After passing three of the five qualification weld tests of the "City and Guilds" welding course around Christmas 2003/2004, in quick succession, it was back to hard work and success running away from me into the uncertainties of "improving technique". Uphill "T"-fillet weld was proving difficult (it is reckoned to be particularly difficult with TIG).
Hand exercises eventually produced something of a break-through, in that in the end a rhythm set in where I could watch and the feeding of the filler wire and progression of the torch took over as a sort of programmed reaction, coupled to what I was seeing happening in front of me. I was also coupling in controlling on the foot pedal for power. But still the result was not perfect, as the following pictures show.
The front face seems fine...
But in spite of obviously reasonable control, the back surface is "burned" with melt-through defects (in oxy-acetylene welding of sheet mild steel, another process with a torch giving nothing but heat and there is an independent filler supply, you cannot avoid them, so you are forgiven them, but with TIG welding you can avoid them, so you have to "make the grade" in this regard).
This is not too bad and it would be difficult for students at a college to produce faultless welds in the duration of a course. But I thought about it and came to this realisation. When you make TIG weld fillet welds, you will find on experimentation that it is simply impossible to make fillets above a certain size without always-present melt-through defects.
This "perfect" argument came to me as I thought about the issue. The "throat thickness", the general definition of which eludes me but in a fillet weld is the thickness from the weld root to the fillet surface, which will be at a 45degree angle if the weld is uniform, needs to match the sheet thickness for the weld strength to match the sheet strength, all other things being equal, which they are to a good approximation if you have welded with reasonable practice. From this realisation, simple trigonometry shows a very simple rule for "leg length", the length of the fillet from its meeting front on one sheet to its meeting front on the perpendicular sheet. See the following diagram:
From this it can be seen, by simple geometry, that the aim fillet size should be that its leg length is twice the thickness of the sheet or plate metal you are welding.
Thus, in this case as photographed, the leg length should be smaller, at about 2/3rd's the size and therefore less than half the mass. At this considerably-larger-than-necessary size, it is not surprising that there are melt-through defects. By the arguments here and represented in the diagram, it is easy and always obvious what fillet size you should be aiming for.