: a voltage offset applied to the stored voltage curve. Documented in volts by three independent makers. Migatronic: the trim value “determines the number of volts by which the voltage synergic curve should be adjusted (up or down)”, minus 9.9 to plus 9.9 v. EWM: “the arc length (welding voltage) can be adjusted for the welding task in hand by plus or minus 9.9 v”. Kemppi: FineTuning “adjusts the arc voltage of the curve within certain limits”.
• Synergic pulse: an arc length target that the controller services by modulating the waveform, most commonly the background current. Kemppi documents this explicitly, FineTuning minus 9.0 to plus 9.0 for “adjusting base current and arc length”, and WiseRoot BaseCurrent minus 50 to plus 50 “for adjusting arc length within specified limits”. The physics is sound: background current doesn’t detach droplets, so changing it changes the burn-off between pulses and therefore the mean arc length, without disturbing the one drop per pulse detachment condition that the peak sets. Explicit for one maker, inferred across the industry.
• Same knob, three jobs on one machine. The Power Feed 15M manual: trim does not apply in cv modes at all, voltage control is used instead; in pulse synergic it adjusts arc length; and in stt “trim changes the heat input by adjusting the tailout and background portion of the waveform”.

The conventions, and why they don’t transfer

There’s an Australian angle in that table. One Australian industrial supplier’s mig catalogue, counted by models stocked, runs Kemppi 10, ESAB 8, EWM Group 5, WIA 4, Lincoln 3, Cigweld 2. The european brands express the control in volts. The american brands express it as a dimensionless multiplier or index. A welder moving between an EWM Picomig and a Lincoln Power Wave in the same shop is turning knobs measured in two units that do not convert. And Cigweld, which has the deepest trade recognition in this country, is ESAB owned and calls it Heat Control, while WIA is Lincoln owned. Same underlying logic, Australian badges, three mental models.

Do not invent a volts equivalence for the Lincoln scale. Lincoln doesn’t publish one, and the manual explains why: “the exact voltage for a given wire feed speed can only be predicted when all the pulsing waveform parameters are known”. Cameron makes the same point independently and adds the one that matters most, that a trim setting of 1.00 produces a different actual voltage the moment you change the electrode, the shielding gas or the wire feed speed. The correct method is empirical. Set the wire feed speed, sweep trim, record the displayed volts at each step, and build a local calibration for that mode, wire, gas and stick-out. And Lincoln, which invented the dimensionless scale, now ships the Power Feed 42 with “Display Trim as Volts” set to yes by default.

Which raises the practical problem. Write plus 2 v on a wps and that instruction only means something on a machine that expresses trim in volts, in the same mode it was measured in, with the same total lead loop length: interconnecting leads, work return and torch. Hand the same sheet to a shop running a different machine and you haven’t made it simpler, you’ve made it ambiguous. The value that matters is arc voltage, measured with a decent tong tester between the wire feeder and the job, not the digital readout. Record the trim by all means, and teach the welders what the number is and is not worth.

This is a large part of why a procedure qualified on one machine does not travel to the next one and give the same result. It is not the welder’s fault. Managing that difference, and training for it, is the welding supervisor’s job.

The physics under the curve

The curve is empirical, but the shape it encodes is not arbitrary. Melting rate is Lesnewich’s burn-off relation:

Two consequences fall straight out of that, and both of them are why this page exists.

First, current is determined, not chosen. Set the wire feed speed and set the extension, and there’s exactly one current at which melting rate equals feed rate. That’s the equilibrium in Norrish’s definition. Inverting the equation gives the current directly, and it’s the arithmetic the tool above runs:

And that’s the whole basis of the synergic idea. If current is a determined function of wire feed speed, then so is the correct voltage, and it can be stored.

Second, sensitivity to stick-out is calculable. Differentiate at constant wire feed speed:

With Richardson’s 1.2 mm coefficients at 15 mm extension that predicts 6.6 a per mm at 300 a. Measured independently: Miller, gmaw spray, 15.9 to 25.4 mm ctwd took 282 a down to 213 a, which is 7.3 a per mm. ewi, gmaw-p at 5.1 m/min wire feed speed, 19.1 to 28.6 mm took 300 a down to 225 a, which is 7.9 a per mm. Within about 10%, from published coefficients, with no fitting. That’s a model you can trust with a decision.

Note that the sensitivity is not constant. It scales roughly with current, so a 5 mm stick-out error costs about 20 a at 200 a and about 40 a at 350 a. The rule of thumb for solid wire in the spray range is about 7 a per mm, or about 2.5% of set current per mm. Below that it’s less. The tool computes it for whatever operating point you’ve set rather than quoting the headline figure at you.

Cored wires are a different physical object, not a variant. Current flows almost entirely in the metal sheath, so the effective conducting area for the joule term is an annulus, not a circle. Since b carries the inverse square of area, b is much larger than the nominal diameter suggests, and the wire burns off faster at the same current. That’s the actual reason 1.2 mm metal cored deposits 5.4 to 6.4 kg/h against 3.6 to 4.5 kg/h for solid wire the same size. It’s also why cored wire synergic curves have to be separate program entries, and why running metal cored on a solid wire program causes undercut. And it’s why Hobart’s FabCOR sheet grades nominal ctwd by diameter, 13 mm at 0.9, 19 mm at 1.2, 25 mm at 1.4 and 1.6. The curve and the nominal stick-out are a matched pair.

Worked example: 1.2 mm solid wire in co2, and one reach into a joint

Set the tool above to carbon steel solid wire, 1.2 mm, co2 (C1), synergic cv. Lincoln’s published SuperArc L-56 rows for that combination give three real breakpoints, and the third one is the operating point most Australian structural shops actually live at.

The arithmetic, so you can check it. Nominal electrode extension at 16 mm ctwd and a 4 mm arc is 12 mm. Solving the burn-off equation at 12 mm gives 204 a against the 200 a Lincoln published, which is 2% out and a fair validation of the coefficients. Go out to 25 mm ctwd and the extension is 21 mm, and the same equation gives 174 a. The local sensitivity is 4.3 a per mm at nominal, falling to 3.1 a per mm as the extension lengthens, so the integrated loss over 9 mm is about 30 a.

Now the compliance consequence, and it’s the reason this is not an academic point. AS/NZS 1554.1 Table 4.11(A) makes a change of more than plus or minus 10% of the specified mean welding current an essential variable, and a change of more than plus or minus 20% in electrical stick-out likewise. Item letters and figures verified against the 2000 edition; buy the current standard before citing item letters against a 2014 date.

So a welder reaching into a joint on a machine nobody has touched has just put the weld outside two essential variables at once. Current 13% down, past the 10% band. Stick-out 25 mm against a nominal 16, which is 56% out, past the 20% band. Zero machine changes. Nothing on the display moved. And deposition is identical, so the bead is the same size and looks right.

The competence problem in one paragraph: the range on the sheet doesn’t weld anything, the welder does, and the only variable on that sheet a welder can actually obey without a meter is stick-out.

Where the story ends

The quarter turn is honest about the factory setting, about trim being an override you make by hand, and about the arc being the feedback. Three places it stops.

It has nothing to say about the shape of the record. A carburettor needle is continuous. A synergic line is four points with straight lines assumed between them, and that discreteness matters: near the ends of a curve you are operating where it was fitted on the fewest data points and where adaptive authority is smallest. Qualify there, or move to a different curve.

It has nothing to say about waveform. In synergic pulse there is no single arc length being held. There is a droplet detached once per pulse by a peak current, with the mean arc length managed by the background current between pulses. A needle cannot carry that, and the tool above does not pretend: select synergic pulse and the voltage chart refuses to draw a centre line, because the centre line does not exist as a published number.

And it has nothing to say about adaptive control. A machine with a genuine adaptive layer, and Lincoln’s R450 declares its envelope as 12.7 to 31.8 mm of electrical stick-out variation, is doing something a carburettor cannot: sensing the arc length from the voltage and current on the pulse ramps and correcting it inside individual pulse periods, so arc length is held independently of extension. Fronius goes further with a penetration stabiliser that holds the current and lets the wire feed speed float, which inverts the whole architecture. A machine in that mode is no longer synergic in the classical sense. Most shops that own the feature have never turned it on.

What the curve cannot see

The machine reads its own output terminals. In all but a handful of adaptive machines, that is the entire extent of its awareness. Four things sit outside it, and all four move the arc away from the number on the wire feeder display. Stick-out is the worked example above. The work return circuit is the 4 v case below. The other two are the gas and the wire, and neither is sensed at all.

The gas: one designation, six percentage points of co2

There’s no such thing as one argon co2 mix. Some countries run 80 argon 20 co2, others 82/18, and then there are the tri mixes, argon with 16 per cent co2 and 3 per cent oxygen. Under ISO 14175 the first two are both M21, a designation that covers anything from 15 to 25 per cent co2. The tri mix is M26, and plenty of machines carry no program for it at all.

Then look inside the designation. Any component above 5 per cent may be delivered within plus or minus 10 per cent of its nominal value. Not ten percentage points, ten per cent of the number. A nominal 20 arrives anywhere between 18 and 22. A nominal 18 arrives between 16.2 and 19.8. Both are fully compliant, and the spread across the two is nearly six percentage points of co2 for one program and one trim setting. The machine has no way of knowing which bottle it is being fed from. It is enough to change how the arc sounds and enough to want a volt of trim, and it is invisible on every document in the shop except the gas certificate.

The direction of the correction is the useful part. Load a spray tuned argon program and run co2 on it and you’ll be winding trim down, several volts, and you still won’t get spray, because co2 is a molecular gas that has to be dissociated before it will conduct and it cannot produce axial spray at any current. Lincoln’s own SuperArc L-56 sheet puts 1.2 mm in 100 per cent co2 at 19 to 21 v and 145 to 200 a, short circuit, and the same wire in 90 argon 10 co2 at 27 to 30 v and 285 to 340 a, spray. Same wire, same page, and co2 sits right down the bottom.

Which leads to the commercial point, and I’ll own it as opinion. Running straight co2 in a modern shop is false economy. You are handing a welder an erratic arc and a harder process at a time when skills are already stretched, and giving back in spatter and metal recovery whatever you saved per cubic metre. The economics of shielding gas deserve their own article, and they will get one.

The wire: two compliant spools that don’t run the same

Two spools can both say ER70S-6 and run differently, because the classification is a range, not a number. AWS A5.18 permits manganese from 1.40 to 1.85 per cent and silicon from 0.80 to 1.15. Those are the deoxidisers, and they set how the pool wets, how the arc forms and how much silicate island is left on the cap. A wire at the bottom of both bands and a wire at the top are both compliant. Add the diameter tolerance in ISO 544, and copper coating on some and none on others, and the accumulation is enough to sit you off the line.

Two mechanisms make that more than a curiosity. Silicon is a powerful resistivity raiser in iron, and resistivity is what heats the electrode extension, so two compliant wires do not burn off at the same rate at the same current: the joule term in the melting rate equation is not the same wire to wire. And the wire is not only the thing being melted. Iron vapour ionises at 7.90 eV against argon’s 15.76 eV, so metal vapour coming off the electrode carries a large share of the current in the arc column, enough that it produces a central minimum in arc temperature. Change the chemistry and you have changed the conducting medium, not just the filler.

Cored wire is not a variant of this, it is a different object, and the physics is in the melting rate section above. Current runs in the sheath rather than the core, sheath thickness is the maker’s own choice, and two cored wires of the same class and diameter do not share a conducting area. Run any of it on a solid wire program and you will be a long way off the line, with undercut to show for it.

None of this is on the panel and none of it is in the program. It arrives as a wire that wants a different trim to the last one, on the same job, and it is the reason trim values do not transfer between shops even when the machines match.

Why the displayed numbers are not the numbers

Three errors stack here, and people usually know about none of them.

One, the averaging error on a waveform. Norrish’s measured spot checks, 5 khz sampling on an AMV 4000 logger:

Read that table three times. First, the error is zero for conventional spray transfer. This is not a gmaw problem, it’s a waveform problem, and a shop running plain cv spray can keep using volts times amps over travel speed. Second, the sign reverses. For pulsed the conventional figure understates the true arc energy by about a quarter. For dip, conventional and controlled alike, it overstates it. If you’re holding an upper arc energy limit on a q&t or tmcp steel and you’re welding pulsed, your paperwork is optimistic and you may already be over the limit. Third, and this is the one to put in front of a client who thinks it’s academic, the downstream consequence is far bigger than the input error: on the conventional dip case Norrish reports the calculated t8/5 came out 46% longer than the naive figure suggests, because the cooling time relation is non-linear. An 8% error in became a 46% error out.

Two, the instrument tolerance. EN 50504:2008 clause 4 Table 1 permits a standard grade power source, one built to EN or IEC 60974-1 with no additional accuracy claim, plus or minus 10% of the true value on displayed current and voltage across the top three quarters of its range. Ten per cent on current and ten per cent on voltage compounds to roughly plus or minus 20% on the product, before you reach the waveform problem at all. IEC 60974-14:2018 Table 1 and Table 2 cell values not verified from a clean rendering; the EN 50504 figures are.

Three, the sensing point. The machine’s voltmeter reads at its own terminals, not at the arc, so it always reads high. A documented case: 4 v lost through leads and connections on a 31 volt arc, which is a 13% error, requiring 35 v at the supply. The machine sent what it was asked for. The arc never got it, so the trim goes up to make up the difference and now there are two faults stacked on each other. With short leads, about 4 m, the difference is negligible. Ammeter readings are unaffected by cable length, being series values. And note the consequence for machines with remote voltage sensing: the same program with the sense lead connected and disconnected is not the same weld. The field check costs nothing: clamp on the work, not the table, and put a hand on the connections after a run. Warm means resistance.

ISO/TR 18491:2026 clause 7.4 gives a procurement test you can use today. An acceptable meter is “often identified by the terms ‘true energy’, ‘true power’, or ‘power factor'”. One marked “kVA”, “DC power” or “average power” is not acceptable. Clause 7.2 and Table 2 also name the sensing point for process 13(x): the connection in the wire feeder. A wps quoting a voltage without a sensing point is quoting an unfalsifiable number.

What a wps has to carry, and what breaks when it doesn’t

ISO 15609-1:2019 clause 4.4.9 expressly permits recording “pulse welding details (machine settings, programme selection and all pertinent information of the process)”. So a shop writing “program 27, trim 0” hasn’t broken a rule. The consequence is in the last two sentences of clause 4.4.10:

Recording only the program number collapses the wps’s range of application onto that machine type. The document stops being a welding procedure specification and becomes a machine specific recipe. Sell the machine, buy a different brand, or take a firmware update that renumbers the synergic lines, and the procedure is gone along with every hour of pqr testing behind it.

Stack that against ISO 15614-1:2017 + Amd 1:2019 clause 8.5.2.3.2, which at level 2 makes the power source manufacturer and the waveform control mode essential variables, and requires a new qualification test on a change of either. Two independent locks: brand locked by the qualification standard, machine type locked by the wps standard.

Norrish states the consequence plainly: where a procedure is qualified using a waveform controlled process, production welding needs “the same power source manufacturer, model, program and synergic lines”.

The transfer mode rules, which run in opposite directions

This catches people, so it’s worth setting out side by side.

• Procedure. ISO 15614-1 clause 8.5.2.3.1, as replaced by Amd 1:2019: “the qualification using short circuiting transfer qualifies only short-circuiting transfer. Qualification using spray, pulse or globular transfer qualifies spray, pulse and globular transfer.” So spray to pulsed is not a requalification. Same architecture as ASME IX QW-409.2, whose boundary per ASME’s own form QW-484A is “spray, globular, or pulse to short circuit-GMAW”, and AS 3992:2020 clause 5.1(c), which drafts “waveform-controlled arc” as a peer of spray and globular.
• Welder. ISO 9606-1:2012 clause 5.2: “qualifying the welder for dip (short-circuit) transfer mode (131, 135 and 138) shall qualify him for other transfer modes, but not vice versa.” The person qualified on dip covers everything. The person qualified on pulse does not cover dip.

So the direction of coverage flips depending on whether you’re talking about the procedure or the person, and the reason is obvious once it’s stated: dip is the harder mode to control and the one that produces lack of fusion, so it’s the harder test. A welder tested on synergic pulse is not qualified for conventional cv dip transfer.

One more trap. Clauses 8.5.2.3.2, .3 and .4 are all prefaced “For level 2”. At level 1 the waveform control mode is not an essential variable at all. A fabricator qualifying to level 1 to satisfy an asme-leaning client gets no waveform lock whatsoever.

The auditor’s list

Record the program so the welder can set the machine. Record the following so an engineer can rebuild the weld on somebody else’s machine.

1. Wire feed speed range, in m/min. The only parameter a synergic machine actually commands and the only machine independent one. ISO 15609-1 clause 4.4.9 requires it only for mechanised and automatic welding, which is a genuine gap for hand held gmaw. Record it anyway. Weld Australia TGN-SG06 section 6.15 makes the technical case, that on a waveform machine you cannot reliably measure the amperage, and section 6.16 makes the supervisory one, that it’s a check on the current meter’s calibration.
2. Contact tip to work distance. Everything above.
3. Travel speed range, and therefore run length and time, not a dial reading.
4. Arc energy range by method B or C of ISO/TR 18491 clause 6, from instantaneous energy or instantaneous power, never method A. State which method on the face of the wps: ISO 15614-1 clause 8.4.7 requires that “the kind of calculation, either heat input or arc energy, shall be documented”.
5. Where the voltage was sensed.
6. Mode of metal transfer, in words. Essential in QW-409.2, in ISO 15614-1 8.5.2.3.1 and in AS 3992:2020 5.1(c). It is not an essential variable in AS/NZS 1554.1 clause 4.11, which is a defect in that standard rather than a licence to omit it.
7. Shielding gas to ISO 14175 including manufacturer and trade name, and consumable make and trade name, not just classification. The synergic line is calibrated to a specific gas and a specific wire. The program means nothing without both.
8. Power source manufacturer, model and firmware version, the waveform control mode by its proper name, and every offset at its actual set value: trim, arc length correction, inductance, dynamic, pulse correction. Plus the pulse parameters where the machine exposes them.
9. Evidence the machine’s own displays are calibrated or validated. On a waveform machine the displays are the only readings you can use, so their calibration is load bearing. IEC 60974-14:2018 is the standard for it. EN 50504:2008 clause 6 set the interval at yearly with a three month initial recheck.

The business case, honestly

The legitimate case for synergic control is variability reduction and faster changeover, not peak performance. It narrows the spread of parameters selected across a variable operator population. ESAB’s framing, “reduces setup and changeover time for everyone”, is defensible.

Two facts bound the claim, and both come from independent sources rather than manufacturers.

The first is twi’s survey of its industrial and professional members. Average repair rates in oil and gas and power run 1 to 3%, with peaks of 25% at specific locations and exceptional cases to 55%. The highest rates concentrate at root runs, fillet welds and areas of limited access. And the three dominant factors are welder skill level, weld location and accessibility, and poor fit-up prior to welding. Read that against the sales proposition. Synergic control addresses none of the three.

The second is that I went looking for independent quantification of synergic control’s own contribution to productivity or quality on manual gmaw, isolated from the pulse waveform or a consumable change, and there isn’t any that I could find. There’s good published work on pulsed against cv heat input and process efficiency, and on adaptive control in robotic welding. Nothing on the curve itself. Every figure in circulation is manufacturer or distributor sourced, and the headline one, three to four times the travel speed, is benchmarked against stick and tig, not against the cv gmaw machines the buyer already owns.

What the curve does control is worth having. Deposition efficiency inside gmaw runs 92 to 98% depending on gas and transfer mode, and that six point spread is decided entirely by which program somebody loaded. On 4.5 tonnes of deposited weld metal it’s roughly 300 kg of wire. Labour is typically 85% of total welding cost and arc-on time averages about 20%, which is 12 minutes of arc in the hour, so a wire with a published deposition rate of 9 kg/h delivers about 1.8 kg/h in the real world. Setup errors are labour hour errors, not consumable errors, and anyone arguing about the cost of a better gas mix against the cost of grinding time has the decimal point in the wrong place.

The documented failure pattern for a synergic fleet is worth knowing before you buy one. A fabricator with 250 welders and a mixed fleet of Power Wave 455 and 655, Idealarc 400 and S500 set out to move to pulse. Fleet heterogeneity defeated standardisation, because no two models use the same pulsing programs, so a shop with three machine generations cannot write one wps. The unbudgeted prerequisite was the electrical circuit: work clamps and power connections have to be right, because added resistance warms connections and degrades the feedback the pulse controller depends on. Vendor opacity blocked the paperwork, and a qa department that cannot document the controlled variables cannot qualify the procedure, so the shop reverts to cv where it can. And nobody had asked the first question, which their welding engineer eventually put to the buyer as: what do you do that makes you think you need pulsing when you’ve been using standard cv for so long?

What worked in that case wasn’t the equipment. Trials on a few machines, a tight identified window of 260 to 280 a and 25 to 27.5 v, the settings locked down, structured training on the actual kit, strict wps compliance, and the objective set as making the correct size weld rather than making it faster.

So be clear about what you have bought. The starting point is real and good equipment is worth having, but good equipment does not translate into faster, better welds on its own. Budget for the change, not just the hardware: commission the leads, clamps and work return, get the right wire and gas, and train the welders on what synergic does, what it does not, and what is likely to go wrong. Do none of that and the new gear creates more problems than it solves. That is change management and process optimisation, and it is the competence the industry is short of. Do not buy synergic machines and expect a solution. It is not written into the curves.

Synergic reduces variability in what the machine commands. It does nothing to variability in what the arc receives. That’s the technical heart of the whole subject, and it’s also the honest answer to a purchase order.